ARX-Based Dynamic Modeling for SVD-LSM Six-Dimensional Force/Torque Sensors

To address the randomness in model order selection during dynamic modeling of six-dimensional force/torque sensors, this paper proposes a high-precision modeling approach based on the ARX model, utilizing the SVD-LSM method. This approach aims to support the design of dynamic compensators for enhanced sensor performance. First, major singular values are filtered by singular value decomposition (SVD) to reduce noise and redundant information in model selection. Then, the model parameters are optimized by the least squares method (LSM) by combining historical input and output data in ARX model construction to achieve accurate model parameter identification. Finally, the effectiveness of the method is verified by experiments, and compared with the traditional particle swarm optimization (PSO) and crayfish optimization algorithm (COA), the SVD-LSM method performs well in terms of order prediction and root mean square error (RMSE), providing higher model accuracy.