The complex environment of grinding makes it difficult for traditional method of signal analysis to extract feature signal

therefore

a feature extraction method based on wavelet decomposition and energy spectrum is proposed. Using wavelet multi-scale and multi-resolution characteristics

the acoustic emission signal is decomposed by multi-scale decomposition; according to characteristics of acoustic emission signal of metal grinding

optimal frequency band is selected to carry out spectrum analysis

and combined with energy spectrum characteristic signal is extracted. The experimental analysis on self-developed grinding monitoring system shows that this method can effectively and accurately diagnose various state of the grinding

and the false positive rate is0. 02 %. Compared with the single spectrum analysis

the precision is higher and the reliability is better

and it has certain engineering practical value.