The Development And Application Of Object-oriented Integrated Software For Power Load Forecasting

With the unprecedented development of electric power industry, accurate load forecasting is beneficial to the development of electric power system planning; can guarantee the safely, stably and economically of electric system; contributes to making the maintenance plan for unit maintenance; help to make power market transaction and modern management electrical power system. Over the years, power system workers have accumulated a lot of experience in load forecasting area and developed many practical and mature forecasting methods. At the same time, take the advantage of computer technology progress, realizing the precious method by computer language will not only liberate Load forecasting workers form complex mathematical calculation and confused data, but also increase the forecasting accuracy.The load changing trend and its forecasting model is deeply analysed by the thesis. According to the different load time limit, the paper classified load forecasting to yearly load forecasting, quarter load forecasting, monthly load forecasting and daily load forecasting, compared and analysed the load changing trend in every time limit, researched the influence of different influencing factor on each load in its time limit.This article built many kinds of targeted load forecasting computation model, at the base of analysing load changing trend, explored the new forecasting ways and method. It also classified and listed each forecasting model with counting principle. Such as daily load forecasting method contains holidays point-to-point times analogy method based on similar days load, and also have intelligent computation methods based on relevant factor-such as artificial neural networks.Based on the load forecasting computation model, a forecasting software is developed. Aims at the strategy for implementation algorithm and its software programming scheme, the thesis formulated the action plan to solve this problem separately and creatively.The software contains four forecasting method library comprised of yearly load forecasting, quarter load forecasting, monthly load forecasting and daily load forecasting, every method library provide more than 20 different basic forecasting methods, each methods can be subdivided to 3 to 5 algorithm with the difference of order and mathematical model. The software nearly contains most of existing mature method.At the base of all single forecasting methods, the software built integrated forecasting model, provided various weighted combination program. The program can choose some single forecasting methods and finish integrated forecasting. User can choose forecasting methods freely and confirm the forecasting program which fit practical situation best.At the same time, the software have thorough data processing function, including data auto input, bad data display and processing, outputting and printing report forms of forecasting result. The software can also calculate and analysis the parameter forecasting results uniformly. The thesis also forecasted load with the software, took crosswise comparison between different forecasted methods.The thought to make software and process is introduced in detail by the thesis. The software follows the thought of unification and modularization. The software is easy to extension method and perfecting function, highlighted the user oriented design concept at the base of guaranteeing forecast accuracy.The software can not only satisfy the needs of related worker in electric power enterprises, but also used at scientific research fields as a general utility tool with its fast computing ability and data analysis function. The software easy to use and install, have friendly man-machine interface, rich in prediction method, perfect in function, have good practicability and promotional value.