Abstract: With the large-scale application of photovoltaic power stations, a key component, the filthy state of photovoltaic panels directly determines the power generation efficiency and economic benefits.However, it brings a huge difficulty weather there is a reasonable cleaning decision for photovoltaic panels cleaning operations.Therefore, with the single chip processor as the core and connecting with the temperature and humidity, solar radiation sensor, PM2.5 sensor, PV voltage detection module design photovoltaic panels filthy state evaluation system achieve the maximum photovoltaic power station profit.Based on improved BP neural network to realize clean decision-making model, it provides a reference for building a new type of power system goal vision and improving a diversified clean energy supply system.