基于改进PSO算法的Logistic模型在饱和负荷预测中的应用

Application of the logistic model based on improved particle swarm optimization algorithm in saturated load forecasting

将改进的粒子群(PSO)算法应用到饱和电力负荷预测中,通过与Logistic时间序列预测模型相结合,对Logistic曲线函数进行优化参数求解。建立了基于该优化算法的Logistic时间序列饱和负荷预测模型,利用某地区电网历史数据进行Logistic时间序列分析。仿真结果表明,该改进算法收敛速度快,全局寻优能力强,克服了传统PSO算法局部搜索能力较差、容易陷入局部最优的缺点。利用它得到的Logistic拟合曲线,相对于传统PSO算法和Marquardt迭代算法的拟合结果,精度有明显的提高,说明该模型能够很好地反映电力负荷整体变化趋势。另外,运用该模型和人均用电量法分别对某地区电网饱和全社会用电量进行预测,结果显示两者预测结果较为接近,而人均用电量法在饱和电力负荷预测中运用已较为成熟,因此可以证明该模型应用到饱和电力负荷预测中是可行的。

Improved PSO algorithm is applied to the saturated power load forecasting, combining with logistic time series forecasting model, used for optimizing the parameters in Logistie curve function. Lo- gistic saturated load time series forecasting model is established based on the optimization algorithm. The logistic time series analysis is carried out on the basis of the historical data of a region power grid. Simulation re- suits show that the improved algorithm has faster convergence speed and stronger global optimization abihty, overcoming the traditional PSO algo- rithm＇ s drawbacks of poor local search ability and being easy to fall into locally optimal point. The Logistic fitting curve based on the improved al- gorithm, compared with the fitting result of traditional PSO .algorithm and Marquardt iteration algorithm, is more accurate obviously. It shows that the model can reflect the whole developing trend of power load well. In addition, the model and per-person electricity consumption method is ap- plied to predict saturated total electricity consumption, having achieved adjacent prediction result, and the application of per-person electricity consumption method in saturated power load forecasting has been rela- tively mature. So it can prove that it is feasible that the improved algo- rithm is applied to saturated toad forecasting.