基于自适应粒子群算法优化支持向量机的负荷预测  

Load forecasting based on adaptive particle swarm optimization algorithm optimizing support vector machine

作　　者：廖庆陵 窦震海[1] 孙锴 朱亚玲 LIAO Qingling;DOU Zhenhai;SUN Kai;ZHU Yaling(Shandong University of Technology,Zibo 255000,China)

机构地区：[1]山东理工大学,山东淄博255000

出　　处：《现代电子技术》2022年第3期125-129,共5页Modern Electronics Technique

基　　金：国家重点研发计划(2017YFB092800);淄博市重点研发计划资助(2019ZBXC498)。

摘　　要：负荷预测是电力系统调度运行的重要基础数据,短期负荷预测的样本数据既有波动性也有随机性。群体优化算法尤其是粒子群算法在负荷预测中运用非常广泛,但常规粒子群算法的惯性参数一般是固定不变的,导致后期搜索效率下降。文中采用改进的自适应粒子群算法提高搜索效率:首先用混沌初始化替代原来的随机初始化,避免了初始种群分布不均;再根据每次迭代适应度的变化更新惯性因子,可以解决后期寻优速度下降的问题;通过差分变异将适应度较差的粒子进行变异,提高了较差个体更新效率;最后利用改进后的自适应粒子群算法优化支持向量机的关键参数c和g,并进行短期负荷预测。通过测试得到改进后的自适应粒子群算法具有较好的优化效果,并且由自适应粒子群算法优化的支持向量机模型具有更好的预测效果。The load forecasting refers to some important basic data for the dispatch and operation of power system.The sample data for short⁃term load forecasting are of volatility and randomness.Swarm optimization algorithms,especially particle swarm optimization(PSO)algorithms,are widely used in modern load forecasting.However,the inertia parameters of the conventional PSO algorithms are generally fixed,which leads to a decreased search efficiency in the later stage.Therefore,an improved adaptive particle swarm optimization(APSO)algorithm is used to improve the search efficiency.The chaotic initialization is used to replace the original random initialization to avoid uneven distribution of the initial population.The inertia factors are updated according to the fitness change of each iteration,which can get rid of the slowdown of the later iteration.The particles with poor fitness are mutated by differential mutation to improve the update efficiency of the poorer population.The improved APSO algorithm is used to optimize the key parameters c and g of support vector machine(SVM)and perform short⁃term load forecasting.After testing,the improved APSO algorithm has a better optimization effect,and the SVM model optimized by the improved APSO algorithm has a better prediction effect.

关 键 词：混沌初始化 群体算法 惯性因子 自适应粒子群算法 差分变异 支持向量机 负荷预测 

分 类 号：TN911.1-34[电子电信—通信与信息系统] TM715[电子电信—信息与通信工程]