论文部分内容阅读
根据刀具磨损状态不同时其不同频带的能量不同,将小波包分解方法和基于神经网络的模糊系统融合器相结合,用于车刀状态诊断。采用小波包将信号分解为不同频带的信号,通过求取不同频带的均方根值提取各特征量,然后将特征向量分别输入BP、SVM、ELM、PNN 4种神经网络分类器,将不同分类器的分类结果应用模糊网络进行优化综合。实验结果表明:多分类融合分类识别效果比单个分类器效果要好,提高了对刀具状态的识别精度。
According to the different energy of different frequency bands when the tool wear state is different, the wavelet packet decomposition method is combined with the fuzzy system fuser based on neural network to be used in turning tool state diagnosis. The wavelet packet is used to decompose the signal into signals in different frequency bands. The feature values are extracted by taking root mean square (RMS) values of different frequency bands. The feature vectors are then input into four neural network classifiers, BP, SVM, ELM and PNN respectively. Classification results of the application of fuzzy network optimization synthesis. The experimental results show that the recognition effect of multi-classification fusion classification is better than that of single classifier, and the recognition accuracy of the tool state is improved.