| As the quantity of electric vehicles continues to grow,the security of electric vehicle charging equipment is becoming more and more prominent.If the core components of electric vehicle charging equipment fail,it will definitely affect the stability and reliability of the whole charging equipment system.Hence,it is important to carry out an in-depth research on the fault diagnosis of electric vehicle charging equipment.Because of the complexity of the physical structure of charging equipment and the fusion of data collected by multiple types of sensors in the process of monitoring different operating states,the fault data has the characteristics of high dimensionality,nonlinearity and high coupling.Establishing accurate physical models.Deep learningbased methods can adaptively extract rich and effective features from complex data and accurately map the complex relationships between fault data and faults,and the model building process is relatively simple.Thus,we apply Deep Belief Network（DBN）to the fault diagnosis of electric vehicle charging equipment,and make a series of improvements and optimizations to it in this paper.Finally,a fault diagnosis system for charging equipment is developed based on the proposed method,and its practicality is verified by putting it into use.The primary focus of this paper is stated as follows.Firstly,the research status of fault diagnosis methods for electric vehicle charging equipment,the current prevailing fault diagnosis techniques,the research significance of fault diagnosis for electric vehicle charging equipment and the primary research content of this paper are introduced.Secondly,the structure and operating mechanism of electric vehicle charging equipment are explained,then the common fault types of charging equipment are derived based on the historical operation status number of charging equipment,and the causes of the common faults of charging equipment are further analyzed,and the main characteristic parameters of charging equipment faults are clarified by the method of random forest,and then the data set is constructed.Then,from the construction and theory of DBN,the two training processes of pretraining and fine-tuning of the network are described,and a DBN-based fault diagnosis method for electric vehicle charging equipment is designed,which is compared with various traditional fault diagnosis methods,like Back Propagation Neural Network（BPNN）,Support Vector Machine（SVM）,etc.,thus verifying that the proposed method has higher fault diagnosis accuracy for electric vehicle charging equipment.Next,the Sparrow Search Algorithm（SSA）is introduced to determine the structural parameters of the network more objectively and accurately,and an improved Sparrow Search Algorithm（ISSA）is designed by integrating the Tent chaos mapping and an adaptive operator with the Sparrow Search Algorithm.Algorithm（ISSA）;meanwhile,to solve the problem that typical DBNs tend to fall into local optimal solutions,an Optimized Deep Belief Network（ODBN）is designed,which uses a Linear Restricted Boltzmann Machine（LRBM）to initialize the output layer parameters of the network.Combining the above algorithms,an ISSA-ODBN-based charging device fault diagnosis method is finally proposed and its effectiveness is experimentally verified.Finally,based on Visual Studio,using C# high-level programming language,a charging equipment fault diagnosis system was designed and developed,and the algorithm proposed in this paper was applied to this system,and then the developed system was tested to verify the practicality of the system for electric vehicle charging equipment fault diagnosis. |
