为提高电网短期负荷预测的精度,对以往学者基于相似日和最小二乘支持向量机(LS-SVM)短期负荷预测方法进行改进,形成一种改进的基于相似日和细菌趋化改进粒子群算法优化最小二乘支持向量机(least squares support vector machine based on improvedparticle swarm optimization for bacterial chemotaxis,PSOBC-LSSVM)的预测模型;克服了标准粒子群算法容易早熟收敛和陷入局部最优的问题,并充分考虑短期负荷的连续性与周期性对选取相似日造成的影响,将二者结合到一起综合考虑,利用改进的粒子群得到二者的最佳匹配值,并将其融合到时间距离这一因子当中;算例表明该方法预测精度较更高,可行且有效.%In order to improve the accuracy of short-term load forecasting,this paper improves the short-term load forecasting method based on similar daily and least squares support vector machine (LS-SVM),and forms a new prediction model based on similarity and bacterial chemotaxis improved particle swarm optimization for least squares support vector machines (PSOBC-LSSVM).This paper overcomes the problem that the standard particle swarm algorithm is easy to prematurely converge and fall into the local optimum.Taking full account of the impact of short-term load continuity and periodicity on the selection of similar days,and using the improved particle swarm,the best matching value of the two is obtained and fused into the factor of date distance.
展开▼
