Research On Classify Forecasting Of Regional Short-term Load

Short-term power load forecasting is of great significance to the operation and maintenance of the power system.Especially for the municipal electric power sector,and the accurate forecast of the power load.Help to arrange power plant planning,system operation,plan and repair work.It can enhance the safety and reliability of the power system,and improve the economy of the operation of the power system.The current short-term load forecasting method is divided into two main categories : first is the prediction of more traditional methods,mainly based on time series method and regression analysis method;other is in the development of artificial intelligence technology,gray theory,expert system method,has the neural network theory,fuzzy prediction theory prediction method.This paper is intended to summarize a set of practical forecasting methods to improve the accuracy of load forecasting.In this paper,taking Zhangjiakou area as an example,the load characteristics in this area were analyzed for the whole load area,affected by climate change,factors such as the impact of the electrified railway,through time series analysis,the traditional prediction method,regression analysis method and other traditional forecasts accuracy is not ideal.However,the intelligent forecasting platform for the construction of load forecasting expert system and artificial neural network system is too large to be used in a single area.The whole area according to the specific character of load load factors and variation characteristics,refined and classified into a plurality of single load and load fluctuation regular sub network,through the accurate prediction of each sub network load,can predict the level of bottom-up to improve short-term load area.Firstly,according to the character of load classification will be divided into industrial area overall load load,load,rural residential electricity and commercial load four categories,then in-depth analysis of the load characteristics of each kind of load,and to further refine the decomposition.Will the overall load in Zhangjiakou area is divided into large enterprises,small and medium enterprises electricity load electricity load,the electrified railway load on the dam,dam life load,rural residents living under load,rural residents in 13 counties of agricultural production,commercial and residential electricity load normal daily load,residential electricity and commercial holiday load a total of 23 sub networks.The load characteristics of each sub network are single and the load variation regularity is strong.The traditional load forecasting methods such as time series and regression analysis can be used to obtain high accuracy data.Finally,the forecast data of each sub network will be synthesized by the method of a certain subnet accumulation.This method is used to carry out short-term load forecasting in cities and regions,and has a great improvement in the overall load forecasting.The forecasting results are very satisfactory,and the overall accuracy rate is above 97.5%.But the daily work is complicated,this paper with the construction of a short-term load forecasting system and forecast area,accumulation method of each sub network built into the system,the system prediction routine,only with simple manual modification,we can get the final prediction results,the method is applicable to the power sector forecast work under the jurisdiction of the short-term load region,has a strong promotion.