Research On The Distributed Pv All-weather System And Output Prediction Method

PV power generation output has obvious intermittent fluctuation characteristics,large-scale photovoltaic power plant has brought great challenges to the power grid scheduling management.The prediction of photovoltaic power generation is one of the key technologies to solve this problem,and comprehensive weather parameters are the basis of PV output prediction.It is of great academic and applied value to carry out the research of meteorological system design and output prediction method for photovoltaic power generation.In this paper,a weather monitoring system of photovoltaic power plant is developed for the problems of scalability and versatility of traditional meteorological data acquisition system.According to the characteristics of different weather sensor output signal and the hardware circuit of the meteorological monitoring system,based on the embedded technology,we developed the sensing terminal for meteorological parameters.The sensing terminal data is transmitted to the RTU via CAN and then pass to the host computer system through the TCP/IP.The system has been put into use in the roof photovoltaic power plant,the actual operation has achieved good results,with high precision measurement,user-friendly,scalable and so on.The photovoltaic power generation is affected by the multivariate meteorological factors,and the selection of the input variables of the forecast model directly affects the prediction accuracy.In this paper,the relationship between photovoltaic power generation and multivariate meteorological factors is studied,and the weight coefficient is defined as an evaluation index to measure the effect of meteorological factors.According to the principle of photovoltaic power generation and the availability of data,the meteorological factors used in the prediction of subsequent photovoltaic output are determined.Aiming at the problem that the traditional predictive model structure is difficult to meet the requirement of prediction accuracy,the weather state pattern recognition model based on K-means clustering is established by extracting the feature quantity reflecting the weather state in the historical data.The model makes full use of the spatial correlation and shape similarity between historical data.Through the simulation experiment,the accuracy of the model has been improved obviously.In order to improve the prediction accuracy of the model,an Elman neural network photovoltaic short-term output prediction model with the introduction of the excitation function is established.The improved Elman neural network solves the problem of insufficient dynamic performance of traditional Elman neural network and improves the modeling efficiency.The actual data of the PV plant monitoring conducted by Portland State University are used to verify the model.The relative root mean square error of the model is 5.44%.By improving the excitation function,the relative root mean square error is reduced to 3.43%and the accuracy is improved by more than 2%.The experimental results show that the model can adapt to the prediction of PV output under various weather types.The meteorological parameter sensing network developed in this paper is universal.The pattern classification method based on weather status and The improved Elman neural network model has the characteristics of conciseness and high efficiency and high prediction accuracy,It has certain application value for the popularization and promotion of distributed photovoltaic power generation system.