Modified Forecast Of Short-Term Power Load Based On QPSO-ELM-KF

Electricity,as the nerve center of contemporary production,life and economic development,is related to the national economy and people’s livelihood,and its importance is self-evident.To ensure the normal operation of the daily power system and meet the requirements of its production activities,power economic dispatch and grid safety analysis,the short-term load forecast of the power system is an indispensable content.When it comes to prediction,accuracy and stability are what we are particularly concerned about.At this stage,a lot of research has been done in load forecasting at home and abroad.In this paper,in order to improve the accuracy and stability of prediction,a short-term load forecasting model of power system which is composed of the quantum particle swarm optimization(QPSO)optimized extreme learning machine(ELM)and then combined with Kalman filter(KF)is proposed.The model first predicts the power load value at each time point through ELM.Among them,according to the characteristics of the QPSO algorithm itself and its advantages in parameter optimization,it is used to optimize the weights of the input layer-hidden layer and the threshold of the hidden layer in the ELM network structure.Then,the KF algorithm is used to further update and optimize the obtained predicted value,so as to obtain the optimal estimated value at each time.Finally,analyze and judge the overall performance of the model.In addition,in order to further improve the accuracy of power load forecasting,this paper modifies the forecast results of the QPSO-ELM-KF combined model for time periods and performs experimental verification.The results show excellent performance.The specific work of this paper is as follows:1.Preprocessing of power load historical data.The basis of model prediction is data.In order to ensure the correctness and accuracy of the research on the law of load changes,this paper analyzes the collected power load data,and screens and processes the missing values and abnormal values in the data.After constructing the learning samples,they are normalized for model analysis.The analysis results are de-normalized to calculate the prediction accuracy of the error judgment model.2.The QPSO-ELM model is preferred.This paper uses the characteristics of PSO and QPSO algorithms combined with ELM to build PSO-ELM model and QPSO-ELM model.By comparing the performance of PSO-ELM and traditional PSO-BP,the advantages of ELM algorithm for load forecasting are analyzed.Then compare the performance of QPSO-ELM and PSO-ELM to analyze the superiority of QPSO for parameter optimization.Finally,the QPSO-ELM model is regarded as the better.3.Construction of QPSO-ELM-KF combined model.Taking into account the impact of data dispersion on the prediction accuracy,the prediction results of the QPSO-ELM model are further optimized by the KF algorithm.Update the error analysis matrix,and complete the optimal estimation at each moment.Thereby reducing the impact of the overall model prediction accuracy reduction caused by the high dispersion point prediction deviation.4.Time-division load forecast revision.In view of the accumulation of errors caused by rolling forecasts,this paper proposes a time-phased load forecast revision based on the QPSO-ELM-KF combined model.Combine the forecast results at each time of the combined model with the results of its segment forecasts to calculate the load correction.By revising the load forecast value at each moment,the overall prediction accuracy of the model is improved.