基于强化学习的高精度电机转角控制仿真  

High Precision Angle Control Simulation Based on Reinforcement Learning

作　　者：刘黎[1] 

机构地区：[1]西南交通大学电气工程学院,四川成都610031

出　　处：《计算机仿真》2014年第3期402-405,共4页Computer Simulation

基　　金：中央高校基本科研业务费专项资金资助项目(SWJTU12CX032)

摘　　要：研究电机转角的高精度控制问题。高精度电机的转角对误差精度要求很高,调度控制存在惯性误差,传统的角度控制方法中,由于运算过程中存在控制时差,造成控制过程粗糙,为了避免上述传统算法的弊端,提出一种基于强化学习的高精度电机控制方法。按照强化学习的相关理论,根据高精度电机转角控制状态数据能够强化控制信号,从而完成高精度电机的转角控制,针对高精度电机转角控制状态数据,对控制结果进行模糊化处理和模糊逆变换处理,从而降低控制误差。实验结果表明,利用改进算法进行高精度电机转角控制,可以极大地提高控制的准确性,从而满足生产过程中的实际需求。In this paper, the high precision control of the motor rotation was studied. The error precision require- ment of high precision motor angle is very stringent, In scheduling control of traditional control method, due to time difference existed in the operation process control, inertia error exists, and control process is rough. In order to avoid the disadvantages of the traditional algorithm, in this paper, a high-precision motor control method based on rein- forcement learning was proposed. In accordance with relevant theories of reinforcement learning, high precision of an- gle control state data can strengthen control signal, so as to complete high accuracy angle control of the motor, angle control status data, for high precision motor. The control results were processed based on fuzzy inverse transforma- tion, which can reduce control error. The experimental results show that the algorithm presented in this paper for high precision angle control, can greatly improve the accuracy of control, so as to meet the actual demand in real produc- tion.

关 键 词：电机转角控制 强化学习 模糊逆变换 模糊化 

分 类 号：F127[经济管理—世界经济]