异步电动机调速系统自适应辨识的CMAC-ADRC算法

CMAC-ADRC Algorithm Based on Adaptive Parameter Identification for Asynchronous Motor Speed Control System

针对异步电动机调速系统快速响应时启动超调量大的问题,提出了一种基于自适应参数辨识的小脑模型神经网络复合自抗扰控制(CMAC-ADRC)的控制算法。将CMAC与ADRC各自的优点相结合,利用CMAC神经网络实现前馈控制,通过在线学习来抑制系统的超调量,增强系统的鲁棒性能,提高系统的快速性能,利用ADRC技术实现反馈控制,进一步增强系统的抗干扰能力。利用参考模型自适应参数辨识技术对转动惯量进行辨识,优化自抗扰补偿系数。以变频器结合异步电动机为控制对象,进行仿真,基于自适应参数辨识的CMAC-ADRC控制算法的干扰响应幅度是一阶优化自抗扰控制下干扰响应幅度的44.57%,是小脑模型神经网络复合比例-微分(CMACPD)控制下干扰响应幅度的17.69%,干扰恢复时间是一阶优化自抗扰控制下干扰恢复时间的50%,是CMAC-PD控制下恢复时间的60%。搭建MCU-CPLD-DSP控制平台进行了实验,基于自适应参数辨识的CMAC-ADRC控制算法的超调量是一阶优化自抗扰控制的45.49%,上升时间是一阶优化自抗扰控制的53.33%,干扰响应幅度是一阶优化自抗扰控制干扰响应幅度的71%,干扰恢复时间是一阶优化自抗扰控制干扰恢复时间的76.47%。

According to the problem of asynchronous motor speed control system, this paper proposes a eerehellar model aritculation controller coupled with active disturbance rejection controller （CMAC - ADRC） control algorithms based on adaptive parameter identification. The respective advantages of CMAC and ADRC were combined. And CMAC neural network was used for feedforward control. Its online learning was applied which suppressed overshoot system, enhanced the robustness and dynamic performance of the system. ADRC was used for feedback control which further enhanced the anti-jamming capability. The inertia was identified by using model reference adaptive parameter identification technique and ADRC compensation factors were optimized. Taking converter and asynchronous motor as control objects, the simulation was carried out. The simulation results showed that the response amplitude caused by disturbance of control system using CMAC - ADRC based on adaptive identification was 44.57% ofthe one using first-order optimization ADRC, and 17.69% of the one using CMAC - PD. Meanwhile, the recovery time of disturbance was 50% of the one using first-order optimization ADRC, and 60% of the one using CMAC - PD. Some experiments were finished on the experiment platform based on MCU - CPLD- DSP. The experiment result showed that with CMAC - ADRC, the overshoot, rising time, response amplitude caused by disturbance, and recovery time of disturbance were 45.49% , 53.33% , 71% and 76.47% of the one using first-order optimization ADRC, respectively.