Research On Power Load Forecasting Based On Lasso-PCA And Improved Adaptive Genetic Neural Network

With the deepening reform of electric power market,load forecasting of power system has become one of the important means to reduce operation costs and improve economic and social values for power companies.Accurate load forecating of power system can enhance the security and stability of power grid operation.Under the momentum of rapid development of power industry,load forecasting of power system will provide help for dispatching operation and resource allocation plan,and load forecasting will play a more and more important role.Therefore,this paper proposed the power load forecasting method based on Lasso-PCA and improved adaptive genetic neural network is studied.First of all,the paper analyzed the power load data characteristics in detail,and pointed out the relationship between external factors and load changes.The paper describes the concrete implementation of load forecasting steps and analysis methods of prediction error.Secondly,in order to reduce the complexity and calculation of the prediction model,a data concision and feature extraction model of Lasso-PCA is proposed.This paper analyzes the concrete experiment to prove the method can compress multiple information variables and feature extraction.Then,the structure theory,algorithm steps,advantages and disadvantages of Back Propagation and Genetic Alogorithms are introduced.Then,it introduced adaptive genetic algorithm,and it is pointed out technique has the slow evolution of population,individual diversity.After that,it proposed an improved adaptive genetic algorithm.Establishing a new calculation formula of cross probability and mutation probability by introducing the individual evolution trend to adjust the concept of parameters,and the Sigmoid function.The IAGA is used to optimize the BP neural network,and it is used to overcome the defects of the BP neural network which is easy to fall into the local optimum.Finally,the paper analyzed the example data,the meteorological principal component,date types and historical load data as the modeling object and establish the BP neuralnetwork prediction model,GA-BP neural network prediction model,AGA-BP neural network prediction model and the IAGA-BP neural network prediction model to predict the actual power load in a certain area of northern China.The analysis shows that the proposed method can accelerate the convergence speed of the model and improve the prediction accuracy.