Research On Fault Diagnosis Method Of Hydraulic Driving System Of Shield Cutterhead Based On Data-driven

The shield machine is the main engineering equipment for subway and tunnel construction.Due to the complicated construction conditions on site and the improper use and maintenance of various components,the fault of the hydraulic drive system of the cutterhead is frequently occurred which is the core component of shield machine.Therefore,in order to ensure the safe and reliable operation of the process of the shield construction,it is of great significance to timely and effectively diagnose faults.In this paper,the hydraulic driving system of the cutterhead of the earth pressure balance shield of Tianjin Rail Transit is researched.The co-simulation Simulink-AMESim model of the constant power control of the hydraulic drive system of the shield cutterhead is constructed according to the mechanism of the failure.The diagnosis algorithm model based on variableweighted-principal-component-analysis(VW-PCA)method and differential-evolution-BPneural-network method are deduced to improve the accuracy of fault diagnosis and the processing efficiency of fault data.Major works accomplished in this study are listed as follows:1.The structure of the hydraulic driving system of the cutterhead are analyzed including working principle and occurrence mechanism of common fault.The mathematical model of the key components of the system is established to find out the relationship between the change of the model parameters and the fault characteristics.For the article,the theoretical basis of the simulating fault is laid by modifying parameters of the AMESim simulation model of the hydraulic driving system of shield cutterhead.2.The simulation model of the hydraulic driving system of the cutterhead was built in the AMESim software environment,at the same time,and the constant power control model of the hydraulic system was built in Simulink,so Simulink-AMESim model was constructed,which the both of two simulation model were joint.By modifying the parameters of the key components,simulating various faults,and selecting eight fault parameters such as oil pressure of output of hydraulic pump,cutterhead speed,viscous friction torque and soft soil load torque,the fault data is extracted,and provides historical fault data for the further researching of fault diagnosis method.3.A variable-weighted-principal-component-analysis(VW-PCA)fault diagnosis algorithm is established.The weighted vector of each type of fault is obtained by using the biased F-value variable-weighting(VW)algorithm,and different weights are assigned to each parameter of the system,thereby highlighting the pairs of parameters and improving the accuracy of the principal component analysis(PCA)method.4.For the VW-PCA fault diagnosis method can only alarm the fault of the hydraulic driving system,and can not identify which the type of fault is.Based on this,the paper proposes a differential-evolution-BP-neural-network method,in order to use the BP neural network to complete the nonlinear mapping the fault input data of the system onto the output of fault category,and uses the differential evolution algorithm(DE)to improve the convergence speed of the BP neural network and improve the efficiency of fault diagnosis.