Research On Power Prediction And Model Fault Diagnosis Of Photovoltaic Power Station

Under the “dual carbon” target,the construction of a new power system with renewable energy as the main part has become the inevitable trend of China’s energy and power transformation and development,among them,vigorously develop photovoltaic power generation is one of the key technical ways.However,in recent years,the explosive growth of installed capacity of photovoltaic power generation system,the intermittent and volatility of its contribution makes it difficult for the power grid to optimize the scheduling precisely,in addition to the harsh operating environment makes the photovoltaic module failure frequency and efficiency decline,these problems bring great challenges to the safe and stable operation of the system and the consumption of renewable energy.Obviously,the study of power prediction model and component fault diagnosis method of high-efficiency photovoltaic power plant is one of the effective means to solve the above problems.In this thesis,the mathematical model construction of photovoltaic power plant,the influence factors,power prediction modeling and fault characteristic analysis have been studied to provide support for the efficient operation and maintenance of photovoltaic power plant.First of all,this thesis takes into account that the mathematical model of photovoltaic power plant is of great significance to the operation and analysis of the unit,on the basis of a brief introduction to the composition of photovoltaic power generation system,this paper discusses in detail the characteristics of photovoltaic modules,Boost and inverter system,control system and other modules,and establishes the mathematical model of photovoltaic power generation system,which lays the foundation for the research of the latter.Secondly,in view of the difficult problem of output power prediction of photovoltaic power plant under complex meteorological changes,the influence of total radiation,direct radiation,scatter radiation,ambient temperature,humidity,component temperature,wind speed,wind direction and other factors on output power is analyzed,and the output power prediction model of photovoltaic power plant based on circular convolutional neural network is established on the basis of multi-regression analysis to avoid the influence of high-dimensional influence factors on the predictive effect of the model.The feasibility verification was carried out in combination with the measured meteorological factors and output power of a photovoltaic power plant in Linze,Gansu Province.Finally,in view of the fault diagnosis of photovoltaic power plant components,the basic characteristics of common ash accumulation,local shadow and aging fault are discussed,three typical fault simulation models are established,the I-V and P-V curves of components under fault and the electrical characteristics of short-circuit current,open-circuit voltage,output power and component temperature are analyzed,and the machine learning-based PV module fault diagnosis algorithm is established,and the feasibility of photovoltaic module fault diagnosis research is verified by combined with the simulation results.Based on the reduction of the power influence factor of the output power of the photovoltaic power plant,a power prediction model of the output power of the photovoltaic power plant based on the circular convolutional neural network is established.The verification of the example data shows that the power prediction model can accurately predict the output power.In addition,the electrical characteristics under the typical fault of photovoltaic modules are found and analyzed,and a fault diagnosis algorithm based on Light GBM is proposed,the simulation results show that the algorithm is accurate and the fault diagnosis effect is good,and the research results can provide guidance and suggestions for optimizing scheduling of photovoltaic grid-connected systems and fault analysis of photovoltaic power plant components.