Parallel Self-Tuning Controllers for Speed Servo System Based on Predictive Model-Mismatch Compensation

When the operating conditions are changed suddenly, the model mismatch for the servo system using model-based methods occurs during the identification convergence, which will lead to the obvious speed spikes and damped oscillations, deteriorating the control performance of the servo system. To solve the problem, a parallel self-tuning scheme based on a generalized predictive control law is proposed in this paper. In the proposed scheme, the controlled model parameters in which an integral-proportional controller (IPC) is considered as the controller of speed loop are first online estimated by a recursive least squares method with a forgetting factor. Then, a model-mismatch compensator (MMC) is designed to obtain the corresponding compensation current, and in addition, the predicted speed is considered as the feedback speed to promote the establishment of an inner loop, which allows for a faster and more thorough model mismatch compensation based on both the excitation torque current and the adaptivity of the MMC. At the same time, through constructing two different quadratic performance indicators, the optimal control laws can be obtained based on a simplified decoupled derivation, thus supplying IPC and MMC with suitable control parameters simultaneously. Simulation and experimental results show that, compared with traditional methods, the proposed scheme can ensure faster convergence speed and better control performance.