神经元控制器在数字调速系统中的实现和应用

神经网络具有自学习、自适应能力 ,用于控制时可不依赖控制对象的数学模型。异步电机矢量控制技术是通过坐标变换 ,实现对定子电流的励磁分量与转矩分量的解耦控制。为实现对交流电机快速和精确控制 ,本文基于单神经元设计出用于异步电机矢量控制的自适应磁链和转速控制器 ,利用神经元的自学习功能在线调节连接权重 ,实现自适应控制。并将此设计应用于由数字信号处理器 (DSP)实现的交流电机矢量控制系统中 ,实验表明此方法设计的控制器结构简单 ,易于数字化实现 ,控制系统动态性能良好。

Realization and Application of Single Neuron Controller in Digital Control System

The neural network with the capability of self-study and self-adaptive control, can control a system without modeling the plant. The technique of induction motor vector control is a way to realize the decoupling control of stator current excitation and torque variable using coordinates transform. The conventional PID controller is of poor robustness when the motor parameters are variable. For the sake of fast and accurate control of induction motor, the adaptive flux controller and speed controller based on neuron for the induction motor vector control using digital signal processor are designed. The adaptive control is realized, using the capability of neuron self-study adjusting the weight online. The result shows that the designed scheme is simply, easy to achieve digitalization and perfect in dynamic performance.