Position Servo Control of Limited angle Torque Motors using BP Neural Network-based Nonlinear Output Compensation

Limited angle torque motors (LATMs) exhibit significant performance advantages in specific fields, but the complex magnetic-electric field coupling involved makes it difficult to accurately model the actuator. To solve this problem, this paper proposes a BP neural network-based method to compensate the nonlinear error output of LATM. By analyzing the nonlinear mechanism of LATM output characteristics, a learning method and system based on model error is designed. Instead of constructing a complex physical model, the method learns the deviation samples in the output characteristics through the BP neural network algorithm and uses feed-forward control to compensate the nonlinear output. Experimental results show that the method can effectively reduce the torque fluctuation of the LATM in the corner range, thus achieving accurate and stable system output characteristics.