Research On Valve Fault Diagnosis Based On Wavelet Packet Decomposition And BP Neural Network

With the improvement of mechanical equipment complexity and automation level,the importance of mechanical equipment fault diagnosis is becoming more and more significant.In the study of intelligent fault diagnosis technology,wavelet analysis and neural network technology are both hot research content and research frontiers.This article first describes the research content and significance of the fault diagnosis technology.The main methods and steps of fault diagnosis are discussed.As the research object,the typical vibration signals of the mesh valve of the 2D-90 MG reciprocating compressor are collected.The vibration domain and frequency signals of the mesh valve are analyzed when the mesh valve fails.The main research is as follows:1.Design the fault simulation experiment of 2d-90 mg reciprocating compressor mesh valve,and the simulation effect of the fault experiment is very close to the actual production situation of reciprocating compressor,which is of great research value;2.Collection and analysis of vibration signal of netted air valve.On the basis of installing acceleration sensor on netted air valve seat,a large number of acceleration signals of netted air valve of reciprocating compressor were collected,which laid a data foundation for the following fault intelligent analysis.3.Time domain waveform and spectrum of the network valve acceleration signal are analyzed,as well as vibration signal characteristics of the network valve in different working states.4.Based on the wavelet packet energy spectrum model,the fault characteristics of reticulated valve were extracted.The wavelet packet decomposition technology is adopted to solve the problem that it is difficult to extract the fault characteristics of the vibration signal of the mesh valve.The original acceleration signal of the mesh valve is decomposed appropriately.And the element's feature vector is constructed with the energy of the subband signal.5.Intelligent identification of mesh valve faults based on BP neural network(DNN);To achieve intelligent identification and diagnosis of mesh valve fault categories,a mesh valve fault identification method based on BP neural network(DNN)is proposed.For get a diagnostic model,the wavelet packet energy spectrum feature vector obtained above is input into the designed DNN to train.Then the trained DNN model is used for intelligent prediction of new samples.The experimental results show that the diagnosis accuracy of the fault diagnosis technology of the mesh valve proposed in this paper is good.It is verified to be effective in this paper to extract the fault characteristics of the mesh valve based on the vibration signal and wavelet packet decomposition technology.And it is feasible of the intelligent identification of the mesh valve fault based on the BP neural network.