Research On Power Load Forecasting Method Based On Grey-markov Chain

Electricity,as an important energy industry,plays an important role in the advancement of society and the development of the national economy.With the development of the country and the improvement of people’s living standards,the demand for electric energy is increasing day by day,and the stability of power supply and the quality of electric energy are facing greater challenges.In order to meet the needs of the development of the power market,and to make the power planning which is close to supply and demand,this paper researched medium and long term power load forecasting,and focused on exploring load forecasting methods.Power load forecasting is affected by a variety of deterministic and uncertain factors such as policies,market economy,meteorology,natural disasters,etc.Therefore,this thesis takes gray theory as the basis and deeply analyzes the change law of historical load data itself.Then we analyze and explore the inner connection between load data and the law of development and change,and build an accurate and highly applicable load forecasting model.First of all,the paper analyzes the applicability of a single forecasting model.The traditional gray prediction model GM(1,1)is only suitable for predicting data series with exponential trend monotonically,and the gray Verhulst model is only applicable to data with "S" type development trend.Due to the influence of various factors,the data curve is often fluctuating and has a complex trend,and the error generated by prediction with a single model is too large.Therefore,to address the problem that a single model is not applicable to fluctuating data series,this paper proposes a combined gray forecasting model to extract the advantages of GM(1,1)and gray Verhulst models.The entropy weight method is applied to the combination model to determine the combination weight coefficients by the degree of error dispersion of the fitted values.Then the combination model is built to improve the adaptability and reliability of the model.After that,due to the randomness of many load influencing factors,the Markov chain in random theory is introduced to correct the individual singular data,and furthermore get the load prediction value with higher accuracy.Finally,the stability analysis of the gray combination prediction model is carried out.Due to the large amount of the original load data,in the process of solving the coefficients of the gray model,when the inversion of the matrix is involved,the condition number will be too large,resulting in the pathological state of the equation set.In this paper,the adjustment unit of measure method is used in order to transform the pathological matrix into a benign matrix,while the least-single product method is given to avoid the matrix inversion process and improve the model stability and accuracy.The present results can provide effective methods for medium and long-term power load forecasting,provide the suitable forecasting models for different data series,and provide a basis for power generation.