Research On The Method Of Constrained Independent Component Analysis Diagnosis Based On Alpha Stable Distribution

In recent years,with the continuous progress of science and technology and the continuous development of industrial demand,the requirements about safety and efficiency of the production equipment attach more and more attention.Failures of critical equipment will not only affect the entire production process,but also may pose a threat to the safety of the production workers.Therefore,the condition monitoring and fault diagnosis of key equipment is very necessary.Fault diagnosis of rotating machinery is usually based on the vibration signal,which through a variety of modern signal processing methods to explore the fault information.The vibration signal collected by the sensor is a mixture of several vibration source signals.How to extract useful information from these complex signals is an urgent problem to be solved.In this paper,the gear box is the research object,which is the representative of the rotating machinery.The vibration mechanism of the main parts of the gear box is studied.According to the linear instantaneous mixture model,and using the theoretical framework of ICA(Independent Component Analysis),a new fault diagnosis method is established.This paper uses alpha stable distribution model to fit the fault signal,and stable distribution parameters to describe the feasibility of non-Gaussian signal.Establishing independence measurement based on fractional lower order statistics,choosing genetic algorithm as the optimization algorithm,designing n-dimensional rotation matrix,so as to establish new ICA method.And on this basis,two kinds of prior information are integrated.The hybrid matrix and signal time domain characteristics are considered respectively in this paper,and the spatial constraint ICA method and hybrid constraint ICA method based on the ? independence measure are established.Finally,the validity of the method is verified by the simulation signal.The performance and efficiency of the ICA method,the spatial constraint ICA method and the hybrid constraint ICA method based on the ? independence measure are tested by the experiment data which was collected from the gearbox with different fault types.Finally,the compartments about different methods are made through the visual spectrum and objective evaluation parameters.