摘要
提出了一种基于HHT的电力系统短期负荷预测模型。针对EMD分解电力负荷时存在模态混叠及对高频IMF预测不准确的问题,采用一阶差分算法对EMD分解进行改进,得到消除模态混叠后的一系列IMF分量及余项。通过对各分量的频谱计算和观察,提取出低频分量,并将其进行重构,各分量选取合适模型进行预测。由于IMF1主要为负荷的随机分量,对其考虑天气、节假日因素,并采用粒子群算法对组合权值进行优化。仿真结果表明此种方法具有较高的预测精度。
A short-term load forecasting model based on HHT is proposed.Due to the mode-mixing phenomenon when the power load data is decomposed by EMD and the problem of the IMF with high frequency is difficult to forecast,the first difference algorithm is used to improve EMD decomposing,then several IMFs and remainder without mode-mixing can be obtained.Through calculating and observing the spectrum of decomposed series,the low frequency IMFs are extracted and reconstructed,which can be forecasted with appropriate forecasting model.Since IMF1 is mainly as the random component of load,factors such as temperature and weekday are taken into consideration,and PSO is used to optimize the combination weights.Simulation results indicate that PSO has higher forecasting accuracy.
出处
《电力系统保护与控制》
EI
CSCD
北大核心
2011年第2期55-60,64,共7页
Power System Protection and Control
基金
教育部新世纪优秀人才支持项目(NECT-08-0825)
教育部霍英东青年教师基金资助项目(101060)
四川省杰出青年基金项目(07JQ0075)