Modeling of Friction Torque Based on Turntable Servo System Using Neural Network with Memory Storage Unit

Cheng Xie,Songlin Chen,Yang Liu

DOI: https://doi.org/10.1007/s12555-021-0973-4

IF: 2.964

2023-01-01

International Journal of Control Automation and Systems

Abstract:Friction is the main factor affecting the performance of the servo systems. The tracking accuracy can be effectively improved by compensating the rotational torque online with the accurate value of friction torque. Therefore, it is necessary to build a friction model, which can accurately describe the friction characteristics of the servo systems during operation. This paper proposes a neural network model based on the mechanical characteristics of friction. The memory storage unit is added to the network structure to store the state information between the current and the next speed reversal time so as to process the dynamic and static friction characteristics separately. To eliminate the quantization noise existing in the angular velocity calculated by the angular position difference, the Savitzky-Golay convolution smoothing algorithm is used to smooth the angular velocity and the genetic algorithm is used to determine the sliding window size and the fitting polynomial order. The simulation results show that the neural network model proposed in this paper can reproduce the characteristics of the dynamic and static friction torque well, and the fitting effect is better than that of the feedforward network and LuGre model under different forms of excitation signals, so it has strong generalization ability and predictive accuracy. In addition, the modeling process can be effectively simplified, thus verifying the effectiveness and superiority of the proposed method.