A Pattern-Recognition-Based Adaptive System for Short Term Load Forecasting

Short term load forecasting is the basis of power generation planning for power dispatch department.The flexibility and adaptivity of load forecasting system is the important assurance for the economic operation of power network.Based on the analysis of the major faceors that influence daily load,a load mode for short term load forecasting is defined.The similarity calculation of two load modes is given using the weighted Hanmin distance,by which the necessary historical load mode samples can be extracted efficiently.The artificial neural network is employed to map the forecasting load.A flexible and intelligent adaptive system for short term load forecasting is developed with object oriented program design method based on C++builder.Field application shows that the system is practical with satisfactory results.