Research On Mechanical Faults Signal Processing And Diagnosis Method

In the fault diagnosis of mechanical equipment,due to the complicated structure of the mechanical equipment itself and the interference of environmental noise,the information reflecting the operating state of the equipment is often overwhelmed by strong noise.Especially in the early stages of mechanical equipment failure,it is more difficult to extract weak fault characteristics.In addition,the signals required for fault diagnosis are primarily provided by sensors placed on the structure,and how to arrange the sensors is critical to the acquisition of fault signals and the diagnosis of the fault.From the perspective of vibration signal acquisition and processing,this thesis studies the optimization of measuring points in the acquisition process,and the diagnosis of early weak faults.The former takes the optimal layout of measuring points of pump body as an example,and the latter takes the bearing fault diagnosis as an example.The main works are as follows:To obtain the best information for fault diagnosis,and achieve a large amount of information with a limited number of sensors while information redundancy can be minimized,Fuzzy C-means clustering method was used to achieve optimal placement of sensors.Firstly,the structure modal analysis was conducted,and mode shapes were extracted.Then,on the basis of the dynamic similarity of the mode shape values in important modes,the degrees of freedom(DOFs)were clustered by using Fuzzy C-means clustering algorithm.The DOFs which had much information were chosen from each cluster as candidate test point.Objective functions were established based on modal assurance criterion(MAC).Genetic algorithm was adopted to solve objective optimization.The sensor’s positions were optimized.Finally,three evaluation criterions,singular value ratio of modal matrix,Fisher information criterion and MAC criterion,are used to form a comprehensive evaluation index to estimate different placement results.Taking a locomotive pump body as an example,the simulation results show that this method can effectively avoid the aggregation of measuring points,and solve the problem of information redundancy while obtaining a large amount of operational state information of reaction equipment.The improved singular value decomposition(SVD)and parameter optimized Variational mode decomposition(VMD)were introduced to diagnose early weak faults.Firstly,the original fault signal was denoised by SVD,and the weak fault was enhanced.The rank of the reconstruction matrix was optimized by the principle of minimum envelope entropy and maximum kurtosis.Secondly,the signal denoised by the improved SVD was decomposed by VMD,and a new index,called ensemble kurtosis,was constructed by combining with kurtosis and the envelope spectrum kurtosis.The parameters of VMD were optimized by the principle of the maximum mean of all intrinsic mode functions(IMFs),and several IMFs were obtained.Finally,according to the kurtosis-Euclidean distance index,the IMF with rich fault information was screened out,and the envelope spectrum of this IMF was obtained.The characteristic frequencies at the prominence of the amplitude were compared with the theoretical values to determine the type of fault.By analyzing the simulated signal and the measured data of bearing fault,the weak characteristic frequency information can be extracted successfully.The results show that the fault diagnosis method based on improved SVD and parameter optimization VMD can effectively realize the early fault diagnosis,and has certain reliability and practicability.