Research on Train Air Conditioning Fault Diagnosis Based on PCA-BP

Train air conditioning is related to the comfort of passengers traveling, due to the climate difference between north and south of China, train air conditioning needs to accept the severe test, so to ensure the stable operation of the train air conditioning and timely detection of faults on the train operation is crucial. Through the simulation of the refrigeration system, the normal state, compressor valve leakage, condenser fouling and evaporator fouling four states of the refrigeration system under each characteristic parameter; will be obtained as a sample of the characteristic parameters as sample data, the use of BP algorithms to train the sample data, and ultimately to achieve the input of a certain state of the characteristics of the parameter to locate the corresponding fault parts; due to the BP algorithm convergence is slower, so the use of PCA downgrading technology for the processing of sample data, eliminating the faults. Due to the slow convergence of the BP algorithm, the PCA dimensionality reduction technique is used to process the sample data, eliminate the redundant information in the sample set, and simplify the sample data to improve the training speed of the BP algorithm as well as the accuracy of fault identification.