基于极点对称模态分解-分散熵和改进乌鸦搜索算法-核极限学习机的短期负荷区间预测

针对确定性负荷点预测存在不同程度误差及难以反映电力需求不确定性的问题,提出一种基于极点对称模态分解(extreme-point symmetric mode decomposition, ESMD)-分散熵(dispersion entropy, DE)和改进乌鸦搜索算法(improved crow search algorithm, ICSA)优化核极限学习机的短期负荷区间预测模型。首先用ESMD将原始负荷时间序列分解为多个特征互异的子序列,降低了原始非平稳负荷序列对预测结果的影响,并计算各子序列的分散熵,将熵值相近的子序列重组为新序列以降低计算规模;其次,基于上下限估计法,利用ICSA算法对核极限学习机(kernel extreme learning machine, KELM)输出权值进行优化,得到最优预测区间上下限,并以此分别对各新序列进行区间预测;最后将预测结果叠加得到最终的预测区间。仿真结果表明,所提模型有效提高了负荷预测区间的质量,为电力系统决策工作提供有力支持。

Short-term load interval prediction based on ESMD-DE and ICSA-KELM

Aiming at the problem that the deterministic load point prediction has different degrees of error and it is difficult to reflect the uncertainty of power demand, A short-term load interval prediction model based on Extreme-point Symmetric Mode Decomposition (ESMD)-Dispersive Entropy (DE) and Improved Crow Search Algorithm (ICSA) Optimized Kernel Extreme Learning Machine (KELM) is proposed.Firstly, ESMD is used to decompose the original load time series into multiple sub-sequences with different characteristics, reduce the influence of non-stationary load sequences on prediction accuracy, calculate the dispersion entropy of each sub-sequence, and recombine the sub-sequences with similar entropy into new sequences. To reduce the scale of calculation;Secondly, based on the lower upper bound estimation method, the ICSA algorithm is used to optimize the KELM output weights to obtain the upper and lower bounds of the optimal prediction interval, and the interval prediction is performed separately for each new sequence.Finally, the prediction results are superimposed to obtain the final prediction interval. The simulation results show that the proposed model effectively improves the quality of the load forecast interval and provides strong support for power system decision-making.