Research On Design And Power Generation Prediction Method Of Loufan Photovoltaic Power Station

In recent years,global energy consumption has been increasing,and countries have vigorously advocated the development of renewable energy.Among them,solar energy is the most important renewable energy.The rational development of solar energy resources is in line with the development direction of energy industry policies and the sustainable development of regional energy structure.The construction of solar photovoltaic power station and the prediction of photovoltaic power generation and power are of great significance to the optimization of energy structure,the suppression of power grid impact and the safety of power grid.Based on the above research background,the main research works of this paper are as follows:(1)According to the construction standard of solar power station,it is determined to build 30 MW solar power plant in Loufan,Taiyuan city.Pvsyst is used to calculate the tilt angle of solar panel.Helios 3D is used to conduct shadow analysis,and the distribution and arrangement of solar array are determined.At the same time,the solar panel model,array structure and inverter selection design are completed,and the boosting mode is determined.Finally,the design of the power station is completed.(2)In order to effectively improve the prediction accuracy of photovoltaic power generation,a new short-term prediction model of photovoltaic power generation based on ant colony algorithm optimization of Long Short-Term Memory(LSTM)was proposed.Firstly,using the ant colony algorithm to optimize the weights and thresholds of the traditional LSTM neural network,and then substituting the optimized weights and thresholds into the LSTM neural network again,and repeating the optimization until the accuracy meets the requirements and output the optimal value.The prediction model after optimizing the LSTM neural network by the ant colony algorithm is used to predict and analyze the photovoltaic power generation of Loufan power station in three weather types:sunny,cloudy and cloudy.Then comparing and analyzing the predicted value of the LSTM neural network model before optimization.The results show that the LSTM neural network prediction model after ant colony optimization has higher accuracy and is more suitable for the short-term prediction of photovoltaic power generation of this power station.(3)The existing photovoltaic power prediction method solves the nonlinear problem of the historical power data of photovoltaic power plants to a certain extent,but cannot take into account the timing problem.In order to make better use of historical valid data,take into account the time series and nonlinear characteristics of data,and improve the short-term power prediction accuracy of photovoltaic power plants,A new short-term power prediction method for photovoltaic power plants based on Kmeans clustering and particle swarm optimization(PSO)optimization convolutional neural network(CNN)and LSTM neural network is proposed.This method combines the respective advantages of CNN and LSTM,First,the approximate daily data is clustered by Kmeans,and then the learning rate and the number of hidden layer nodes of the CNN-LSTM neural network model are optimized by PSO.Finally,the prediction model of photovoltaic power generation power is built by LSTM.By comparing the LSTM model and the CNN-LSTM model,it is verified that the prediction method is more accurate.The design and construction of large-scale solar photovoltaic power plants have promoted the increase of the proportion of clean energy.The prediction model based on deep learning improves the accuracy of power generation and power generation prediction,suppresses the impact of solar power plants on the power grid and is conducive to the reliable grid connection of large power plants.