Active disturbance rejection control with adaptive RBF neural network for a permanent magnet spherical motor.

Abstract

In response to the issues of low tracking accuracy and poor robustness in the trajectory tracking control of a permanent magnet spherical motor (PMSpM), an active disturbance rejection control (ADRC) scheme combining neural networks is put forward in this research. The unknown total disturbance is approximated by employing a radial basis function (RBF) neural network, with weights updated by an adaptive law and compensated for through the nonlinear feedback loop. This approach addresses the problem of performance degradation of the extended state observer under severe total disturbance, thereby ensuring accurate tracking of the PMSpM. Comparative simulations are accomplished to evaluate the performance of the RBF-ADRC scheme in enhancing disturbance rejection capability and robustness. Experimental results from the planar circular motion experiment on the PMSpM test platform validate the application value of the scheme.