Research On PV Station Fault Detection Based On Improved Clustering Algorithm And Association Rules

Intelligent operation and maintenance of photovoltaic station mainly uses the monitoring system to monitor the operation status.At the same time,the photovoltaic intelligent diagnosis method is used to diagnose the faults of the photovoltaic station,so as to achieve the purpose of intelligent operation and maintenance.Intelligent operation and maintenance technology can reduce the maintenance cost of the photovoltaic station and guarantee the users'income.At present,fault detection based on power prediction is difficult to carry on,and the accuracy is greatly affected by the environment.Therefore,this paper studies the comparative method of adjacent power stations which has much engineering feasibility.In this paper,the photovoltaic station is taken as the research object,and the rule relationship between photovoltaic data and the problem of power plant fault detection and location are studied.The main work of this paper is as follows:(1)By studying the factors affecting photovoltaic power generation,the input characteristic parameters of the association rule mining model are determined.After cleaning and discretizing the characteristic parameter data,the association rule mining model based on Apriori algorithm is used to mine the rules between photovoltaic power generation data,which provides a basis for judging the subsequent fault detection model.(2)In order to improve the clustering performance of DBSCAN clustering algorithm,the DBSCAN algorithm is improved,and the improved algorithm has higher performance.(3)According to the results of association rules mining,the input characteristic parameters of the fault detection model are determined.The photovoltaic power generation fault detection model based on improved A-DBSCAN clustering algorithm is used to detect the fault.The validity and efficiency of the model are verified by experiments.The fault detection method which has a high accuracy proposed in this paper solve the problem of the lack of fault sample data and can be independent of the fault sample data.This makes an important reference significance for the current study of photovoltaic fault detection methods.