Co-integration Analysis Of The Relationship Between Electricity Demand Forecasting And Industrial Adjustment

Power system load forecasting is the basis for power network planning work; theprediction accuracy will directly affect the quality of network planning，and healthydevelopment of the power industry and the whole economy．The main role oflong-term load forecasting is to provide basic data for power and power networkplanning．Therefore，study the forecast model that which can reflect load trends overan extended period，improve accuracy of long-term load forecasting has greatpractical and economic sense．Since2000，the rapid development of China’s economy，the adjustment ofindustrial structure，and implementation of energy conservation policy，so that thepower consumption structures are more reasonable，and thus the power consumptionof the whole society has also undergone a big change．But the impact of the economiccrisis and adjustment policies mutation factor also makes greater fluctuation inelectricity consumption．Therefore，from the perspective of the trend of industrialdevelopment，and consider the mutation factors，research the relationship and trendbetween electricity consumption of the whole society and the trend of industrialdevelopment，has important practical significance for power grid planning．The thesis is based on the discussion and analysis of the characteristics of thechanges of power load，and the main factors that impact the power system loadforecasting．By analysis the impacts that the industrial adjustment of China on thechanges of electricity structure，and the impact on the changes of the powerconsumption． We know the power consumption of the country is essentiallymaintaining the same trends with the economic development and the industrialadjustment．The thesis elaborates the concept of the variable co-integration of parameters，the methods of co-integration test， and the establishment of error correctionmodel．Meanwhile，based on the study of the neural network of radial basis function(RBF)， and using the characteristics of function approximation of RBF network．Propose the method that firstly determines the possible structural breakpoints by Chow test，and then take the possible mutation points as the input data ofRBF neural network for fitting and analysis the power consumption，and taking theminimum point in time of sum squared errors and mean square error，as the pointmutation of co-integration between variables，finally establish the mathematicalmodel of variable co-integration of parameters and error correction model betweenthe power consumption and industrial adjustment．Through practical examples，verifythe method is more effectively to improve the accuracy of load forecasting than theregression model which not considered the mutation factors，and also verify thefeasibility of this forecasting method is used to forecast the power load and theconclusions is effectiveness．...