Research On Fault Diagnosis Method Of Rotating Machinery Based On Autocorrelation And EMD

Modern mechanical equipment has rotating mechanical parts,which has an important influence in the field of engineering.The running state of rotating machinery will directly affect the performance of mechanical equipment,so the research of rotating machinery fault diagnosis meets the needs of production development.Before the diagnosis research,it is necessary to process the mechanical vibration signal and extract the feature.Traditional signal processing and feature extraction methods include Fourier transform,holographic spectrum,fractal dimension,etc.,but the main function of these methods is to analyze signals,and the amount of information feature extraction is often unable to meet the requirements.The traditional fault identification methods include empirical mode decomposition,approximate entropy of inherent time scale decomposition,Support vector machine and so on.To some extent,these methods realize the fault recognition of rotating machinery,but only improve the method of fault diagnosis and classification,or extract a single feature of vibration signal,so as to improve the recognition rate.To solve these problems,this paper studies signal processing,feature extraction and modal identification.The main contents are as follows:Firstly,in order to eliminate the noise of the signal,based on the autocorrelation and empirical mode decomposition,a new denoising method based on autocorrelation and improved set empirical mode decomposition is proposed.Firstly,the original signal is de-noising by autocorrelation,and the eigenmodes are obtained by selecting the corresponding improved set empirical mode decomposition and empirical mode decomposition,and the correlation coefficient is calculated.Secondly,the eigenmodes with large correlation coefficients are reconstructed.Finally,clear fault signal is obtained.The experimental results of different rotating machinery show that this method can effectively suppress the noise of rotating machinery and extract the vibration signal with obvious impact.Secondly,in order to extract effective feature information,based on the theory of information entropy and wavelet packet decomposition,this paper proposes a feature extraction method based on wavelet packet energy entropy.Firstly,the signal is decomposed into four layers of wavelet packets,and then the energy value of each frequency band is calculated.Finally,the entropy value is calculated by using theinformation entropy,and 16 clear fault characteristic parameters are obtained.Through the experiment comparison of different feature extraction methods,this method can effectively extract feature parameters,so that there are obvious differences between different fault signals.Finally,in order to achieve the purpose of fault identification and improve the accuracy of fault identification.In this paper,based on BP neural network and each optimization algorithm,through the improvement of optimization algorithm,a fault diagnosis method based on chaos particle swarm optimization neural network is proposed.This method uses particle swarm optimization to optimize BP neural network and train to identify faults.Then,on the basis of PSO neural network,using the characteristics of chaos algorithm,avoiding local optimization,improving the accuracy of fault identification.The experimental results of different rotating machines show that this method can effectively improve the accuracy of fault identification,and finally achieve fault identification.