| In recent years,the photovoltaic industry has developed rapidly.With the continuous increase of global photovoltaic installed capacity,problems such as the difficulty in operation and maintenance of photovoltaic power stations,unstable power generation,and poor controllability have been exposed,which have become key issues that the photovoltaic industry needs to solve urgently.Based on the grid-connected photovoltaic power plant simulation model,actual photovoltaic power plant fault data and power generation data,combined with the popular machine learning and deep learning methods,this article has done following research on the photovoltaic power plant intelligent operation and maintenance and power prediction technology.(1)This article briefly introduces the development status,existing problems,and solutions of the photovoltaic industry.(2)This paper introduces the main classification and composition structure of grid-connected photovoltaic power stations,equivalent physical models of photovoltaic cells and photovoltaic arrays,maximum power tracking control(MPPT)and voltage source converter control(VSC)for photovoltaic power generation,and Based on this,a simulation model of a typical grid-connected photovoltaic power plant is build.(3)This paper proposes a method for evaluating the operating efficiency of photovoltaic power plants based on data envelopment analysis(DEA)and BP neural network,which can evaluate the operating conditions of the power plant in real time.(4)This paper proposes a hierarchical-SoftMax-based photovoltaic array fault detection method and a weight-tree-based photovoltaic power station fault detection process to improve the operation and maintenance efficiency of photovoltaic power stations.(5)This paper proposes a photovoltaic power plant power prediction method based on weather clustering and parameter adaptive adjustment.Compared with traditional prediction methods,the accuracy of power prediction is improved. |
PDF Full Text Request