Research On Online Fault Detection Of Photovoltaic Array Based On Optimized Support Vector Machine

As large-scale photovoltaic installations are put into use,photovoltaic arrays are the core components of photovoltaic systems,and it is particularly important to monitor electrical parameters and diagnose faults in real time.The existing traditional methods and intelligent algorithms have limitations on the fault detection of photovoltaic arrays.In this paper,by establishing a photovoltaic simulation model and analyzing the fault output characteristics,a method is proposed for the four faults of open circuit,short circuit,shading and abnormal aging that often occur in photovoltaic arrays.A Least Squares Support Vector Machine(LSSVM)photovoltaic array fault diagnosis method based on fault parameter input.The main contents of the paper are as follows:(1)For the four faults of open circuit,short circuit,shading and abnormal aging that often occur in photovoltaic arrays,analyze the causes of the faults and establish a Simulink simulation model to obtain the fault state criteria and output change rules of photovoltaic cells and photovoltaic modules,which are the basis for photovoltaic arrays.Fault diagnosis provides a theoretical basis.(2)Aiming at the few fault data and feature types of photovoltaic arrays,a smallsample fault diagnosis system with Support Vector Machine(SVM)as the core is proposed.By analyzing the basic theory and classification mechanism of SVM,the problems of lacking pertinence in parameter selection of SVM and the large influence of sample size on the solution speed of SVM are improved by using LSSVM.LSSVM improves the performance of the algorithm by simplifying the quadratic programming problem.The comparison of the classification test results between the two algorithms and the kernel function shows that LSSVM has higher classification accuracy,and the parameters of the kernel function affect the classification performance of the algorithm.(3)Aiming at the defects of Artificial Bee Colony(ABC)algorithm,which is easy to fall into the local optimal solution and slow down the convergence speed in the later stage,an adaptive weight factor is introduced,combined with the chaotic optimization strategy,and an improved artificial bee colony(Improved Artificial Bee Colony)is obtained.Colony,IABC)algorithm optimizes the kernel function parameters of LSSVM.The simulation results of the basic ABC and IABC algorithms by the test function show that IABC obtains good global optimization results on the basis of effectively solving the slow convergence rate of the basic ABC,which proves the effectiveness of the algorithm optimization.(4)Combine the formal features of the LSSVM kernel function with the design constraints between the parameters,and then use the IABC algorithm to optimize the optimal parameter range to establish a photovoltaic array fault diagnosis model based on the IABC-LSSVM algorithm.The fault data obtained by the photovoltaic array simulation model Carry out training and diagnosis,and compare the diagnosis results with the diagnosis results of the ABC-SVM algorithm and the PSO-RBF algorithm.The test results show that the improved algorithm can obtain better diagnostic results for photovoltaic array faults,and the classification accuracy and algorithm stability are greatly improved.