Research On Fault Diagnosis Of Photoelectric Detection System Based On Support Vector Machine And Neural Network

The photoelectric detection system is a system that detects optical signals through photoelectric conversion and circuit processing methods,and is widely used in industrial,medical,astronomical and military fields.With the continuous development and wide application of photoelectric detection systems,its accuracy and stability requirements are getting higher and higher,so fault diagnosis research on photoelectric detection system is necessary.In this paper,the fault diagnosis research of the photoelectric detection system composed of the photomultiplier tube(PMT)module and the current amplifier is carried out through simulation,and the fault diagnosis schemes for the PMT module and the current amplifier are proposed and verified by simulation.The main research work and innovation of the paper are as follows:First,based on the characteristics of the output waveform of the PMT module during the pulse amplitude test,its fault diagnosis research is carried out.The pulse amplitude test of the PMT module often uses the scintillator under the action of the nuclear radiation source as the pulse light source,and the nuclear pulse signal collected at its output can be approximately represented by a double exponential function.In this paper,the nuclear pulse generation circuit is designed and the Gaussian shaping of the signal is carried out using CR-(RC)~4 circuit on the Multisim14.0 platform.The nuclear pulse waveform data obtained through simulation can simulate the output pulse waveform of each fault state during the pulse amplitude test of the PMT module,and provide an excitation signal for the current amplifier circuit to conduct fault diagnosis research on its circuit soft fault.Secondly,the simulation output waveforms of the PMT module and the current amplifier circuit under different fault conditions are obtained through simulation and a fault model is established.The wavelet packet decomposition is used to extract the features of the fault datas,and then the support vector machine(SVM)and BP neural network are used to classify the fault datas.Aiming at the selection of penalty parameters and kernel function parameters of SVM,particle swarm optimization(PSO)and improved particle swarm optimization(IPSO)are used to optimize SVM.Finally,SVM model,PSO-SVM model,IPSO-SVM model and BP neural network are used to classify the fault data of photoelectric detection system.The simulation results show that the fault diagnosis scheme combining the feature extraction method of wavelet packet decomposition and the fault classification method based on SVM and BP neural network can accurately and quickly complete the fault diagnosis of the photoelectric detection system,which makes a useful attempt to the intelligent diagnosis of the photoelectric detection system.