| The load forecasting is always an important base for the operation and scheduling of power system, as well as for the well-running of power market. With the power market open, a precise, intelligent, reliable and real-time load forecasting is in need.In this paper, the writer tried to show a detailed process of using wavelet analysis to forecast short-term power load by building the relatively simple and fast mathematic modes. By analyzing the load character, the writer showed the periodicity of load time series; by analyzing the theory and arithmetic of wavelet analysis, the writer showed the feasibility of load forecasting by wavelet analysis.This paper was a further study on one-hour short-term load forecasting after the writer developing a load forecasting software package by grey mode in order to find out the advantage of wavelet analysis in load forecasting.On the basic theory of the relationship between Lipchiestz exponent and maximum module, the paper showed the way to choose a suitable wavelet function before analyzing the curve.On the basic theory of wavelet and part irregularity, the writer checked out the irregular signals and adjusted them to a more smooth level so that the regularity of signal could be obvious.By deposing the signal into convergence of low frequencies and high frequencies, which reflect the periodicity of series, the writer used PAR mode to deal with the mid and low frequency time series while dealing with high frequency series by exponential smoothing method. The reconstruction process allows the last forecasting result.The simulating result proved the feasibility and advantage of this method. |
