Design Of Fault Diagnosis System For Train Air Conditioning

Train air conditioning is related to the comfort of passengers.Due to the large climate difference between north and south in China,train air conditioning needs to be severely tested.Therefore,ensuring the stable operation of train air conditioners and timely detection of faults is essential for train operation.In order to better understand the operating characteristics of train air conditioning,the key components of the train refrigeration system,such as compressors,condensers,accumulators,throttling devices and evaporators,are modeled and simulated,and the compressor speed is reduced,the condenser is fouled,and the evaporator is evaporated.The trend of each characteristic parameter when the three kinds of faults occur occurs,and the sample data is provided for the data processing part of the following.In order to realize the fault diagnosis function of train air conditioner,BP and SVM algorithm data classification function is used to build BP neural network and SVM algorithm training model on MATLAB platform,and its data classification function is used to classify and identify fault data,and fault recognition accuracy rate They reached 96.67%and 93.33%respectively.In order to optimize the training speed and fault identification accuracy of BP algorithm,the PCA and LLE algorithms are used to reduce the dimensionality of the fault data.After the dimension reduction,the BP algorithm is used for classification and recognition.The final result shows that the BP neural network fault recognition accuracy rate is 95%after the LLE algorithm is reduced,and the training speed is not optimized.The BP nerve processed by the PCA algorithm is not optimized.The network recognition rate has reached 100%and the training time has been reduced by 6 times.The introduction of neural network and SVM algorithm to identify train air conditioning faults will effectively increase the intelligence level of fault identification.Only need to input relevant parameters at a certain moment to diagnose its specific state,and locate the fault occurrence location,which will reduce the demolition.Check the frequency of the machine to improve the service life of the air conditioner.The combination of PCA algorithm and BP algorithm is introduced to effectively improve the training speed and fault recognition rate.