Optimization of Chassis Stability Network Control System of Substation Inspection Robot Based on Multimodal Sensor Fusion

Abstract

Substation inspection robots are prone to attitude instability, path deviation, and low inspection efficiency when operating in complex terrains due to the instability of multi-source sensor data and the inadequacy of chassis control strategies. To address this issue, this paper presents an optimization method for chassis stability network control system based on multimodal sensor fusion. The extended Kalman filter (EKF) is used to filter and fuse multi-source sensor data such as vision and inertia to generate high-precision environmental perception information. The deep learning model CNN-LSTM (Convolutional Neural Network - Long Short-Term Memory) is then utilized to further enhance the multimodal data processing capability. Then, fuzzy logic control combined with sliding mode control is used to optimize chassis posture adjustment, and the spatio-temporal graph convolutional network (ST-GCN) is applied to predict the chassis’s dynamic behavior. Finally, the system’s real-time and robustness are improved based on distributed control and edge computing. The experimental results show that the chassis tilt angle stability of the optimized system is within 1.5° in the simulation test; the data processing latency time is controlled within 13ms; the inspection task completion rate on slippery roads reaches 95%. The study demonstrates that this method effectively improves the stability and inspection efficiency of the inspection robot in complex terrain, and provides a reliable solution for the development of substation inspection technology.