Assessment of urban rail train drivers’ emergency handling capability based on a physio-psycho-machine-environment-management multidimensional framework

This study addresses two major limitations in the current evaluation system for urban rail train drivers’ emergency handling capability: the lack of clearly defined criteria, and an overemphasis on technical skills to the neglect of psychological factors. We innovatively construct a multidimensional evaluation framework based on the Physio-Psycho-Machine-Environment-Management (PPMEM) model. Through a systematic analysis of the core components of emergency response capability and its influencing factors, a mechanism model rooted in “Human-Machine-Environment-Management” theory is established. Empirically, 30 key influencing factors were identified and categorized into seven dimensions: cognitive, physiological, skill-based, psychological, equipment, environmental, and managerial. A mixed-methods approach was adopted. During the qualitative phase, a system of influencing factors was determined through field studies and in-depth expert interviews. In the quantitative phase, a questionnaire survey was administered to employees of Kunming Rail Transit Operations Co., Ltd. (N = 538 valid responses), and a multidimensional evaluation model was developed using structural equation modeling (SEM) with Amos 26 Graphics. The results indicate that the total effects of latent variables on emergency handling capability, in descending order, are: psychological factors (β = 0.214) > physiological factors (β = 0.212) > environmental factors (β = 0.205) > equipment status (β = 0.126) > cognitive factors (β = 0.105) = skill-based factors (β = 0.105) > managerial factors (β = 0.102). Notably, psychological, physiological, and environmental factors all exhibited effect sizes exceeding the significant threshold of 0.2, constituting a core group of determinants for emergency response performance. Therefore, metro operators should prioritize improvements in drivers’ workload management, mental health support, and environmental adaptability, supplemented by targeted skill and cognitive training, as well as policy refinement. These measures will contribute to a systematic enhancement of emergency response capabilities. The findings provide both a theoretical foundation and practical guidance for strengthening emergency management systems in urban rail transit.

1 Introduction

Urban rail transit serves as the backbone of urban public transportation, and its operational safety is critically linked to the safety of passengers’ lives and property, as well as social stability. According to the 2024 Statistical Analysis Report on Urban Rail Transit Operations released by the China Association of Metros, the annual passenger volume of metro systems in China has reached 32.2 billion, with an average annual growth rate of operational mileage exceeding 15% (Metros, 2024). A single metro train can carry between 1,200 and 1,800 passengers. Any service disruption can therefore lead to significant social impact. As the primary operator in charge of train control, frontline monitor of line conditions, and first responder to onboard failures, metro train drivers are required to possess strong emergency handling capabilities. They must rapidly perform tasks such as fault identification, information reporting, troubleshooting, and incident resolution under extreme time constraints (Sun, 2020; Yang et al., 2023). The effectiveness of emergency response plays a decisive role in restoring normal operations and mitigating the impact of failures. However, current evaluations of drivers’ emergency response capabilities predominantly focus on technical skills, paying insufficient attention to non-technical factors such as physiological, psychological, and cognitive elements. Yet, once drivers are qualified for independent operation, the influence of these non-technical factors becomes considerably pronounced (You et al., 2021; Collins, 2023). Given the variability of emergency scenarios in urban rail systems, a single evaluation standard often fails to meet the need for accurate assessment (Wang et al., 2024; Yang et al., 2024).

In recent years, research on emergency response has primarily focused on methodological design. For example, researcher (Li et al., 2022) used simulation to model different states of emergency gates, escalators, and automatic fare gates before and after emergency instructions to study the impact of response time on evacuation efficiency. Researcher (Kroll et al., 2020) designed a holistic hazard test for emergency response drivers, incorporating both hazards and multiple-choice questions. Researcher (Wei et al., 2022) applied active inference predictive processing to simulate drivers’ emergency braking responses to automated system failures within a perceptual-cognitive-behavioral framework. Researcher (Yang et al., 2025) investigated subtleties in train drivers’ hazard detection by analyzing eye movements and responses to simulated foreign object intrusions at various speeds. Researcher (Lv et al., 2022) utilized electroencephalogram (EEG) signals, combined with driving performance quality, and employed DPCA clustering algorithms to classify driving fatigue and label EEG feature datasets. Nevertheless, existing studies often adopt oversimplified dimensions for influencing factors. For instance, researcher (Davidich et al., 2025) emphasized psychophysiological and vehicle dynamic factors but omitted management-related elements; researcher (Ma and Yao, 2025) developed a human-machine cognitive load balance model under time pressure without incorporating environmental factors; researcher (Lu and Sun, 2020) constructed a multi-objective model for emergency resource allocation during metro emergency rescue based on scenario analysis, yet overlooked human factor indicators. Several scholars (Tolani et al., 2006; Wang et al., 2022; Xu et al., 2024; Xiao and Terzi, 2025) and others (McClintock, 2019; Xing et al., 2021; Klocek et al., 2023; Santis et al., 2024) have conducted preliminary studies on “human-machine system” theoretical frameworks and complex systems theory, though few have applied these in the context of urban rail transit. Thus, there remains considerable room for improvement in the systematic integration and quantitative analysis within this field.

In terms of human error research, initial efforts focused on error classification and causation, revealing relationships between cognitive bias and organizational deficiencies (Leicher and Mulder, 2016; McWhirter et al., 2021). Subsequent work deepened mechanistic understanding by employing quantitative models to analyze error probabilities (Fan et al., 2022; Lin et al., 2022), and utilized neuroscientific tools (e.g., eye-tracking, EEG) to explore biological mechanisms underlying signal misjudgment (Wang et al., 2014; Kovesdi et al., 2018). With technological advances, research expanded to address new risks in human-machine interaction, providing evidence that automated cockpits can reduce situational awareness (Wang T. et al., 2020; Kashevnik et al., 2022; Yi et al., 2023). Current cutting-edge studies aim at predictive intervention, integrating multimodal data to develop dynamic error prediction systems (Zhao and Kasim, 2023; Zhou and Guo, 2024). The underlying logic follows a progressive chain from “phenomenon description → mechanism deconstruction → technical response,” evolving from static factor analysis (e.g., fatigue, management) to dynamic coupling studies of “human-technology-environment,” with methodologies shifting from empirical observation to data-driven modeling, ultimately supporting proactive safety protection.

The PPMEM (Physiological-Psychological-Machine-Environment-Management) framework serves as a comprehensive analytical tool in system safety, with theoretical roots tracing back to systems engineering and complex system safety theories. Evolving from the early SHEL (Software-Hardware-Environment-Liveware) human factors model (Perboli et al., 2021; Doi, 2024) and incorporating analytical approaches from latent failure analysis (Werth and Christopher, 2021; Andreou et al., 2025), this model establishes a holistic perspective for system safety analysis through its five core dimensions. Its systematic theoretical structure demonstrates distinctive value in safety analyses within high-reliability industries such as aviation and nuclear power, effectively elucidating the mechanisms through which system risks emerge under the coupling effects of multiple factors. However, specific limitations become apparent when applying this framework to the specific context of metro drivers’ emergency response capability: the delineation of the “Human” dimension remains overly broad, while the boundaries between the “Procedure” and “Management” dimensions lack sufficient clarity, somewhat diminishing the model’s explanatory power regarding dynamic operational processes. Building upon these identified theoretical shortcomings, this study introduces an innovative extension of the PPMEM framework. Through in-depth field investigations and expert interviews, the original five macro-dimensions have been reconstructed into seven refined dimensions encompassing cognition, physiology, skill, psychology, equipment, environment, and management. This restructuring enables precise capture of elements specifically relevant to rail transit emergency scenarios. The integration of the “Physiological,” “Psychological,” and “Management” dimensions facilitates a more systematic and profound exploration of the underlying mechanisms influencing safety performance within complex sociotechnical systems, thereby providing an innovative practical exemplar for deepening the application of traditional system safety theories in specific industrial contexts.

In summary, existing research on the emergency response capabilities of urban rail train drivers remains inadequate, lacking a comprehensive and systematic evaluation system, and falling short in providing objective assessment factors. Therefore, this study innovatively proposes a multidimensional evaluation framework based on the Physio-Psycho-Machine-Environment-Management (PPMEM) model, establishing a system of influencing factors comprising 7 dimensions and 30 observed variables. This system enables an objective, quantitative, and comprehensive assessment of urban rail train drivers’ emergency handling capabilities. It can provide data support for driver selection, job assignment, and shift scheduling, facilitating the scientific allocation of human resources and enhancing operational safety.

2 Materials and methods

2.1 Study subjects

This study selected 566 employees from Kunming Rail Transit Operations Co., Ltd. to conduct a questionnaire survey, collecting data on the influence of physio-psycho-machine-environment-management multidimensional factors on train drivers’ emergency handling capability. The survey was carried out with approval from the company, and all participants provided informed consent, with assurance that their personal information would remain confidential and not be disclosed in any published results. The research team included rolling stock engineers, train driver team leaders, operational technology specialists, depot managers, and electric train drivers—with the latter constituting the majority of respondents. The driver participants covered a range of professional seniority, all possessing valid train operation qualifications and the ability to work independently. Their driving experience varied from 2 to 10 years. Data were collected through both online and offline channels. A total of 566 questionnaires were distributed, all of which were returned, resulting in 538 valid responses—a validity rate of 95.05%, which meets the standard requirement for social science survey data. Descriptive statistics of the valid sample are presented in Table 1.

Table 1

Table 1. Descriptive statistics of sample data.

2.2 Connotation of emergency handling capability

The emergency response capability of train drivers refers to their comprehensive competence in rapidly identifying risks, making accurate decisions, and effectively executing countermeasures during unexpected operational events, such as equipment failures, passenger emergencies, or natural disasters (Zhu et al., 2024). This capability encompasses not only technical operational proficiency but also the synergistic integration of cognitive, psychological, and physiological dimensions (Xie et al., 2021). The PPMEM (Physiological-Psychological-Machine-Environment-Management) system developed in this study represents an adaptive construct that inherits and refines the HMEM (Human-Machine-Environment-Management) framework, specifically tailored to the operational safety characteristics of urban rail transit. The PPMEM framework fully preserves the four core system tiers of HMEM—Human, Machine, Environment, and Management—maintaining its inherent systemic analytical perspective. To better align the framework with the context of driver capability assessment, we integrated and refined the original HMEM dimensions. Specifically, the “Human” dimension was deconstructed to emphasize its distinct “Physiological” and “Psychological” attributes, thereby forming the five foundational pillars of the PPMEM framework. Based on the Physio-Psycho-Machine-Environment-Management multidimensional framework, this study constructs a “Human-Machine-Environment-Management” (HMEM) system. Its core components include: At the system interaction level: the ability to dynamically capture multi-source abnormal signals for risk perception, the capacity to anticipate cascading effects under high-pressure situations for decision-making, and precision in human-machine collaborative operations; At the capability support level: physiological and psychological stability under stress (e.g., maintaining operational accuracy and managing cognitive bandwidth), as well as the ability to transfer knowledge to atypical scenarios; At the system support level: reliance on executable emergency plans, human factors engineering design, and multi-agent coordination mechanisms.

Essentially, emergency response capability positions the train driver as the core decision-making node within the system, operating through a four-level response loop: “physiological function → cognitive processing → technical operation → system interaction.” This process demonstrates dynamic adaptability within the critical response window of ≤3 min. Key characteristics include: dynamic adaptation to evolving incident phases, systemic balance under multiple stressors, cognitive flexibility beyond predefined procedures, and goal-oriented operational recovery. The ultimate effectiveness is validated through closed-loop verification using quantitative indicators such as system disturbance absorption rate and recovery cost control.

2.3 Dimensional analysis of influencing factors

The emergency handling capability of train drivers constitutes a complex and integrated system, centered on the effective synthesis of both technical and non-technical skills during unexpected events. Since drivers already possess basic operational skills before qualification, non-technical factors—such as physiological, psychological, and cognitive elements—significantly influence the application of those skills. Accordingly, this study incorporates cognitive, physiological, skill-based, psychological, equipment-related, environmental, and managerial dimensions to achieve a precise assessment.

Dimension 1: Cognitive Factors. Cognitive factors primarily involve information processing and decision-making (Alzayed and Alsardi, 2025). Drivers must rapidly identify abnormal signals through visual scanning and pattern recognition in complex environments. Maintaining working memory stability under time pressure, accurately interpreting situations, and predicting potential developments are essential for high-quality decision-making.

Dimension 2: Physiological Factors. Physiological factors pertain to the impact of physical condition on emergency response (Bortkiewicz et al., 2022). Fatigue from prolonged duty significantly increases reaction time; studies show that braking response delays become notable after more than 4 h of continuous operation. Circadian rhythms also affect alertness levels. Maintaining optimal physiological condition is fundamental to effective emergency handling.

Dimension 3: Skill Factors. This dimension includes both proficiency in executing standard operating procedures and the ability to manage unanticipated scenarios. Drivers must be adept at handling emergency equipment—for example, switching to degraded operation modes during ATP system failures. They should also apply professional knowledge creatively in unpracticed situations.

Dimension 4: Psychological Factors. Psychological factors focus on emotion regulation and stress coping during emergencies. Critical incidents trigger stress responses, observable in physiological indicators such as heart rate variability. Drivers with high psychological resilience recover more quickly and maintain sound judgment. Research indicates that elevated anxiety levels correlate with increased operational errors (Bejar et al., 2021; Marín-Berges et al., 2025).

Dimension 5: Equipment Factors. Equipment performance and interface design directly shape a driver’s emergency response. The rationality of human-machine interface design affects information acquisition efficiency; the timeliness of alarm systems influences response speed; and the reliability of auxiliary driving systems underpins operational safety (Biassoni et al., 2016). Clarity of auditory alerts and coordination of multi-modal warnings also affect judgment accuracy, while maintenance quality determines system stability.

Dimension 6: Environmental Factors. Environmental influences include both physical and operational conditions (Ma et al., 2024). Visibility in tunnel sections, track curvature, ambient noise, and lighting conditions affect perception and judgment. Passenger density and distribution alter the complexity of emergencies, while weather conditions and time of operation introduce additional challenges.

Dimension 7: Managerial Factors. Effective management systems are essential for supporting drivers’ emergency capabilities. Scientific training programs, regular drills, psychological support mechanisms, and dynamically updated emergency plans together form an institutional foundation for effective response.Based on the above multidimensional analysis, the influencing factors on train drivers’ emergency handling capability are summarized into seven categories: cognitive, physiological, skill-based, psychological, equipment-related, environmental, and managerial, along with their sub-factors. A conceptual model of these influencing relationships is illustrated in Figure 1.

Figure 1

Figure 1. Influencing mechanism of train drivers’ emergency response capability.

Building upon the HMEM framework and informed by characteristic analysis of factors influencing train drivers’ emergency response capabilities, we have delineated seven distinct dimensions: The “Human” dimension was operationalized as Physiological Factors and Psychological Factors to precisely capture drivers’ internal states. The “Machine” dimension was specified as Equipment Factors. The “Environment” dimension was differentiated into Environmental Factors and Cognitive Factors, reflecting the human cognitive processing of environmental information. The “Management” dimension was refined into Management Factors and Skill Factors, representing the direct manifestation of management and training outcomes at the individual level. Figure 2 illustrates this evolutionary pathway, clearly depicting the complete logical relationships from the foundational HMEM framework to the refined PPMEM construct and subsequently to the seven specific assessment dimensions.

Figure 2

Figure 2. Conceptual diagram of the logical relationships among HMEM, PPMEM, and the seven dimensions.

2.4 Structural equation modeling framework

The construction of the evaluation system indicators is guided by four core principles: Scientific rigor: Indicators must reflect the essential characteristics of the object being evaluated, supported by reliable theories and data. Systematicness: Coverage of multidimensional elements and their interrelationships to avoid one-sidedness. Operationalizability: Ensuring data accessibility, feasible measurement methods, and reasonable costs. Goal orientation: Close alignment with the core objectives of the evaluation, emphasizing key performance metrics. The overall framework seeks to balance theoretical rigor with practical applicability, ensuring the scientific validity and effectiveness of the indicator system. The principles of Scientificity, Systematicity, and Operability were derived from core literature in research design and scale development, following established paradigms for systematic scale construction (Churchill, 1979) and scale development guidelines (Anderson et al., 2023). These represent universally recognized standards for developing high-quality measurement instruments. The Goal-Oriented Principle was explicitly proposed and emphasized by our research team based on the specific context of this study, informed by expert interviews. This principle ensures all indicators are not only methodologically sound but also directly serve the ultimate research objective of “assessing and enhancing drivers’ emergency response capability,” thereby maintaining alignment between the indicators and core research goals.

The development of evaluation indicators for train drivers’ emergency response capability originated from a systematic deconstruction of capability components integrated with operational safety requirements. Beyond professional skills, significant attention was given to implicit non-technical factors. Based on literature analysis (Zhu et al., 2023a; Zhu et al., 2023b; Yi et al., 2023; You et al., 2021) and in compliance with national and industry standards, the secondary indicators were systematically categorized. To ensure the identified influencing factors were both accurate and representative, these factors were further refined through field investigations and expert interviews. Through analytical processing, the factors were ultimately categorized into seven dimensions.

For ease of reference, each influencing factor is assigned a capital letter code, as detailed in Table 2.

Table 2

Table 2. Influencing factors of urban rail train drivers’ emergency response capability.

Structural Equation Modeling (SEM) is a multivariate statistical technique that integrates factor analysis and path analysis, enabling the simultaneous examination of complex relationships between observed and latent variables. In evaluating the emergency response capability of urban rail transit train drivers, SEM treats abstract constructs—such as “cognitive ability,” “psychological quality,” and “operational skill”—as latent variables, and establishes quantitative relationships based on questionnaire data. This study employs Amos 26 Graphics software for SEM construction and analysis. In accordance with SEM conceptual requirements and the operational specifications of Amos 26 Graphics, the model was developed according to the stepwise procedure illustrated in Figure 3.

Figure 3

Figure 3. Structural equation modeling (SEM) development procedure.

Prior to model construction, an initial theoretical model was established based on the aforementioned research. The number of factors and their interrelationships were determined according to the identified influencing factors. A path diagram was subsequently drafted to represent hypothesized relationships among latent variables, and was iteratively adjusted to enhance interpretability and theoretical coherence, thereby ensuring the model effectively explains variable meanings.

2.5 Research hypotheses

According to structural equation modeling (SEM) theory, model construction consists of two main components: the measurement model and the structural model. The measurement model describes the relationships between observed variables and latent variables, while the structural model captures the relationships among the latent variables themselves.

Model 1

Model 1. Theoretical Model of Latent Variables.

Figure 4

Figure 4. Latent variable model and measurement model of endogenous latent variables. (a) Latent variable model. (b) Measurement model of endogenous latent variables.

Model 2

Model 2. Measured Variables and Measurement Model.

Figure 5

Figure 5. Measurement model of exogenous latent variables.

3 Results

3.1 Questionnaire development

The design of the questionnaire in this study is grounded in empirical evidence obtained during a preliminary systematic research phase. Through in-depth field investigations at the operational frontline of Kunming Rail Transit and structured interviews with experts—including safety managers, senior instructors, and highly experienced drivers—we systematically identified and defined the key factor indicator system influencing emergency response capability. This first-hand information, gathered directly from practice, forms the core foundation for the questionnaire’s content validity and concretely reflects the industry’s central concerns and operational requirements regarding safety and quality management. Subsequently, the analytical findings were translated into specific items following established scale development protocols from psychological and behavioral research, employing a structured self-report format. The final instrument underwent rigorous reliability and validity tests, thereby ensuring the scientific soundness and reliability of the measurement tool. A questionnaire was designed to investigate the influence of latent variables—derived from the “Human-Machine-Environment-Management” framework—on the emergency response capability of urban rail train drivers. The questionnaire was structured to collect data corresponding to each observed variable through targeted questions. It comprised two main sections: basic demographic information of the respondents and specific items related to the research variables, enabling comprehensive data collection for statistical analysis.

Given that the structural equation model includes seven exogenous latent variables, measurable factors associated with each latent variable were defined to quantify their interrelationships. Building on the observed variables identified in previous analyses, the questionnaire was designed to facilitate empirical investigation. All items were formatted using a 5-point Likert scale. The demographic section captured basic information such as name, age, gender, and years of experience. The main survey items were developed based on the research variables to gather feedback on drivers’ emergency response capabilities.

3.2 Descriptive statistics of the sample

The collected data were processed and analyzed according to standard statistical procedures. Responses were coded numerically: “Strongly Disagree” = 1, “Disagree” = 2, “Neutral” = 3, “Agree” = 4, and “Strongly Agree” = 5.

Descriptive statistics of the sample are presented in Table 3. The skewness of all measured items ranged from −2.103 to 0.705, and kurtosis values fell between −0.305 and 5.782. Since the absolute values of skewness were less than 3 and those of kurtosis were below 10, the data were considered suitable for further analysis. The mean score for emergency response capability was 4.57 ± 0.48. The scores for each dimension were as follows: cognitive factors (4.57 ± 0.48), physiological factors (4.54 ± 0.47), skill factors (4.54 ± 0.49), psychological factors (4.53 ± 0.46), equipment factors (4.45 ± 0.71), environmental factors (4.47 ± 0.46), and managerial factors (4.48 ± 0.45). All values met the prerequisites for subsequent analytical procedures. Detailed descriptive statistics for each dimension are provided in Table 3.

Table 3

Table 3. Score distribution of variables.

3.3 Reliability and validity testing

In this study, Cronbach’s alpha coefficient was employed to assess the internal consistency reliability of the questionnaire. This method is widely used for evaluating the reliability of surveys measuring attitudes or opinions. The formula for calculating Cronbach’s alpha is as follows:

$\alpha =\frac{K}{K-1}\left(1-\frac{{\displaystyle {\sum }_{Y_i}^k}{\sigma }_{Y_i}^2}{{\sigma }_X^2}\right)$

where: K—number of items, ${\sigma }_X^2$—variance of the total sample scores, ${\sigma }_{Y_i}^2$—variance of individual measurement items

Reliability analysis results for the survey data are presented in Table 4.

Table 4

Table 4. Reliability test results of the sample.

The overall Cronbach’s α coefficient for the 34 variables included in the questionnaire was 0.980. As shown in Table 4, the Cronbach’s α values for all individual variables exceeded the threshold of 0.8, indicating high internal consistency and satisfactory reliability among the 30 measured variables.

Content Validity: Theoretical analysis confirms that the items included in the questionnaire and measurement scales were derived from key modules and processes reflecting the target construct —“urban rail train drivers’ emergency response capability.” The selected items comprehensively cover the major aspects of the measured construct. Furthermore, the measurement items were developed and refined through field investigations, theoretical analysis, and interviews with professional technicians and management staff. Thus, the questionnaire demonstrates strong content validity.

Construct Validity: To evaluate the construct validity of the measured variables and the theoretical model, the Kaiser–Meyer–Olkin (KMO) measure and Bartlett’s test of sphericity were applied. The results showed a KMO value of 0.908, and Bartlett’s test reached statistical significance (p < 0.05), confirming the suitability of the data for factor analysis. Subsequent factor loading analysis revealed that all variables had factor loadings exceeding 0.6, indicating strong associations between the items and their corresponding factors. Therefore, the scale demonstrates well-designed independent and mediating variables, with both reliability and validity meeting acceptable standards.

3.4 Path structure of the structural equation model

Based on the data analysis, the structural equation model was fitted according to modeling requirements. Using the theoretical and measurement models of urban rail train drivers’ emergency response capability improvement, along with the interrelationships among the seven exogenous latent variables, Amos 26 Graphics was employed as the analytical tool to construct the structural equation model in line with the research framework and hypotheses. The initial analysis revealed that the chi-square/degrees of freedom (χ²/df) ratio of the original model exceeded 3. Although indices such as the Root Mean Square Error of Approximation (RMSEA), Root Mean Square Residual (RMR), Comparative Fit Index (CFI), and Incremental Fit Index (IFI) fell within acceptable ranges, the Goodness-of-Fit Index (GFI) and Tucker–Lewis Index (TLI) were below 0.9 but still within tolerable limits. Consequently, the model was modified. The revised structural equation model is shown in Figure 6.

Figure 6

Figure 6. Modified structural equation model.

The modified model demonstrated a satisfactory overall fit, with a χ²/df ratio of 2.833 (within the recommended range of 1–3), RMSEA = 0.047 (<0.08), RMR = 0.016 (<0.08), and GFI, CFI, IFI, and TLI all exceeding 0.9. These results indicate that the modified structural equation model for improving drivers’ emergency response capability is well-adapted and shows a strong alignment with the empirical data. Detailed fit indices are presented in Table 5.

Table 5

Table 5. Goodness-of-fit test of the modified model.

By importing the processed and adjusted survey data into the software, a standardized analysis of the modified structural equation model was conducted. The resulting path diagram is shown in Figure 7. Analysis using Amos 26 Graphics indicated that all seven latent variables exhibited positive standardized regression coefficients, confirming positive correlations among them (Table 6). The standardized regression coefficients reflect the strength of association between observed variables—higher values indicate stronger relationships between the corresponding factors.

Figure 7

Figure 7. Standardized SEM analysis results.

Table 6

Table 6. Path coefficients and significance tests for latent-measured variables in the modified model.

3.5 Analysis of results

As shown in Table 6, the seven exogenous latent variables exert varying degrees of influence on the target endogenous latent variable—“driver emergency response capability”—with total effects of 0.105, 0.212, 0.105, 0.214, 0.126, 0.205, and 0.102, respectively. The magnitude of influence based on standardized path coefficients can be classified into three levels: Highly significant: standardized path coefficient ≥0.3; Moderately significant: standardized path coefficient between 0.2 and 0.3; Minor influence: standardized path coefficient <0.2. Among the exogenous latent variables analyzed in this study, physiological factors, psychological factors, and environmental factors exhibit moderately significant influences on emergency response capability. In contrast, cognitive factors, skill factors, equipment factors, and managerial factors show minor influences. Therefore, all seven hypothesized factors are confirmed to positively affect drivers’ emergency response capability.

A weighted method was applied to analyze the measured variables. The degree of influence of each factor was calculated by multiplying the path coefficients of the 30 measured influencing factors by the total effect values of their corresponding endogenous latent variables. The computed influence values are summarized in Table 7. For example, to calculate the influence value of the latent variable “AS1 Abnormal Signal Identification Speed” on “Emergency Response Capability,” the path coefficients from the standardized structural equation model (Figure 7) were referenced. The influence logic follows: “AS1 Abnormal Signal Identification Speed → A Cognitive Factors”, “A Cognitive Factors → H Emergency Response Capability”. Using the corresponding standardized path coefficients, the influence value is calculated as:0.853 × 0.105 = 0.09. This method was applied iteratively to determine the influence values of all 30 factors, as detailed in Table 7.

Table 7

Table 7. Effect magnitudes of standardized path coefficients.

The total effects of the seven exogenous latent variables on emergency response capability, in descending order, are: Physiological factors (0.214), Psychological factors (0.212), Environmental factors (0.204), Equipment status (0.126), Skill factors (0.105), Cognitive factors (0.105), Managerial factors (0.102). Key observations from the influence levels and rankings include: (1) Psychological factors exhibit the strongest influence (total effect: 0.214), with psychological resilience and stress stability (factor loadings >0.8) being the most impactful sub-dimensions; (2) Physiological factors show a nearly identical influence (total effect: 0.212), where endurance during continuous operation and circadian adaptability (factor loadings >0.8) are the dominant sub-factors; (3) Environmental factors also play a critical role (total effect: 0.205), with slope interference, curve interference, and lighting conditions (factor loadings >0.8) being the most influential. (4) Although cognitive, equipment, skill, and managerial factors demonstrated relatively minor overall influences on individual drivers, with total effect values of 0.105, 0.126, 0.105, and 0.102 respectively (all below the 0.2 threshold), several sub-factors with notably high factor loadings (exceeding 0.8) warrant particular attention. These include abnormal signal identification speed, decision-making accuracy, and timeliness of solution selection among cognitive factors; cab layout design, external signage guidance, and fault feedback performance among equipment factors; and years of cumulative experience, response capability to unexpected events, and safety risk awareness among skill factors.

4 Discussion

This study developed a comprehensive and practical physio-psycho-machine-environment-management (PPMEM) evaluation index system through field investigations and expert interviews. By incorporating objective physiological and psychological factors into the assessment framework for urban rail train drivers’ emergency response capabilities, we established a multidimensional evaluation system. Using structural equation modeling (SEM), the influencing factors were objectively and quantitatively evaluated. Empirical analysis based on data from Kunming Metro Operation Co., Ltd. revealed significant differences in the key factors affecting drivers’ emergency response capabilities. The results indicate that psychological, physiological, and environmental factors are the most critical determinants, with total effect values all exceeding 0.2 and being highly comparable. In contrast, equipment status, skill proficiency, cognitive ability, and management systems demonstrated relatively lower total effects (all below 0.13), positioning them as secondary influencing factors. These findings provide both theoretical and practical guidance for enhancing emergency management systems in urban rail transit.

The findings of this study not only validate the relationships between various influencing factors and emergency response capability but, more importantly, provide strong empirical support for the expansion and optimization of the HMEM framework into the seven-dimensional PPMEM model. First, our results reveal that psychological (β = 0.214) and physiological (β = 0.212) factors constitute the core drivers influencing emergency capability, with effect sizes significantly surpassing most other dimensions. This finding is particularly illuminating: it confirms the necessity and validity of deconstructing the broadly defined “Human” factor in HMEM into two distinct, operationalizable dimensions of “Psychological” and “Physiological” attributes. Had the original framework been retained, quantifying the differential impacts of these intrinsic characteristics on driver performance would have been challenging, potentially obscuring the most critical influencing mechanisms. Second, the results demonstrate an indirect pathway through which “Management” factors (β = 0.102) exert influence via “Skill” factors (β = 0.105). This precisely validates another key refinement we made to the HMEM framework—the explicit integration of “Skill” as a constitutive element within the “Management” dimension. It indicates that management effectiveness is not directly manifested but is realized through shaping executable driver competencies, a nuanced pathway relationship that the original HMEM framework could not adequately delineate. In summary, the empirical evidence from this study substantiates that the refinement and reconstruction of the HMEM framework, based on the operational characteristics of rail transit, represent not merely a theoretical improvement but a crucial enhancement to the model’s explanatory and predictive power. This refinement successfully transforms a macro-level system safety framework into an effective assessment model capable of precisely diagnosing deficiencies in metro drivers’ emergency response capabilities and guiding targeted interventions.

The analysis suggests that the sudden, high-pressure, and time-sensitive nature of emergency scenarios places paramount importance on psychological readiness, physiological condition, and environmental adaptability, which directly shape a driver’s instantaneous response and decision-making capacity. These constitute irreplaceable immediate human-factor elements (Dantas et al., 2021; Wu et al., 2024). In contrast, equipment, skills, cognition, and management exert their influence indirectly through long-term training or systemic optimization, and their effects are often attenuated in acute emergencies, resulting in their secondary roles. Ultimately, emergency response capability hinges critically on the real-time interaction between the human operator and the environment, rather than on pre-controlled physical or institutional conditions (Liu et al., 2019; Wang D. et al., 2020; Zhou et al., 2020). However, most metro operators still emphasize long-term training and system optimization, paying insufficient attention to these immediate human factors. We recommend that metro companies prioritize factors that directly affect drivers’ stress response speed and decision stability, enabling more scientific and efficient human resource management and ensuring operational safety.

A limitation of this study is its focus on data from a single metro operator. Although it systematically reveals the influencing factors within the “Human-Machine-Environment-Management” framework, future research should incorporate comparisons across different cities considering variations in network complexity, equipment systems, and management cultures. Multi-regional comparative studies would further refine the understanding of drivers’ emergency response capabilities.

Furthermore, this study primarily focuses on developing a holistic model of factors influencing emergency response capability. It did not investigate potential perceptual differences across driver subgroups within this model. Future research should employ specifically designed sampling strategies and statistical methods, such as multi-group analysis, to uncover potential moderating effects. Such investigations would provide a robust foundation for developing personalized training programs and targeted management interventions.

5 Conclusion

This study constructed a comprehensive and practical physio-psycho-machine-environment-management evaluation index system, incorporating objective physiological and psychological factors into the assessment of urban rail train drivers’ emergency response capabilities. The framework enables a deeper understanding of the main influencing factors. Using a combination of empirical data and analytical methods, we weighted seven latent variables and 30 specific factors. The results demonstrate that psychological, physiological, and environmental factors are the primary influences, while equipment status, skill proficiency, cognitive ability, and management systems are secondary factors. These findings offer a theoretical foundation and practical guidance for strengthening emergency management systems in urban rail transit operations.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement

Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study. The studies were conducted in accordance with the local legislation and institutional requirements.

Author contributions

JY: Conceptualization, Data curation, Investigation, Methodology, Software, Writing – original draft, Writing – review and editing. JH: Formal Analysis, Project administration, Supervision, Writing – original draft, Writing – review and editing. WL: Resources, Supervision, Validation, Writing – review and editing. XH: Data curation, Methodology, Writing – review and editing. PL: Conceptualization, Investigation, Writing – review and editing.

Funding

The author(s) declared that financial support was received for this work and/or its publication.

Acknowledgements

We would like to thank the metro drivers of Kunming Rail Operation Co., Ltd., in Yunnan Province of China for supporting this survey, and we would also like to thank the team for their support and help in writing the manuscript.

Conflict of interest

Author WL was employed by Kunming Rail Transit Operation Co., Ltd.

The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

This work was supported by the Kunming Metro Operation Co., Ltd. 2024 Mental Health Assessment and Intervention Program for Electric Train Drivers (Project No.KDY-CW-FW-E-2024-292). The funder had the following involvement in the study: providing and organizing the raw data, as well as participating in the study design, data collection, supervision, and manuscript review processes.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Alzayed, M. A., and Alsardi, N. (2025). Dispatch under pressure: an investigation into the cognitive load of kuwait’s emergency responders. J. Eng. Res. doi:10.1016/j.jer.2025.06.004

CrossRef Full Text | Google Scholar

Anderson, L. S., McCallum, R. S., Castleman, D. M., and Fast, M. (2023). Development and validation of the scale of emotional functioning: Education (SEF:ED). Psychol. Sch. 60 (8), 2692–2716. doi:10.1002/pits.22911

CrossRef Full Text | Google Scholar

Andreou, E., Gagliardini, P., Ghysels, E., and Rubin, M. (2025). Spanning latent and observable factors. J. Econ. 248, 105743. doi:10.1016/j.jeconom.2024.105743

CrossRef Full Text | Google Scholar

Bejar, M., Regaieg, N., Gdoura, D., Aloulou, J., and Amami, O. (2021). Anxious driving behavior among taxi drivers. Eur. Psychiatry 64 (S1), S184–S185. doi:10.1192/j.eurpsy.2021.488

CrossRef Full Text | Google Scholar

Biassoni, F., Ruscio, D., and Ciceri, R. (2016). Limitations and automation. The role of information about device-specific features in ADAS acceptability. Saf. Sci. 85, 179–186. doi:10.1016/j.ssci.2016.01.017

CrossRef Full Text | Google Scholar

Bortkiewicz, A., Gadzicka, E., and Viebig, P. (2022). Physiological reaction to the night work in truck drivers. Saf. Health A. T. Work 13, S17–S18. doi:10.1016/j.shaw.2021.12.757

CrossRef Full Text | Google Scholar

Churchill, G. A. (1979). A paradigm for developing better measures of marketing constructs. J. Mark. Res. 16 (1), 64–73. doi:10.1177/002224377901600110

CrossRef Full Text | Google Scholar

Collins, M. D. (2023). Train driver selection: the impact of cognitive ability on train driving performance. Int. J. Sel. Assess. 32 (2), 249–260. doi:10.1111/ijsa.12461

CrossRef Full Text | Google Scholar

Dantas, A., Banh, D., Heywood, P., and Amado, M. (2021). Decoding emergency settlement through quantitative analysis. Sustainability 13 (24), 13586. doi:10.3390/su132413586

CrossRef Full Text | Google Scholar

Davidich, Y., Földes, D., and Galkin, A. (2025). Enhancing emergency response efficiency through advanced urban logistics: the role of driver psychophysiology and vehicle dynamics in mitigating socio-economic impacts. Transp. Res. Interdiscip. Perspect. 31, 101464. doi:10.1016/j.trip.2025.101464

CrossRef Full Text | Google Scholar

Doi, T. (2024). The relationship between the living environment and remote working: an analysis using the SHEL model. PeerJ 12, e17301. doi:10.7717/peerj.17301

PubMed Abstract | CrossRef Full Text | Google Scholar

Fan, H., Enshaei, H., and Jayasinghe, S. G. (2022). Human error probability assessment for LNG bunkering based on fuzzy bayesian Network-CREAM model. J. Mar. Sci. Eng. 10 (3), 333. doi:10.3390/jmse10030333

CrossRef Full Text | Google Scholar

Kashevnik, A., Ponomarev, A., Shilov, N., and Chechulin, A. (2022). Threats detection during human-computer interaction in driver monitoring systems. Sensors 22 (6), 2380. doi:10.3390/s22062380

PubMed Abstract | CrossRef Full Text | Google Scholar

Klocek, A., Premus, J., and Řiháček, T. (2023). Applying dynamic systems theory and complexity theory methods in psychotherapy research: a systematic literature review. Psychotherapy Res. 34 (6), 828–844. doi:10.1080/10503307.2023.2252169

PubMed Abstract | CrossRef Full Text | Google Scholar

Kovesdi, C., Spielman, Z., LeBlanc, K., and Rice, B. (2018). Application of eye tracking for measurement and evaluation in human factors studies in control room modernization. Nucl. Technol. 202 (2-3), 220–229. doi:10.1080/00295450.2018.1455461

CrossRef Full Text | Google Scholar

Kroll, V., Mackenzie, A. K., Goodge, T., Hill, R., Davies, R., and Crundall, D. (2020). Creating a hazard-based training and assessment tool for emergency response drivers. Accid. Analysis and Prev. 144, 105607. doi:10.1016/j.aap.2020.105607

PubMed Abstract | CrossRef Full Text | Google Scholar

Leicher, V., and Mulder, R. H. (2016). Individual and contextual factors influencing engagement in learning activities after errors at work. J. Workplace Learn. 28 (2), 66–80. doi:10.1108/jwl-03-2015-0022

CrossRef Full Text | Google Scholar

Li, H., Wang, Y., Jiang, J., and Zhou, R. (2022). Metro station evacuation safety assessment considering emergency response. SIMULATION 98 (10), 919–931. doi:10.1177/00375497221095071

CrossRef Full Text | Google Scholar

Lin, C., Xu, Q. f., and Huang, Y. f. (2022). An HFM-CREAM model for the assessment of human reliability and quantification. Qual. Reliab. Eng. Int. 38 (5), 2372–2387. doi:10.1002/qre.3081

CrossRef Full Text | Google Scholar

Liu, X., Wang, Z., Zhang, S., and Liu, J. (2019). Analysis of influencing factors in emergency management based on an integrated methodology. Adapt. Behav. 27 (5), 331–345. doi:10.1177/1059712319858623

CrossRef Full Text | Google Scholar

Lu, Y., and Sun, S. (2020). Scenario-based allocation of emergency resources in metro emergencies: a model development and a case study of nanjing metro. Sustainability 12 (16), 6380. doi:10.3390/su12166380

CrossRef Full Text | Google Scholar

Lv, C., Nian, J., Xu, Y., and Song, B. (2022). Compact vehicle driver fatigue recognition technology based on EEG signal. IEEE Trans. Intelligent Transp. Syst. 23 (10), 19753–19759. doi:10.1109/tits.2021.3119354

CrossRef Full Text | Google Scholar

Ma, Y., and Yao, H. (2025). Research on understandability and cognitive load equilibrium of human-machine interface under time pressure. Displays 89, 102891. doi:10.1016/j.displa.2024.102891

CrossRef Full Text | Google Scholar

Ma, J., Wang, Y., Xia, M., Guo, Z., Li, Z., Zhang, J., et al. (2024). The influence of train driver's foreign body penetration experience on hazard perception sensitivity: the mediating role of sustained attention. Cognition, Technol. and Work 26 (1), 153–168. doi:10.1007/s10111-023-00744-4

CrossRef Full Text | Google Scholar

Marín-Berges, M., Villa-Berges, E., Lizana, P. A., Gómez-Bruton, A., and Iguacel, I. (2025). Depression, anxiety and stress in taxi drivers: a systematic review of the literature. Int. Archives Occup. Environ. Health 98 (1), 135–154. doi:10.1007/s00420-024-02117-4

PubMed Abstract | CrossRef Full Text | Google Scholar

McClintock, P. V. E. (2019). Introduction to the theory of complex systems. Contemp. Phys. 60, 318–319. doi:10.1080/00107514.2019.1663936

CrossRef Full Text | Google Scholar

McWhirter, L., King, L., McClure, E., Ritchie, C., Stone, J., and Carson, A. (2021). The frequency and framing of cognitive lapses in healthy adults. CNS Spectrums 27 (3), 331–338. doi:10.1017/s1092852920002096

PubMed Abstract | CrossRef Full Text | Google Scholar

Metros, C. (2024). China association of metros.

Google Scholar

Perboli, G., Gajetti, M., Fedorov, S., and Giudice, S. L. (2021). Natural language processing for the identification of human factors in aviation accidents causes: an application to the SHEL methodology. Expert Syst. Appl. 186, 115694. doi:10.1016/j.eswa.2021.115694

CrossRef Full Text | Google Scholar

Santis, E. D., Martino, A., and Rizzi, A. (2024). Human versus machine intelligence: assessing natural language generation models through complex systems theory. IEEE Trans. Pattern Analysis Mach. Intell. 46 (7), 4812–4829. doi:10.1109/tpami.2024.3358168

CrossRef Full Text | Google Scholar

Sun, M. (2020). Evaluation criteria and application of emergency handling ability for metro train drivers. Urban Mass Transit 23 (12), 143–147. doi:10.16037/j.1007-869x.2020.12.029

CrossRef Full Text | Google Scholar

Tolani, D. K., Ray, A., and Horn, J. F. (2006). Integrated decision and control of human-engineered complex systems. Int. J. General Syst. 35 (3), 275–294. doi:10.1080/03081070600660962

CrossRef Full Text | Google Scholar

Wang, Y.-T., Huang, K.-C., Wei, C.-S., Huang, T.-Y., Ko, L.-W., Lin, C.-T., et al. (2014). Developing an EEG-Based on-line closed-loop lapse detection and mitigation system. Front. Neurosci. 8, 321. doi:10.3389/fnins.2014.00321

PubMed Abstract | CrossRef Full Text | Google Scholar

Wang, D., Wan, K., and Ma, W. (2020a). Emergency decision-making model of environmental emergencies based on case-based reasoning method. J. Environ. Manag. 262, 110382. doi:10.1016/j.jenvman.2020.110382

PubMed Abstract | CrossRef Full Text | Google Scholar

Wang, T., Chen, Y., Yan, X., Li, W., and Shi, D. (2020b). Assessment of drivers comprehensive driving capability under man-computer cooperative driving conditions. IEEE Access 8, 152909–152923. doi:10.1109/access.2020.3016834

CrossRef Full Text | Google Scholar

Wang, F.-Y., Guo, J., Bu, G., and Zhang, J. J. (2022). Mutually trustworthy human-machine knowledge automation and hybrid augmented intelligence: mechanisms and applications of cognition, management, and control for complex systems. Front. Inf. Technol. and Electron. Eng. 23 (8), 1142–1157. doi:10.1631/FITEE.2100418

CrossRef Full Text | Google Scholar

Wang, Z., Du, X., and Zhang, X. (2024). Research on the construction of emergency management system based on urban rail intelligent maintenance. Mod. Urban Transit (9), 16–22. doi:10.20151/j.cnki.1672-7533.2024.09.004

CrossRef Full Text | Google Scholar

Wei, R., McDonald, A. D., Garcia, A., and Alambeigi, H. (2022). Modeling driver responses to automation failures with active inference. IEEE Trans. Intelligent Transp. Syst. 23 (10), 18064–18075. doi:10.1109/tits.2022.3155381

CrossRef Full Text | Google Scholar

Werth, B. L., and Christopher, S.-A. (2021). Potential risk factors for constipation in the community. World J. Gastroenterology 27 (21), 2795–2817. doi:10.3748/wjg.v27.i21.2795

PubMed Abstract | CrossRef Full Text | Google Scholar

Wu, Y., He, S., and Shi, J. (2024). A dynamic decision-making approach for cabin unlawful interference emergency disposal using dynamic Bayesian network. Sci. Rep. 14 (1), 19002. doi:10.1038/s41598-024-69842-1

PubMed Abstract | CrossRef Full Text | Google Scholar

Xiao, J., and Terzi, S. (2025). Large language model-guided graph convolution network reasoning system for complex human-robot collaboration disassembly operations. Procedia CIRP 134, 43–48. doi:10.1016/j.procir.2025.03.007

CrossRef Full Text | Google Scholar

Xie, H., Lin, W., Lin, S., Wang, J., and Yu, L.-C. (2021). A multi-dimensional relation model for dimensional sentiment analysis. Inf. Sci. 579, 832–844. doi:10.1016/j.ins.2021.08.052

CrossRef Full Text | Google Scholar

Xing, H., Mo, S., Liang, X., and Li, Y. (2021). Water resources allocation based on complex adaptive system theory in the inland river irrigation district. Sustainability 13 (15), 8437. doi:10.3390/su13158437

CrossRef Full Text | Google Scholar

Xu, Z., Hong, C. S., Zurita, N. F. S., Gyory, J. T., Stump, G., Nolte, H., et al. (2024). Adaptation through communication: assessing Human-AI partnership for the design of complex engineering systems. J. Mech. Des. 146 (8), 081401. doi:10.1115/1.4064490

CrossRef Full Text | Google Scholar

Yang, J., Gu, C., Liu, Z., and Yu, Z. (2023). Research on influencing factors of metro drivers' emergency handling performance. Mod. Urban Transit 1, 71–75. doi:10.20151/j.cnki.1672-7533.2023.01.014

CrossRef Full Text | Google Scholar

Yang, J., Zhang, Z., Qin, Y., Chen, Y., Cheng, Y., and Du, H. (2024). Development status and countermeasures of urban rail transit emergency drill system. Railw. Transp. Econ. 46 (3), 164–172. doi:10.16668/j.cnki.issn.1003-1421.2024.03.21

CrossRef Full Text | Google Scholar

Yang, L., Xia, M., Wang, Y., Fang, D., Jing, X., and Ma, J. (2025). Eye movement predictors of hazard response performance in train drivers under different speed conditions. Sci. Rep. 15 (1), 24024. doi:10.1038/s41598-025-09008-9

PubMed Abstract | CrossRef Full Text | Google Scholar

Yi, B., Cao, H., Song, X., Wang, J., Guo, W., and Huang, Z. (2023). How human-automation interaction experiences, trust propensity and dynamic trust affect drivers’ physiological responses in conditionally automated driving: moderated moderated-mediation analyses. Transp. Res. Part F Traffic Psychol. Behav. 94, 133–150. doi:10.1016/j.trf.2023.01.024

CrossRef Full Text | Google Scholar

You, X.-d., Zhu, L., Liu, Z.-g., and Huang, Y.-c. (2021). Experimental study on the relationship between fatigue and unsafe behavior of urban rail transit drivers. Transp. Res. Rec. J. Transp. Res. Board 2675 (10), 1151–1160. doi:10.1177/03611981211014888

CrossRef Full Text | Google Scholar

Zhao, J., and Kasim, E. (2023). Study of dynamic solutions for human–machine system with human error and common-cause failure. Mathematics 11 (12), 2771. doi:10.3390/math11122771

CrossRef Full Text | Google Scholar

Zhou, J.-L., and Guo, Z.-M. (2024). A hybrid SNN-STLSTM method for human error assessment in the high-speed railway system. Adv. Eng. Inf. 60, 102408. doi:10.1016/j.aei.2024.102408

CrossRef Full Text | Google Scholar

Zhou, D., Fan, C., and Chen, A. (2020). Evolution mechanism and driving factors of unconventional emergencies in megacities: an empirical study based on 102 cases in the world. Nat. Hazards 103 (1), 513–530. doi:10.1007/s11069-020-03998-7

CrossRef Full Text | Google Scholar

Zhu, G., Fan, J., Huang, X., Zhang, N., Wu, B., and Wei, Y. (2023a). Knowledge acquisition method of urban rail transit safety event case base for intelligent emergency response. IEEE Trans. Automation Sci. Eng. 22, 1–9. doi:10.1109/tase.2023.3337135

CrossRef Full Text | Google Scholar

Zhu, G., Huang, X., Yang, R., and Sun, R. (2023b). Relationship extraction method for urban rail transit operation emergencies records. IEEE Trans. Intelligent Veh. 8 (1), 520–530. doi:10.1109/tiv.2022.3160502

CrossRef Full Text | Google Scholar

Zhu, G., Mu, L., Sun, R., Zhang, N., Wu, B., Zhang, P., et al. (2024). Emergency control method of multi-modal passenger flow in urban rail transit. IEEE Trans. Automation Sci. Eng. 22, 1–11. doi:10.1109/tase.2023.3322031

CrossRef Full Text | Google Scholar