Research On Chaotic Theory And Its Application In Short-Term Electric Power Load Prediction

This paper summarized and analyzed the basic conception, status and application of chaos theory, the methods to distinguish the chaos time series, reconstruction of phase space, the methods to compute the parameters, such as Lyapunov exponent, etc.This paper studied the chaotic trend combined model of electric power day-load record. Because the day-load trend was obvious, it was decomposed to two parts: trend series and leaving series. On the basis of the weighting first-order local predict method, this paper brought in the degree of incidence, raised by big percentages the accuracy of prediction. This paper also brought in the cosine value of the trajectories evolvement direction angle, as part of the weighting. Accordingly the degree of correlation of trajectories was more accuracy. This paper analyzed and studied the Lyapunov exponent prediction method and the weighting first-order divergence degree prediction method.In the end, this paper combined the chaotic theory and the neural network, trained the recursion network again and again, then we got the satisfying predict result.