模糊神经网络在无刷直流电机中的应用

Application of fuzzy neural networks in Brushless DC Motor

针对电动汽车中无刷直流电机负载波动较大的特点,提出了4层模糊神经网络模型,该模型融合了模糊逻辑和神经网络的长处,模糊推理和解模糊化均通过神经网络来实现。模糊化层将输入特征量转化为模糊量,选取的隶属函数使神经网络的权值表示一定的知识,在输出层通过解模糊得到具体的控制量。实验结果表明,该模糊神经控制系统具有较好的带负载能力和抗负载变化能力,达到了预期的效果。

Aiming at the Brushless DC Motor（BLDCM） in electric vehicles whose load fluctuates irregularly, an four-layer fuzzy neural networks is presented. By fusing the advantages of fuzzy logic and neural networks, both the fuzzy inferenee and the defuzzification of this model were realized by neural networks. The input variables were translated into fuzzy variables by fuzzy layer. The selected membership function made neural network weight values have definite knowledge meaning. Finally, the reliable control values were gained by proper defuzzification on the output layer. The results show that it exhibits robustness and good adaptation capability, and it can be practically implemented.