Load Detection of Industrial Robots in Manufacturing Environment Based on Improved FNO Network

Fangyong Gao¹Dechang Xie²Yu Xie^3*

• ¹Software Engineering Institute of Guangzhou, Guangzhou, China
• ²Guangxi Minzu Normal University, Chongzuo, China
• ³Guangxi Vocational&Technical Institute of Industry, Nanning, China

In response to the insufficient accuracy in load detection of industrial robots, a load detection technology based on Fourier neural network is proposed. The robot dynamics model is analyzed and a load detection model is constructed. Secondly, a Fourier neural operator is introduced to extract spatial physical information. In addition, the study introduces an attention mechanism layer to enhance key load information and correct the influence of external environment through error compensation. In the load detection experiment, the proposed model showed the best prediction accuracy compared to similar models. For example, when the load was 2kg, 2.5kg, and 3kg, the predicted loads were 2.0044kg, 2.5102kg, and 3.0150kg, respectively. In addition, the proposed model performed well, with an average time of 0.82ms. In the fusion error compensation, the maximum relative error of the proposed model after error correction remained within 3.25%, showing the best performance. In the revised time comparison, the single sample time of the proposed model was 5.1ms, which was better than that of similar techniques. The technology has good application effects. The proposed model will provide technical support for parameter recognition and control optimization of industrial robots.