论文部分内容阅读
针对BP神经网络易陷入局部极小值的缺点,提出了混沌神经网络的学习算法。利用混沌的遍历性和随机性,采用混沌变量全局粗搜索与混沌变量局部细搜索相结合,得到神经网络权值的全局最优值。利用该算法对直接转矩控制(DTC)系统进行转速辨识,仿真结果表明用混沌优化BP神经网络的速度辨识器不仅具有较好的跟踪能力,还提高了运算效率,使系统具有良好的静动态性能。
Aiming at the disadvantage that BP neural network is easy to fall into local minima, a learning algorithm of chaotic neural network is proposed. By using the ergodicity and randomness of chaos, the global optimum of neural network weights is obtained by combining the global coarse search of chaos variables with the local fine search of chaotic variables. The algorithm is used to identify the rotational speed of direct torque control (DTC) system. The simulation results show that the speed recognizer based on BP neural network with chaos optimization not only has better tracking ability, but also improves computational efficiency and makes the system have good static and dynamic performance.