Research On Health Diagnosis Method Of Vehicular Power Supply Based On Deep-learning Theory

The vehicle power supply is the main source of electric energy in the field operations of modern military weaponry.The health status diagnosis is a technical method to determine the running state of the vehicle power supply.It can evaluate the operating state of the equipment and diagnose the faul t in time.It is beneficial to the operation,maintenance and maintenance of the vehicle power supply.The development is better suited to the power supply of the vehicle,improving the safety,reliability and economic requirements of its operation.Under this background,the key technologies of vehicle power health assessment,fault diagnosis and other key technologies are studied.In view of the outstanding advantages of deep learning method in data feature extraction,this paper focuses on the deep belie f network in deep learning method.The research results of this paper have important theoretical and applied value for improving the management level of vehicle power equipment.The main work of this article includes:1)DBN-MD based health state assessment for vehicle power supplyIn view of the problem that the vehicle power manufacturer and its military users need to assess the health status of the vehicle power supply and have less research results,this paper presents a DBN-MD based assessment method for the health status of the vehicle power supply.First of all,the DBN method is used to train the health and abnormal running data of the vehicle power supply off-line to identify the health status of the vehicle power supply.In line with the DBN health status recognition model,the abnormal data in the real-time data are indicated,and the health degree of the vehicle power source is evaluated by Mahalanobis distance,and it is timely and high.The effective evaluation of the health status of the vehicl e power supply shows that the evaluation of the health status of the vehicle power supply by the method of DBN-MD has good effect.2)Research on vehicle power fault diagnosis method based on DBNIn this paper,the fault diagnosis method of vehicle power s upply based on DBN is studied in this paper to solve the problem that the fault mechanism of the vehicle power supply is complex and the knowledge experience is insufficient,and the traditional shallow neural network is difficult to be satisfied.With the aid of several common fault data collected by the vehicle power supply simulation system,the pre training and reverse tuning of DBN are used to build a deep diagnosis neural network for the corresponding fault of the vehicle power supply,and the effecti ve intelligent diagnosis of several common faults of the vehicle power supply is realized.The advantage of this method is that it can integrate the fault feature extraction and fault diagnosis of the vehicle power supply and get rid of the dependence of t he traditional shallow fault diagnosis method on the large number of signal processing technology and the diagnosis experience.The simulation experiment also further shows the effectiveness and suitability of the method in the fault diagnosis of the vehic le power supply.3)Research on vehicle power fault diagnosis method based on PCA-DBNFor network training and fault diagnosis based on multiple measurements,the accuracy and time of modeling and diagnosis are affected by many variables and not independent of each other.In this paper,a fault diagnosis method based on PCA-DBN is proposed.This method is used to establish the vehicle power depth.When the degree network fault diagnosis model is used,the PCA is used to reduce the dimension of the data set to eliminate the more effective information.Then,the DBN is pre trained and fine tuned,and then the fault diagnosis of the vehicle power supply is also used.The simulation results show that the proposed method has better fault diagnosis accuracy and timeliness than that of single DBN and PCA combined with shallow network.