Research On Fault Diagnosis Of Gun Servo System Based On Wavelet Packet And Optimized Support Vector Machine

Servo system is an important part of artillery weapon equipment.Its basic purpose is to achieve high precision automatic aiming,and to improve the system response ability,automation,intelligence and autonomy of artillery in combat.Its running state directly affects the overall performance of artillery.Therefore,the fault diagnosis of artillery servo system is of great practical significance to improve the maintenance and security of weapon equipment.Based on the analysis of the common fault modes,fault mechanisms and effects of gun servo system,this thesis makes an in-depth study on its fault diagnosis from two aspects: fault feature extraction and fault pattern recognition.The composition of artillery servo system and two common fault modes,the fault of inverter in servo driver and the fault of rolling bearing in reducer,are briefly introduced,and the fault characteristics of the two faults are emphatically analyzed.The inverter fault modes in servo driver are modeled and simulated by MATLAB/Simulink.Then,wavelet packet is selected to extract signal features,and the fault signal is decomposed into three layers by wavelet packet decomposition method,and the energy value of wavelet packet band obtained after wavelet packet reconstruction is used as its feature vector,which provides good training samples and test samples for subsequent pattern recognition.In view of the fact that the parameter selection of Support Vector Machines(SVM)depends on experience selection because of the lack of a large number of theoretical and data sources,we can use the Grey Wolf Optimizer(GWO)to optimize the relevant important parameters of SVM.Considering the limitation of experimental conditions,in order to obtain the fault information of servo system,fault simulation becomes an essential link.Therefore,this thesis combines the data of case western reserve university experimental platform and the simulation data of servo driver,and uses wavelet packet to input the extracted feature vectors into support vector machine and support vector machine optimized by grey wolf optimization algorithm,and carries out fault diagnosis simulation verification on classification accuracy and parameter optimization time.By comparing the simulation results,it can be proved that the diagnosis method based on wavelet packet and GWO-SVM proposed in this thesis can not only effectively extract fault features and accurately identify faults,but also has significant advantages in fault classification accuracy.The system is developed with the help of Graphical User Interfaces(GUI)tools based on MATLAB software,and the overall framework of the system is designed,which realizes the functions of fault feature extraction and fault pattern recognition of the artillery servo system.The feasibility test of the designed fault diagnosis GUI system shows that the system is practical,simplifies the fault diagnosis process,and improves convenience and man-machine interaction.