一种模糊神经网络改进算法及其在振动强度识别中的应用

对传统模糊自适应Hamming网络算法进行了改进，通过引入新的模糊算法对传统算法中的类别选择函数进行改进，以提高网络的正确识别率，为了实现模式识别中网络的有序输出，对输出层获胜神经元的选取方法也进行了相应的改进。改进后的算法用于空调压缩机壳体振动强度的识别，利用模糊自适应Hamming神经网络综合考虑各测点振动、噪声信号所包含的信息，对壳体振动强度区域实现自动划分。通过改进师前、后两种算法在不同警戒参数下的试验结果发现，采用改进后的算法大大提高了网络的正确识别率，并能够很好地实现网络的有序输出。

An Improved Algorithm of Fuzzy Neural Networks and Its Application to Vibration Intensity Identification

We put forward an improved algorithm of fuzzy adaptive Hamming neural networks, which overcomes the shortcomings of traditional one's shortcomings that are random order recognition and lower discrimination capability by introducing a new category choice function and modifying the method of choosing winner neuron in output layer with the characteristics of vibration and noise signal taken into account, fuzzy adaptive Hamming networks with improved algorthm is employed for identifying shell's vibration intensity, and contrast tests are carried out under differnet vigilance parameters. The results show that the method can improve the networks discrimination power greatly, and can also realize the sequential recognition process.