Location:Home > Engineering science > Hydraulic engineering > Hydrology and Water Resources > An Improved Method for Short-term Load Forecasting Based on FFT and Chaos

An Improved Method for Short-term Load Forecasting Based on FFT and Chaos

Downloads: []
Tutor: WuYongGang
School: Huazhong University of Science and Technology
Course: Hydrology and Water Resources
Keywords: Short - term load forecasting,FFT,Chaos,Phase Space Reconstruction,Weighted rank
CLC: TM715
Type: Master's thesis
Year:  2011
Facebook Google+ Email Gmail Evernote LinkedIn Twitter Addthis

not access Image Error Other errors

The short-term load forecasting is the basis of the power system, the prediction accuracy of a direct impact on the safe operation of the entire power grid and electricity load allocation. Short-term load forecasting gradually attracted attention, and studies have shown that load with periodic and chaotic characteristics of chaos, so the \hot spots. FFT-based chaotic time series prediction method using FFT method separating the first load cycle, chaos and the remaining load forecast in chaos prediction method of phase space reconstruction based on weighted rank local step prediction model has been widely used for the prediction of chaotic time series. In this algorithm, the to all separation cycle frequency value and the introduction of load forecasting value, making the algorithm frequency value of the life-cycle to extract excessive and cumulative error to solve these problems, its improvement research, the main work is as follows: 1) load time series can be divided into periodic, chaotic and random noise of three parts, the portion of the separation cycle, the use of the FFT of the load is converted into the frequency domain, by automatically identifying the cycle frequency components in the frequency domain, all of the cycle frequency component inverse transform to when The domain periodic component will cause excessive cycle parts extracted. Proposed for this problem, the periodic component separation cycle, chaotic and random part of this component of the contribution, according to the chaotic, random spectral characteristics to determine the amount of periodic separation. 2) the weighted rank local multi-step prediction in neighboring points in the multi-step predictive value, and this value substituted into the algorithm predicted, would cause cumulative error. To solve this problem, this paper proposes an additional factor in the weights of the neighboring points adaptive correction to reduce the accumulated error. 3) In this paper, the research and analysis of the residual load, the remaining load of random noise is dealt with. The remaining Chaos load at the same time the use of wavelet transform chaotic noise separation, and then processed separately. In addition, through the improvement of the load forecast algorithm tests prove that the improved algorithm has better accuracy. Load forecasting, Hainan power grid load forecasting, for example, its use of improved forecasting methods to predict to be satisfied with the results, to improve the prediction accuracy.
Related Dissertations
Last updated
Sponsored Links
Home |About Us| Contact Us| Feedback| Privacy | copyright | Back to top