Research Of Cross-platform Short-term Load Forecasting System

Short-term load forecasting is one of the most important routine works for power dispatch centers. High accuracy of short-term load forecasting is a guarantee of the economics and security for power system operation.As is well-known, power load will fluctuate with variations of temperature, rainfall, wind and typical days. It means a big uncertainty and is very difficult to forecast the day-ahead load accurately. This paper provides innovative ideas for the designation of load forecasting algorithms.Research directions focus on the following key points:The first one is Multi-Points Extrapolation & CART (Classification and Regression tree) short-term load forecasting algorithm. Based on the research that power load could be divided into different parts, the main innovation of the method is to forecast the different parts of the power load with different method. The forecasting result of this method displays that it has a brilliant performance.The second one is Similar-Point of short-term load forecasting. It offers the definition of similar-point. Based on the tools of decision tree and Genetic algorithm, the method establishes an expert system to forecast day-ahead load.The third one is short-load forecasting application's cross-platform. The application is developed with Java. It could run on different operating systems such as Windows, Unix, Linux, Solaris etc. More important, the flexibility and openness of the application is expanded by advanced software development ideas such as design patterns and so on.The last one is database cross-platform. Through using the innovative database technology Hibernate, it is convenient to change in different database platforms for short-load forecasting application, such as Oracle, Sqlserver, DB2 etc. This innovation promotes the flexibility of the application.