Analyze And Research Of Solar PV Array Failure Prediction

In today's increasingly serious environmental pollution,photovoltaic power plants are rapidly gaining share in the power generation market with a clean,rich,environmentally friendly and renewable advantage.However,the fault occurrence of photovoltaic power stations running at home and abroad mainly takes regular diagnosis and maintenance,which leads to the long outage time and high maintenance cost,unable to accurately and timely determine the fault location and reasons;unable to predict the failure information in time to repair.Based on the actual situation,this paper designs a monitoring system to realize real-time monitoring and fault location.The improved BP NN algorithm for PSO is proposed.The prediction model for PV fault is established.The prediction of PV fault array is realized.The specific study contents:First,starting from the PV array structure,the principle of photovoltaic power generation,and based on the mathematical model of the fault condition were summed up and the simulation analysis,find the different fault types of data characteristics of the different fault types of quantitative indicators.Secondly,based on the influence factors and faults of photovoltaic power generation,a BP neural network algorithm with self-learning ability and nonlinear mapping function is selected to predict the failure.In order to improve the prediction accuracy of this prediction model,a new algorithm is proposed to optimize BP neural network using particle swarm optimization.A nonlinear dynamic adjustment strategy is used to optimize the inertia coefficient of particle swarm algorithm to achieve global optimization and local optimization.Then,a photovoltaic fault monitoring system was established to analyze the performance of the monitoring system.Hall sensor designed the system.The monitoring interface is developed and the generation efficiency of the real-time monitoring system is developed.The fault node can be found in the first time,so as to facilitate the maintenance of the staff,so as to improve the efficiency of photovoltaic power generation.Finally,based on the analysis of the fault characteristics of PV system,the weight and threshold of BP network were trained using the improved PSO algorithm.The prediction model of PSO-BP photovoltaic fault was established.It can accurately and timely predict the location and time of the PV fault.Compared with the traditional BP neural network,this can overcome the problem of fitting and local minimum value,which has good robustness and generalization,improves the prediction accuracy and reduces the prediction error.