基于免疫神经网络的电力电子电路故障诊断

Fault Diagnosis of Power Electronic Circuit Based on Immune Neural Network

任何单一的故障诊断方法都有其优缺点,根据生物免疫学的原理,将人工神经网络和免疫算法结合起来,形成免疫神经网络。将免疫系统应用于人工神经网络隐层数据处理中,选择一种合适的确定隐含层到输出层的权值,计算得到人工神经网络的输出,并根据输出结果诊断出故障的类型。实验结果表明,免疫神经网络能够以较小的网络规模实现对多种故障模式的准确识别,它具有较快的处理速度、良好的容错性和强大的自适应能力。

Any single fault diagnosis method has its advantages and disadvantages. According to the biological immunology principle, the artificial neural network was combined with the immune algorithm to form the immune neural network. The immune system was applied to choose a hidden layer in artificial neural network data, then a suitable layer weight from the hidden layer to the output layer was selected, and artificial neural network output was calculated, the output result showed the type of diagnosis. The result indicates that the immune neural network can realize many kinds of faults pattern recognition accurately with a smaller network scale, and it has high process-speed, good fault-tolerant performance and formidable self-adapting ability.