Broad Extrapolation Of Very Short Term Load Forecasting

Load forecasting is very significant to power system control,operation and schemes. To improve the accuracy can not only strengthen the security, but also can perfect the economical efficiency of power system operation. As a kind of load forecasting, very short term forecasting mainly predict the load from 5min to within 1h. As a crucial basic indicator, Mains frequency demands a balance between generating and power utilization. Economic dispatch,safety monitoring on line and AGC will adjust generating unit output of power plants in power grids ,according to the results of load forecasting, to make a new balance of power system and to keep mains frequency .Very short term forecasting is becoming more and more important, So the study of this subject has important practical significance.Electrical load is a system with strong periodicity and randomicity .On the one hand, it changes regularly by following certain tendency; on the other hand, it has a strong wave character affected by many factors. In connection with load characteristic, analysis and grab load's regularity, give attention to all the factors that are influential in load forecasting are necessary. Very short term forecasting data contains extrinsic influence, so the main influence is intrinsic characteristic of load.This paper collects 12 kinds of very short term prediction algorithms and analyzes their principles and relative merits. Considering main feature of very short term load forecasting, choose two methods named load derivation and increment extrapolation. It makes a principle analysis of the two methods, programs and calculates the results using Matlab software, realizes the application of these two methods, makes a statistical analysis of average relative error and error distribution. Furtherly, it proposes a new load forecasting method broad extrapolation by taking transverse influence and longitudinal influence into account defines and analyzes weighting factor. It realizes broad extrapolation using Matlab program. Numerical examples illustrate the efficiency of the proposed method .Simultaneously, select the appropriate weight factor and make some new discovery.It analyzes factors that influence load forecasting in view of load's inherent law; broad extrapolation exemplified the importance of load's inherent law.