Short-term Power Load Forecasting Based On PCA-GWO-BP Model

The safe and stable operation of the power system is closely related to the development of national economy and the safety of people and properties in the society.Accurate short-term load forecasting is essential for guiding the power system to make power generation and maintenance plans,coordinating unit operation and scheduling load distribution.With the growth of brilliant grid and the pay a visit to more and more distributed force sources,the factors touching the preciseness of force burden prediction are becoming more and more compound,making the force burden more non-linear,which also puts forward higher requirements on the preciseness of short-term burden prediction.The key skill of motor-driven burden is the prediction skill and the extract of mathematical pattern,and the chief work of this dissertation is to study and improve the BP nervous net pattern,so as to improve the forecast preciseness.This treatise firstly holds the study backdrop and meaning of short-term electric burden forecasting,and extracts the condition position of internal and overseas electric burden forecasting ways,and resolves the merit and no merit of each prediction way.Among the present prediction ways,BP(Back Propagation)nervous net has been widly used in short-term force burden prediction because of its adaptability,research methods and mistake step,and has got more satisfactory outcome.However,the BP nervous net pattern itself also has some problems,such as low prediction exactness slow,junction and liable to decline into partial minima.In order to resolve these problems,this thesis uses a BP nervous net use based on principal component analysis,which depends on the better entire find function of Grey Wolf Optimizer(GWO)for The weight threshold of the BP nervous net is optimized to improve the shortcomings of the BP nervous net arithmetic.Secondly,so as to assure the preciseness of short-term burden prediction for Lixin County,it is usually required to premeditate a variety of burden affect elements,which cause in a large number of dimensions of historical data,and these influencing elements will be key in pattern as data,which will lead to a growing practice burden of the pattern.Therefore,Principal Component Analysis(PCA)is used to decrease the number of keys into the pattern and substitute a large amount of data with a small number of data keys in,which can remove superfluous messages while keeping the most significant message,thus improving the short-term prediction accuracy.Finally,the PCA-GWO-BP nervous net mode economic in this paper is validated from the feature of the electric burden in Lixin County and integrates with the true regional high and low temperatures and daily type material using the Python programming language platform.By contrasting and analyzing the outcome of PCA-GWO-BP prediction with those of traditional GWO-BP,GA-BP and BP,the outcome shows that the PCA-GWO-BP nervous net pattern predicts less mistakes and the predicted and actual worth are more persistent.Therefore,the PCA-GWO-BP pattern is used to forecast the short-term electric burden in Lixin County,which is more appropriate for the real local circumstances and provides a basis for the future production and scheduling of the strength grid by Lixin County Force Strength Supply Company.