ORIGINAL RESEARCH article

Front. Mater., 19 October 2026

Sec. Metamaterials

Volume 13 - 2026 | https://doi.org/10.3389/fmats.2026.1980178

Optimization of multi-frequency resonant coupling acoustic package for electric motor based on a Kriging-genetic algorithm

  • 1. College of Textiles, Donghua University, Shanghai, China

  • 2. School of Mechanical and Automotive Engineering, Shanghai University of Engineering Science, Shanghai, China

  • 3. Shenda(shanghai) Technology Co., Ltd., Shanghai, China

Abstract

As the core component of electric vehicle (EV) power systems, the electric motor (E-motor) is an important source of noise, which significantly affects passenger comfort. Nevertheless, due to weight- and volume-related limitations, the design of E-motor acoustic packages faces great challenges in low-frequency broadband noise reduction. Aimming to reduce the low-frequency noise below 2000 Hz of the E-motor effectively, this paper proposes a multi-frequency resonant coupled acoustic metamaterial acoustic package (MRPAM-AP) based on the multi-frequency resonant coupling mechanism. The sound transmission loss (STL) characteristics and the effect of structural parameters on STL are analyzed by finite element method (FEM) models. To achieve optimal STL characteristics, the MRPAM-AP topology parameters are optimized using the Kriging-genetic algorithm optimization method. The noise reduction performance is validated via E-motor noise tests. The experimental results illustrate that the MRPAM-AP can significantly reduce E-motor low frequency noise. The proposed acoustic package and optimization method provide a novel approach for the industrial application of acoustic metamaterials.

1 Introduction

With the advancements in the automotive industry, electric vehicles (EVs) have gradually replaced those operating with traditional fuels. The noise level of the electric motor (E-motor), the core component of electric vehicles, severely affects market competitiveness. The noise sources of the E-motor are relatively complex, including mechanical components and electromagnetic noise, which leads to low-frequency (below 2000 Hz) and wideband noise characteristics (Xu et al., 2026; Wang et al., 2025). Noise control methods for E-motors include passive and active control methods (Ding et al., 2024; Dalela et al., 2022). Acoustic packages have been widely used in passive noise control due to their advantages of simplicity, low cost, and reliable performance. Common acoustic package structures are composed of a high-density sound insulation layer and a porous sound absorption layer with sound absorption and insulation properties among others. Nevertheless, due to the limitations of the mass law (Fahy and Gardonio, 2007), traditional acoustic packages made of homogeneous materials struggle to control low-frequency noise (Fu and Wang, 2025). Achieving both lightweight and low-frequency broadband noise barriers is difficult. Thus, exploring novel acoustic package structures with low-frequency sound insulation characteristics offers broad research and practical value.

The acoustic metamaterial theory, along with the application of local resonance acoustic metamaterial structures, has gradually attracted the attention of researchers. The elastic wave bandgap, negative stiffness, and negative mass characteristics of metamaterials can effectively overcome the mass law obstacle, thereby achieving subwavelength elastic wave control with small volume and large wavelength. Typical metamaterial structures include thin films (Jiang et al., 2023; Wang et al., 2024; Cai et al., 2023; Cai et al., 2024), thin plates (Varanasi et al., 2017; Kaidouchi et al., 2023; Ang et al., 2017; Ang et al., 2019; Wen et al., 2019), and Helmholtz resonators (Liu et al., 2020). Thin-plate acoustic metamaterial structures provide a new approach to low-frequency sound insulation, since they can control low-frequency sound waves while maintaining a small thickness.

Thin-plate acoustic metamaterial (TP-AMM) structures have evolved from single-membrane structures to various forms such as composite multilayer boards, sandwich layers, and lattice structures with complex topological configurations. For instance, Tian et al. (Tian et al., 2020) studied the sound transmission loss (STL) of sandwich-like composite laminates, and discussed the effects of thickness, bulk density, damping layer, and layup structure on sound insulation performance. Their research results indicated that increasing material thickness and density can effectively enhance sound insulation performance. Liu et al. (Liu, 2015) analyzed the STL of six different three-layer plate structures in a diffusion field using the transfer matrix method. They revealed that, under consistent quality and thickness, the STL of the three-layer board structure was better than that of the double-layer one. Xie et al. (Xie et al., 2023) proposed a sandwich structure comprising two sound absorbing layers and a sound insulation layer between them, and experimentally determined the STL of different structures. The results showed that their proposed structure achieved excellent sound insulation performance. In another study, Xia et al. (Xia et al., 2023) designed a composite structure composed of perforated plates, porous materials, thin steel plates, damping layers, and steel plates, and experimentally analyzed the sound insulation performance of structures with different layer thickness and material combinations. It was found that increasing the thickness of the porous material on the steel plate side improves the low-frequency sound insulation, while increasing the thickness of the flat plate improves the sound insulation in the high-frequency range. Wang et al. (Wang et al., 2021; Wang et al., 2023) combined double-layer metamaterial plates with porous materials and investigated their STL performance. Their results indicated that this combination can produce high sound insulation in the ultra-wide low-frequency range. In addition, low-frequency broadband sound insulation of such structures can be achieved through methods such as structural parameter optimization, multi-local resonance coupling, and multi-mechanism coupling (Zhou et al., 2020; Roca et al., 2021). It is particularly noteworthy that, due to the excellent processability and economy, TP-AMM has been introduced into the automotive industry, and has demonstrated significantly higher sound insulation performance than mass law (Tomita et al., 2026). This also presents a novel approach for the design of sound absorption and insulation structures in vehicles.

In this paper, a multi-frequency resonant coupled acoustic metamaterial acoustic package (MRPAM-AP) is designed to reduce the low frequency noise below 2000 Hz of E-motor. A STL calculation model is constructed to analyze the effects of different structural parameters. The Kriging-genetic algorithm is used to optimize the structural parameters of the acoustic package. Finally, a noise tests are conducted to verify the sound transmission characteristics of the optimized MRPAM-AP.

2 MRPAM-AP structure design and modeling

2.1 Structure and parameters

In an electric vehicle power system, an acoustic package usually wraps around the E-motor housing (Figure 1a), the MRPAM-AP structure proposed in this paper comprises three parts: a MRPAM plate, sound absorbing material and a square cover (Figure 1b). The MRPAM consists of a thin plate and three types of resonators (A, B, and C) with different structural parameters (Figure 1c). The resonators are arranged on the thin plate in a rotating manner. The MRPAM is made of thermoplastic polyurethane (TPU) and sound absorbing material is made of polyurethane foam; their properties are listed in Tables 1, 2. The MRPAM geometric parameters are detailed in Table 3.

FIGURE 1

TABLE 1

MaterialDensityYoung’s modulusPoisson’s ratio
TPU1,100 kg/m31.9 Ɨ 109 Pa0.35

MRPAM material properties.

TABLE 2

PorosityFlow resistivityTortuosity factorViscous characteristic lengthThermal characteristic length
0.9740,813 Ns/m42.933.8 Ɨ 10āˆ’5 m7.2 Ɨ 10āˆ’4 m

Sound absorbing material properties.

TABLE 3

DescriptionParameterValue
Thin plate thicknesstp1 mm
Resonator a radiusr120 mm
Resonator a heighth18 mm
Resonant element a lengthl140 mm
Resonator B radiusr211 mm
Resonator B heighth210 mm
Resonant element B lengthl240 mm
Resonator C radiusr312 mm
Resonator C heighth312 mm
Resonant element C lengthl340 mm
Square cover heightH30 mm
Square cover thicknessbc1 mm
Square cover lengthac120 mm
Sound absorbing material heighttab15 mm

MRPAM plant geometric parameters.

2.2 Sound insulation dynamic model of MRPAM

Dynamic models are widely used to characterize the sound insulation of acoustic metamaterials (AMMs). Different from homogeneous plant, the MRPAM can be equivalent to a multi-degree of freedom (Multi-DOF) system as shown in Figure 2, where M0 represents the mass of the main structure and mt represents the mass of the different resonators. Since the resonators have different sizes and masses, their spring stiffness kt differs as well. Each resonator is connected to the main structure through a spring, and the main structure is excited by a force F.

FIGURE 2

The displacements of the main structure and the resonators are x and xt, respectively; thus, the matrix u in Equation 1 can be given as:

The acceleration matrix a in Equation 2 is defined as:

Considering a single direction of motion (z-direction), the mass matrix M, stiffness matrix K, and selection vector b can be expressed as Equations 3, 4:

Therefore, the dynamic Equation 5 of the multi-DOF system MRPAM can be expressed as:

Without considering the rigidity of the plate, the equivalent stiffness of the nth resonator is Equation 6 (Fatehi et al., 2021):where E is the Young’s modulus, v is the Poisson’s ratio, hn is the radius, and rn is the height of the nth resonator.

The MRPAM plate of nth resonator element can be simplified as simply supported square thin plate (Roca et al., 2021). According to the Kirchhoff–Love theory, the equivalent stiffness of the MRPAM plate considering the contact boundary of the resonator can be expressed as Equations 7, 8:where is the correction function considering the contact boundary of scatterers, which can be obtained by mechanical experiment or FEM calculation. ln and tp are the side length and thickness of the square plate, respectively.

Then the equivalent stiffness of a single resonator in Equation 9 can be expressed as:

The local resonance frequency of the nth scatterer in Equation 10 can be obtained as:

The equivalent mass density meff in Equation 11 can be represented as:

The transmission coefficient of the structure under single-angle sound wave incidence is Equation 12:where ρ0 represents the density of air, c0 denotes the speed of sound in air, and θ is the incidence angle of sound waves.

3 STL analysis of MRPAM-AP

3.1 FEM model

The FEM model for STL was implemented through COMSOL Multiphysics 6.4. Boundary conditions of FEM model are illustrated in Figure 3a. MRPAM-AP was set in the center of the model, and input and output air fields were arranged on the upper and lower sides of the MRPAM-AP. When a plane wave with a sound pressure of Pin enters the MRPAM-AP from the input field along -Z direction, PML boundaries were set on both sides of the model in the Z direction. As the core of the model, the MRPAM plate mesh model is shown in Figure 3b.

FIGURE 3

When a plane wave with a sound pressure of Pin enters the MRPAM-AP from the incident sound field, part of the energy is reflected and dissipated by the structure, while the remaining energy is transmitted through the lower air domain. For a single-angle incident sound wave, the incident sound power Win and transmitted sound power Wout can be expressed as Equation 13:where Sin and Sout are the surfaces of the incident and transmission sides, respectively, ρ0 is the air mass density, and Pin and Pout are the incident and transmission sound pressures, respectively. The STL of the MRPAM-AP with single-angle incidence can be expressed as Equation 14:

To further verify the sound insulation effect of the MRPAM-AP, a homogeneous plate-AP with the same surface density and sound absorbing material was designed. The STL curves of the homogeneous plate-AP and MRPAM-AP structures are depicted in Figure 4. Except for the 265–455 Hz frequency range, MRPAM-AP exhibited higher acoustic transmission loss than the homogeneous plate. In the 200–1750 Hz range, the average STL of the MRPAM-AP was 14.32 dB higher than the homogeneous plate.

FIGURE 4

3.2 Effect of MRPAM-AP structural parameters on STL

From the dynamic model proposed in Section 2.2, the sound insulation characteristics of MPRAM are directly influenced by the resonator. The key structural parameters of the resonator, including resonator radius Δri and height Δhi differences, floor thickness tp, and sound absorption layer thickness tab were selected to discuss their impact on STL.

The STL exhibited a significant increasing trend in the entire frequency range with increasing Δri, and the STL peak moved to a higher frequency (Figure 5a). This is consistent with the regulatory property of MRPAM mentioned in Section 2.2. The effect of Δhi on sound insulation was complex. In the frequency range below 1,000 Hz, the STL increased with increasing Δhi. In the range above 1,000 Hz, the STL exhibited a decreasing trend with increasing Δhi. At the same time, the STL peak moved to a lower frequency and the peak value decreased as well (Figure 5b).

FIGURE 5

Figure 6 reveals the effects of tp and tab on sound insulation. With the increase of tp, the peak frequency shifted towards the low-frequency range, and below 1,225 Hz, the sound insulation increased significantly with increasing tp. Furthermore, this indicates that the square cavity resonant cover and sound-absorbing layer significantly alter the regulation mechanism of the structural parameters.

FIGURE 6

The peak STL frequency remained relatively stable and had less effect on the frequency range below 1,000 Hz with increasing tab. In the range above 1,000 Hz, the STL increased with increasing tab. This agrees with the inherent characteristics of sound-absorbing materials. Nonetheless, in industrial applications, the thickness of the acoustic package is limited; thus, tab needs to be selected according to the actual requirements.

4 Structural parameter optimization

Genetic optimization (GA) has been validated as an effective method for the optimization of metamaterials (Cao et al., 2024; He et al., 2023; Zhang et al., 2021; Wang and Liu, 2021). However, due to their global optimization characteristics, the optimization efficiency of traditional GA optimization methods is low. Combining the surrogate model with GA can effectively improve optimization efficiency. In this section, the Kriging-GA optimization method is proposed to optimize the MRPAM-AP structural parameters. The STL performance of the optimized structure is then validated by actual E-motor NVH experiments.

4.1 Kriging surrogate model

The Kriging surrogate model is an interpolation technique based on Gaussian processes (Aghaei et al., 2024; Khatouri et al., 2022; Peng et al., 2023). Its principle is to establish a variation function derived from the variation characteristics of existing observational data. This allows for the optimal and unbiased estimation of attribute values at unknown points. In addition, this model provides both an estimate of the function and the corresponding prediction error for that estimate.

The predicted values of the Kriging model in Equation 15 are represented as: where μ denotes the mean of a random process, ε(x) is a normal distribution with a mean of 0 and a variance of σ2, x=(x1 … xd).

For x and x’, the correlation function between ε(x) and ε(x’) is in Equation 16:where pi is a parameter that affects the smoothness of y(x) at x and Īøi can be used to evaluate the importance of xi to y(x).

For Nt training data (x,y) with inputs and outputs, the hyper-parameters μ, σ, Īø1 … Īød, and p1 … pd can be obtained through the maximum likelihood function in Equation 17:where y represents the output matrix of NtƗ1 and I denotes the identity matrix of Nt dimensional. The calculation Equation 18 for R is:

Then, μ and σ2 can be calculated by maximizing the likelihood function in Equations 19, 20:

The parameters μ and σ2 can be eliminated by substituting and into the likelihood function. Subsequently, the and in the maximum likelihood function can be determined. The flowchart of the Kriging surrogate model (Peng et al., 2023) is illustrated in Figure 7.

FIGURE 7

4.2 Training results

According to Latin hypercube sampling (LHS), the minimum sample size for N variables is (N+1) (N+2)/2 (Xie et al., 2025; Zaryab et al., 2025). In this section, the sample size was set as 55 for 9 variables. Table 4 lists the range of topological parameters based on the MRPAM-AP characteristics. A sampling plan consisting of 80 samples was designed (Figure 8). In this plan, the area S represents the area of the STL curve that exceeds the mass law within the target frequency range, as calculated by FEM.

TABLE 4

VariableMinimum valueMaximum value
Square resonator cover height H5 mm30 mm
Square resonator cover thickness bc1 mm5 mm
Thin plate thickness tp1 mm3 mm
Resonator a radius of r11 mm16 mm
Resonator a height of h11 mm29 mm
Resonator B radius of r21 mm16 mm
Resonator B height of h21 mm29 mm
Resonator C radius of r31 mm16 mm
Resonator C height of h31 mm29 mm

Sound package design variables and their ranges.

FIGURE 8

After training, the Kriging model was evaluated using the following parameters: coefficient of determination R2, root mean square error (RMSE), relative average absolute error (RAAE) and relative maximum absolute error (RMAE), which are determined by the following Equations 21–24:where n is the number of evaluation samples, y is the response value of a sample, is the predicted value of a sample, and represents the average response value.

After training, eight sets of samples were selected to evaluate the accuracy of the Kriging model. The results of the Kriging and FEM models are compared in Figure 9. The R2, RMSE, RAAE, and RMAE values were 0.9703, 0.0154, 0.0133, and 0.0274, respectively. When R2 ≄ 0.9, RMSE ≤ 0.2, RAAE ≤ 0.2, and RMAE ≤ 0.3 (listed in the Table 5), the model can be considered as interpretable (Xiao et al., 2012). This also means that the constructed Kriging model has sufficient accuracy.

FIGURE 9

TABLE 5

FEA valuesPredicted values
69,957.270,257.2
74,204.175,807.4
76,984.278,636.1
78,683.180,022.9
79,764.279,996.7
89,339.090,162.4
91,965.193,559.9
95,517.198,134.3

FEA and predicted values of Kriging model.

4.3 Kriging-GA optimization results

During optimization, the area exceeding the sound insulation of homogeneous plate-AP S was taken as the optimization objective, which can intuitively reflect the STL of the MRPAM-AP over a wide frequency range and comprehensively evaluate its performance. The Kriging-GA optimization flowchart is illustrated in Figure 10. The key parameters of Kriging-GA optimization are listed in Table 6. The design variables and objective function are represented as Equations 25, 26:

FIGURE 10

TABLE 6

Parameter nameCalculated valueAccept value
R20.9703≄0.9
RMSE0.0154≤0.2
RAAE0.0133≤0.2
RMAE0.0274≤0.3

Evaluation parameters of Kriging model.

The optimization process and results are revealed in Figure 11. With the increase in iterations, the generation best values rapidly rise. When iterations reach 23, the generation best values tend to be stabilized. and meet the termination condition of fitness change < 0.5% for 50 consecutive generations. The topology parameters before and after optimization are presented in Table 7.

FIGURE 11

TABLE 7

Parameter nameValue
Population size20
Crossover probability0.8
Mutation probability0.2
Maximum evolutionary generation100
Termination conditionFitness change <0.5% for 50 consecutive generations

Key parameters of Kriging-GA optimization.

Figure 12 plots the STL before and after optimization. It can be observed that, after optimizated, the MRPAM-AP exhibits higher sound insulation than homogeneous materials above 315 Hz; the absorption peaks at 415, 735, 1,090, 1,535, and 1905 Hz are indicated by arrows of different colors in Figure 12b. And the effective working frequency, where the STL is higher than that of the homogeneous plate-AP, increases from 79.4% to 92.6%, and the maximum STL peak shifts from 66.4 to 90.3 dB, demonstrating the effectiveness of the optimization. The structural parameters of the acoustic package before and after optimization are compared in Table 8.

FIGURE 12

TABLE 8

Before/after optimizedH (mm)bc (mm)tp (mm)r1 (mm)h1 (mm)r2 (mm)h2 (mm)r3 (mm)h3 (mm)
Before153110811101212
After29.53.2214.914.615.424.518.610.8

Acoustic package structural parameters before and after optimization.

4.4 Noise experimental verification

Noise experiments were conducted in a semi-anechoic chamber with a Dyno system (Horiba DYNAS3) for the E-motor. Sound collection and analysis were achieved with 4 microphones (PCB 378B02) and a data acquisition system (Siemens Simcenter SCADAS with LMS Test. LAB). The microphones were positioned at the front, top, left, and right sides of the E-motor, each at a distance of 1 m. The microphones placement is illustrated in Figures 13a,b, and the experimental site is shown in Figure 13c. At the maximum rated torque (150 N m), acoustic data were collected and analyzed for the E-motor with and without the MRPAM-AP at speeds ranging from 2000 to 5,000 rpm. The motor parameters and test conditions are listed in Table 9. The experimental results are demonstrated in Figure 14.

FIGURE 13

TABLE 9

Type of motorRated speedRated torqueTest condition (rpm)
Permanent magnet synchronous motor (PMSM)5,000 rpm150 N. m2000, 3,000, 4,000, 5,000

The motor parameters and test conditions.

FIGURE 14

After the MRPAM-AP installing, the noise level of the E-motor decreased significantly under all conditions, especially above 315 Hz (marked with blue dashed line). At the absorption peak frequencies in FEM result (marked with arrows in Figure 14), the reduction of noise is more significant (especially at 735 Hz, 1,535 Hz, and 1905 Hz). This is also consistent with the calculation results of the FEM model in Section 4.3.

Table 10 presents the 4-point average root mean square (RMS) values of the overall sound pressure level at frequencies below 2000 Hz for four different speeds, with and without the acoustic package. The RMS significantly decreased at all test speeds with the acoustic package, especially at low speeds (2000 rpm), where the RMS decreased by a maximum of 11.2 dB (A).

TABLE 10

Groups2,000 rpm3,000 rpm4,000 rpm5,000 rpm
Without MRPAM-AP/dB(A)64.1370.9271.0867.75
With MRPAM-AP/dB(A)52.9361.8163.5461.51
Difference/dB(A)11.209.117.546.06

RMS sound pressure levels for various motor speeds with and without the sound package.

5 Conclusion

In this study, an MRPAM-AP acoustic package for E-motor was proposed. The effects of different topology parameters on the STL characteristics were investigated using FEM. The Kriging-GA method was proposed to optimize the topology parameters of the MRPAM-AP, and the noise-reduction performance of the optimized acoustic package was experimentally validated. The main conclusions can be drawn as follows.

  • The MRPAM-AP structure was introduced, which comprises three parts: a MRPAM plant, sound absorbing material and a square cover. A multi-DOF dynamic model was proposed to reveal the mechanism of low-frequency sound insulation of the structure.

  • The STL characteristics of MRPAM-AP were analyzed via FEM. The MRPAM-AP structure overcame the homogeneous plate-AP limitation at frequencies above 455 Hz. The effects of different topological parameters on STL were analyzed. The results indicated that the structural parameters had significant and complex effects on STL characteristics.

  • A Kriging-GA optimization method was employed to optimize the topological parameters. The relationship between topological parameters and STL performance was effectively modeled using the Kriging surrogate model. After optimization, the effective working frequency range (STL higher than the mass law) and STL peak increased significantly.

  • The sound reduction characteristics were validated via E-motor NVH experiments. The optimized MRPAM-AP achieved effective noise reduction above 248 Hz, and the maximum sound insulation reached 11.2 dB(A). This indicates the effectiveness of the proposed structural design and optimization method.

Overall, the MRPAM-AP structure and Kriging-GA optimization method can effectively address the mass law problem and achieve significant noise reduction of E-motor. The proposed structure and optimization method can effectively enhance the application and performance of automotive noise control solutions.

Nevertheless, owing to current level of research, the following research needs to be further promoted.

  • Due to limitations in the available testing conditions, the sound-insulation characteristics of the MRPAM-AP were verified through E-motor NVH experiments. This approach does not allow the frequency-STL relationship to be quantified directly. More accurate STL testing methods, such as standing wave tube testing, should be implemented in further research.

  • Sound insulation dynamic model of MRPAM was established in Section 2.2. The formulas explain the dynamic characteristics of the multi-resonators structure and the low-frequency sound insulation mechanism. However, at the current stage of the research, it is still difficult to establish a performance prediction model based on the structural parameters. More accurate parameter performance prediction models for the plate-resonators need to be considered in future work.

Statements

Data availability statement

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.

Author contributions

JW: Conceptualization, Data curation, Formal Analysis, Investigation, Methodology, Resources, Software, Supervision, Visualization, Writing – original draft, Writing – review and editing. BS: Conceptualization, Funding acquisition, Methodology, Resources, Supervision, Validation, Visualization, Writing – review and editing. HG: Funding acquisition, Investigation, Project administration, Resources, Supervision, Validation, Visualization, Writing – review and editing. PS: Conceptualization, Data curation, Formal Analysis, Investigation, Methodology, Software, Supervision, Visualization, Writing – review and editing, Writing – original draft. SH: Data curation, Investigation, Methodology, Visualization, Writing – review and editing. XX: Funding acquisition, Project administration, Resources, Writing – review and editing. NL: Data curation, Methodology, Software, Writing – review and editing. MY: Data curation, Formal Analysis, Methodology, Software, Writing – original draft, Writing – review and editing.

Funding

The author(s) declared that financial support was not received for this work and/or its publication.

Conflict of interest

Author Xiaolong Xie was employed by Shenda(shanghai) Technology 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.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. Part of the content in Figure 1 was generated by generative AI.

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

  • 1

    AghaeiP.HakanenJ.MiettinenK. (2024). A surrogate-assisted a priori multiobjective evolutionary algorithm for constrained multiobjective optimization problems. J. Glob. Optim.90, 459–485. 10.1007/s10898-024-01387-z

  • 2

    AngL. Y. L.KohY. K.LeeH. P. (2017). Broadband sound transmission loss of a large-scale membrane-type acoustic metamaterial for low-frequency noise control. Appl. Phys. Lett.111 (4), 041903. 10.1063/1.4995405

  • 3

    AngL. Y. L.KohY. K.LeeH. P. (2019). Plate-type acoustic metamaterials: evaluation of a large-scale design adopting modularity for customizable acoustical performance. Appl. Acoust.149, 156–170. 10.1016/j.apacoust.2019.01.027

  • 4

    CaiT.HuangS.GuoH.SunP.YuanT. (2023). Low frequency sound insulation performance of membrane-type acoustic metamaterial with eccentric mass block. Phys. Scr.98 (7), 075016. 10.1088/1402-4896/ACDE19

  • 5

    CaiT.HuangS.GuoH.YuanT.SunP.LiuN. N. (2024). Sound insulation performance of membrane-type acoustic metamaterial based on defect state structure. Phys. Scr.99 (2), 18. 10.1088/1402-4896/ad1da3

  • 6

    CaoL. L.WuJ. H.ZhangZ.ZhangC. Z.WanW. X.BaoJ. D.et al (2024). Design optimization of elastic metamaterials with multilayered honeycomb structure by Kriging surrogate model and genetic algorithm. Struct. Multidiscip. Optim.67 (5), 82. 10.1007/S00158-024-03768-0

  • 7

    DalelaS.BalajiP. S.JenaD. P. (2022). A review on application of mechanical metamaterials for vibration control. Mech. Adv. Mater. Struct.29 (22), 3237–3262. 10.1080/15376494.2021.1892244

  • 8

    DingH.GuoH.SunP.HuangS.YuanT.WangY. (2024). In-plane dynamic crushing of a novel hybrid auxetic honeycomb with enhanced energy absorption. Mech. Adv. Mater. Struct.31 (19), 4635–4653. 10.1080/15376494.2023.2204082

  • 9

    FahyF.GardonioP. (2007). Sound and Structural Vibration. Oxford Academic Press.

  • 10

    FatehiE.SarvestaniH. Y.AshrafiB.AkbarzadehA. H. (2021). Accelerated design of architectured ceramics with tunable thermal resistance via a hybrid machine learning and finite element approach. Mater. Des.210, 110056. 10.1016/J.MATDES.2021.110056

  • 11

    FuY.WangX. (2025). Advancements and trends in vehicle sound package for noise control: a comprehensive review. Adv. Mech. Eng.17 (6), 1–17. 10.1177/16878132251345867

  • 12

    HeZ. G.YuanK.XiongG. H.WangJ. (2023). Inverse design and experimental verification of metamaterials for thermal illusion using genetic algorithms. Chin. Phys. Lett.40 (10), 104402. 10.1088/0256-307X/40/10/104402

  • 13

    JiangR. j.ShiG.HuangC. M.ZhengW. G.LiS. (2023). Acoustic insulation characteristics and optimal design of membrane-type metamaterials loaded with asymmetric mass blocks. Mater16 (3), 1308. 10.3390/MA16031308

  • 14

    KaidouchiH.KebdaniS.SlimaneS. A. (2023). Vibro-acoustic analysis of the sound transmission through aerospace composite structures. Mech. Adv. Mater. Struct.30 (19), 3912–3922. 10.1080/15376494.2022.2085348

  • 15

    KhatouriH.BenamaraT.BreitkopfP.DemangeJ. (2022). Metamodeling techniques for CPU-intensive simulation-based design optimization: a survey. Adv. Model. Simul. Eng. Sci.9 (1), 1–31. 10.1186/s40323-022-00214-y

  • 16

    LiuY. (2015). Sound transmission through triple-panel structures lined with poroelastic materials. J. Sound. Vib.339, 376–395. 10.1016/j.jsv.2014.11.014

  • 17

    LiuY.XuW.ChenM.YangT.WangY.HuangX.et al (2020). Three-dimensional fractal structure with double negative and density-near-zero properties on a subwavelength scale. Mater. Des.188, 108470. 10.1016/j.matdes.2020.108470

  • 18

    PengW.ZhangJ.ShiM.LiJ.GuoS. Y. (2023). Low-frequency sound insulation optimisation design of membrane-type acoustic metamaterials based on kriging surrogate model. Mater. Des.225, 111491. 10.1016/j.matdes.2022.111491

  • 19

    RocaD.CanteJ.Lloberas-VallsO.PĆ miesT.OliverJ. (2021). Multiresonant layered acoustic metamaterial (MLAM) solution for broadband low-frequency noise attenuation through double-peak sound transmission loss response. Extreme Mech. Lett.47, 101368. 10.1016/j.eml.2021.101368

  • 20

    TianH. L.WangM.CaiJ. J.SunH. (2020). Influence of layup structures on sound insulation of composite laminates. Compos. Mater. Sci. Eng. (04), 90–95.

  • 21

    TomitaS.AritaS.NakanoS.AizawaK.ItoK.TakeguchiS.et al (2026). Molded metamaterial plate for reduction of sound transmission through automotive door. Sci. Rep.16 (1), 20165. 10.1038/S41598-026-47579-3

  • 22

    VaranasiS.BoltonJ. S.SiegmundT. (2017). Experiments on the low frequency barrier characteristics of cellular metamaterial panels in a diffuse sound field. J. Acoust. Soc. Am.141 (1), 602–610. 10.1121/1.4974257

  • 23

    WangL.LiuH. T. (2021). Parameter optimization of bidirectional re-entrant auxetic honeycomb metamaterial based on genetic algorithm. Compos. Struct.267 (1), 113915. 10.1016/j.compstruct.2021.113915

  • 24

    WangS.XiaoY.GuoJ.ZhangH.WenJ. (2021). Broadband diffuse field sound insulation of double layer metamaterial plates lined with porous material. Appl. Phys. Lett.119 (8), 119. 10.1063/5.0060100

  • 25

    WangS.XiaoY.GuJ. H. W.HuC.ZhangH.WenJ. (2023). Double-panel metastructure lined with porous material for broadband low-frequency sound insulation. Appl. Acoust.207, 109332. 10.1016/j.apacoust.2023.109332

  • 26

    WangC.CaiL.GaoM. C.JinL.SunL. C.TangX. Y.et al (2024). Manufacturing of membrane acoustical metamaterials for low frequency noise reduction and control: a review. Mech. Adv. Mater. Struct.31 (25), 7008–7023. 10.1080/15376494.2023.2242363

  • 27

    WangD.WangX.WangB.NieJ.LiJ.WangX. (2025). Dynamic eccentricity impact on electromagnetic vibration and acoustic noise of interior permanent magnet synchronous motor with different numbers of parallel branches. IEEE Trans. Transp. Electrification.11 (1), 3337–3348. 10.1109/TTE.2024.3438262

  • 28

    WenZ. H.WangD. W.MaL. (2019). Sound transmission of composite sandwich panel with face-centered cubic core. Mech. Adv. Mater. Struct.28 (16), 1663–1676. 10.1080/15376494.2019.1700433

  • 29

    XiaZ. W.WuX. J.YangQ.YangZ. Q.WangZ. Y.WangY. F. (2023). Analysis and test of sound insulation performance of multilayer structures. Noise Vib. Control43, 129–134+141.

  • 30

    XiaoY.WenJ.WenX. (2012). Sound transmission loss of metamaterial-based thin plates with multiple subwavelength arrays of attached resonators. J. Sound. Vib.331 (25), 5408–5423. 10.1016/j.jsv.2012.07.016

  • 31

    XieX. X.YangJ. B.KangY. Z.ShangguanW. B.ZhangQ. (2023). Sound transmission loss of absorber-barrier-absorber structures with kapok fibers as absorption layers. J. Ind. Text.53 (000), 22. 10.1177/15280837231157270

  • 32

    XieH.ZhongQ.ZengW. (2025). An adaptive multi-scale kriging framework with committee-based infill for engineering design optimization. Sci. Rep.15 (1), 42226. 10.1038/s41598-025-26363-9

  • 33

    XuZ.ShiD. S.GuD. W.LiuZ. Q.ChenG. N.HuC. C.et al (2026). Innovative approach to mitigating vibration and noise in an external rotor permanent magnet synchronous motor. J. Vib. Control32 (3-4), 646–663. 10.1177/10775463241302376

  • 34

    ZaryabS. A.MannoA.MartelliE. (2025). SCR, an efficient global optimization algorithm for constrained black-box problems. Optim. Eng.26 (3), 1863–1893. 10.1007/s11081-024-09943-y

  • 35

    ZhangJ. M.WangG. W.WangT.LiF. (2021). Genetic algorithms to automate the design of metasurfaces for absorption bandwidth broadening. ACS Appl. Mater. Interfaces13, 7792–7800. 10.1021/acsami.0c21984

  • 36

    ZhouX.WangL.QinL.PengF. (2020). Improving sound insulation in low frequencies by multiple band-gaps in plate-type acoustic metamaterials. J. Phys. Chem. Solids.146, 109606. 10.1016/j.jpcs.2020.109606

Summary

Keywords

acoustic package, Kriging-genetic algorithm, multi-frequency resonant coupling, sound transmission loss, topology parameter optimization

Citation

Wang J, Sun B, Guo H, Sun P, Huang S, Xie X, Liu N and Yao M (2026) Optimization of multi-frequency resonant coupling acoustic package for electric motor based on a Kriging-genetic algorithm. Front. Mater. 13:1980178. doi: 10.3389/fmats.2026.1980178

Received

27 August 2026

Revised

22 September 2026

Accepted

23 September 2026

Published

19 October 2026

Volume

13 - 2026

Edited by

Shaoping Xiao, The University of Iowa, United States

Reviewed by

Yifeng Fu, Jiangsu University, China

Guofeng Zheng, Chongqing Jiaotong University, China

Updates

Copyright

*Correspondence: Baozhong Sun, ; Hui Guo,

Disclaimer

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.

Outline

Figures

Cite article

Copy to clipboard


Export citation file


Share article

Article metrics