经验模态分解和遗传程序设计算法在短期负荷预测中的应用

摘要

The forecasting technique is one of the key factors for improving the accuracy of load forecasting. Thus, the Empirical Mode Decomposition (EMD) and the Genetic Programming (GP) are applied to short-term load forecasting. In the specific forecasting process, the load data are decomposed by EMD, and each Intrinsic Mode Function (IMF) obtained is forecasted by GP according to time-sharing. Finally, the results forecasted are obtained by constructing the corresponding forecasting results of each IMF. The percentage error of the final result forecasted is less than 4％, which manifests that this forecasting scheme proposed in this paper can meet the forecast requirements in practical application and is feasible.%负荷预测精度的高低关键因素之一取决于预测技术.为此,提出将经验模态分解和遗传程序设计算法相结合用于电力系统的短期负荷预测.具体预测过程是对负荷样本进行经验模态分解,然后对分解后的各本征模态分量分别利用遗传程序设计进行分时预测,并通过对各本征模态分量的预测结果进行重构来得到最终预测结果.预测结果的误差基本都在4%范围内,说明此方法能够满足实际预测要求,具有一定可行性.