Research On Short-term Load Forecasting Based On Photovoltaic Output And Electric Vehicle Grid Connection

In response to the energy shortage,many countries have started to introduce renewable energy on a large scale,but due to the influence of light,temperature,and other weather conditions,the new energy output is more volatile,making the instability of the power system increase.Meanwhile,with the rapid development of electric vehicles,the charging load of electric vehicles has caused a certain impact on the power grid.Therefore,accurate load prediction is helpful to enhance the adaptability and robustness of the power grid,which becomes an important research direction to solve energy problems.Based on this,this paper proposes a short-term load forecasting model based on PV output and EV grid connection.By studying the impact of PV output and EV charging load on the grid,we establish various corresponding load forecasting models,analyze and compare the advantages and disadvantages of the models,and finally achieve the purpose of improving the accuracy of load forecasting to ensure the efficient and stable operation of the power system.The main work of this paper is as follows:(1)To address the problem that traditional load forecasting does not consider the impact of new energy grid connection on load forecasting,the impact of grid connection of photovoltaic output and electric vehicles on electric load forecasting is fully considered,and a new model of short-term load forecasting based on electric vehicles and photovoltaic output is proposed to be considered.(2)Considering the problem that the abnormal data in massive power data are difficult to handle,the density estimation algorithm is proposed to identify and correct the abnormal data;for the problem that the algorithm operation efficiency is affected by a large amount of data,the similar day extraction based on K-means algorithm clustering is proposed to realize the effect of data dimensionality reduction and improve the efficiency of the forecasting model.(3)In terms of PV output,a model with kernel parameters and disciplinary parameters of the whale algorithm optimized support vector machine(WOA-SVM)is proposed to predict PV output under three scenarios of sunny,cloudy and rainy days,respectively,and the higher accuracy of the WOA-SVM prediction model algorithm is proved by introducing mean absolute percentage error(MAPE)to compare with other three prediction models;in terms of electric vehicle load Combining the type of electric vehicles,replenishment mode,starting charging time,and charging power,the Monte Carlo algorithm is proposed to simulate and generate the electric load curves of 10,000electric private cars,electric buses,and electric cabs,and the total electric vehicle load curve is obtained by superimposing the three types of load curves;in terms of electric load prediction,by using the beta function,adding the adaptive weighting factor,modifying the nonlinear The average absolute percentage error(MAPE)of the improved whale algorithm optimized support vector machine model for short-term electric load forecasting is 0.73%and the correlation coefficient(R~2)is 99%,which is similar to that of the particle swarm optimization support vector machine(PSO-SVM)and the improved whale algorithm optimized Long Short Term Memory Recurrent Neural Network(IWOA-LSTM)and the original Whale Optimization Support Vector Machine algorithm(WOA-SVM),the MAPE decreased by 2.03%,0.93%,and 0.21%,and the R~2 improved by 8.34%,4.77%,and 0.18%,respectively.This study has higher accuracy compared to the traditional prediction method.The study shows that the large-scale integration of PV output and EV charging load into the grid puts higher requirements on load forecasting and load balancing.The results of this study can provide valuable information and decision support for the management and operation of the power system.