Research And Implementation Of Deep Learning-based Monitoring And Cleaning Cycle Of PV Panel Ash Accumulation

The abundant solar resources provide favorable conditions for the photovoltaic(PV)power generation industry,which has a diverse market and great potential.However,the accumulation of dust on the surface of PV modules can impact their power generation efficiency,thereby affecting the profitability of PV power plants.Therefore,this study focuses on the phenomenon of dust accumulation on PV module surfaces.Based on the data collected from PV power plants,a deep learning model is employed to predict the clean power generation,and market research is conducted to determine the relevant factors for cleaning.Mathematical modeling is used to analyze the loss of power generation caused by dust accumulation,aiming to determine the optimal cleaning cycle for PV power plants.In this paper,the power generation data is obtained from a PV power station in Ningxia,China.Preprocessing techniques include outlier detection using the Isolation Forest algorithm and missing value imputation using Euclidean distance.The weather data is numerically processed using the Particle Swarm Optimization algorithm.A multilayer feed-forward neural network(backpropagation,BP)model is utilized to predict the theoretical clean power generation.To address the issues of the BP neural network,such as being prone to local optima,slow learning speed,and low computational accuracy,improvements are made using genetic algorithms and ant colony algorithms to optimize the initial weights and thresholds of the BP neural network.Furthermore,the weather data is categorized to further reduce the prediction error of the model.Finally,the predicted results are compared with the actual power generation to monitor the dust accumulation and determine the cleaning cycle and cost.Moreover,a distributed PV dust monitoring and cleaning cycle system is designed and implemented.It can display the distribution of PV power stations in Ningxia,the prediction of theoretical clean power generation,actual power generation,dust accumulation monitoring information,and cleaning alerts.