Aiming at the problem of internal leakage fault easily occurred in the travelling hydraulic system of YJ900 transported frame integrated grider machine,AMESim was used to establish the hydraulic simulation model.Pump leakage,motor leakage and simultaneous leakage of the pump and motor were introduced into the model,the hydraulic motor inlet and outlet data were collected as samples.The data samples were divided into training samples and test samples.The training samples were input into the MATLAB to build a BP neural network fault diagnosis model and the test samples were employed to test the fault model.The influence of the number of neurons and the number of training samples on the success rate of fault diagnosis was studied.Particle swarm optimization (PSO) algorithm was used to optimize the initial weight and bias of BP neural network,by which the success rate of fault diagnosis under the condition of less training samples was significantly improved