Research And Development Of Lack-of-power Index Forecasting System

With the development of the GNP of China, local electrical load also increases year by year, especially the promotion of the citizen's living standard, residential load, especially the proportion of air-condition load which causes the serious situation of short of electricity in China recent years becomes larger and larger. So DSM measures are being taken all over China. Such measures include load management and improvement of terminal energy efficiency. But most of them are emergency measures. We can do much better if we can predict the future situation. So it's necessary for power system to be assisted by an earlywarning system that can be used as a reference for daily production and control.So-called lack-of-power index is calculated by local future load demand and the capability of power supply. Then local lack-of-power degree will be acquired. By the index the system can ascertain the local grade of absent electricity power and give some advices to the decision-maker. Meanwhile citizens and enterprises can know the index and degree through media. This system is similar to a small earlywarning system.For climate has great influence on the day maximum load, the model considers these factors such as temperature, humidity and so on. In order to decrease the number of independent variable and simplify the model, the paper uses Comfort Index of Human Body to replace previous conventional climatic-gene. Meanwhile this paper proves that using Comfort Index of Human Body in load forecasting is feasible and convenient by comparing an example.Max-load prediction of a few days later is key point of the earlywarning system. Linearity Regression; Artificial Neural Network; Gray Prediction are conventional arithmetic. But we find that forecasting accuracy is rather worse especially the maximum of relative error is bigger by these arithmetics. So a forecasting arithmetic based on load-decompose, which divides day maximum load into normal load and weather sensitive load, is mentioned here. It uses composite model combining exponential increasing trend and cycle trend to forecast the normal load and uses artificial neural network to forecast weather sensitive load. The final result proves the method is feasible.At last the paper introduces the process of construction of this system simply, including main modules and function of the lack-of-power index system as well as some websites are shown.