鲶鱼粒子群算法优化支持向量机的短期负荷预测

Short-time load prediction based on support vector machine optimized by catfish particle swarm optimization algorithm

为了准确、有效地预测短期负荷,提出了一种鲶鱼粒子群算法优化支持向量机的短期负荷预测模型(BFPSO-SVM)。基于混沌理论对短期负荷时间序列进行相空间重构;将支持向量机参数的组合看作一个粒子位置串,通过粒子间互作找到最优支持向量机参数,并引入"鲶鱼效应",克服基本粒子群算法的缺点;根据最优参数建立短期负荷预测模型,并对模型性能进行仿真测试。仿真结果表明,相对于其他预测模型,BFPSO-SVM不仅加快了支持向量机参数寻优速度,而且提高了短期负荷预测精度,更适用于短期负荷预测的需要。

In order to accurately, effectively predict short-term load, this paper proposes a short-term load prediction （BFPSO-SVM） based on support vector machine optimized by catfish particle swarm optimization algorithm. The short-term load time series are reconstructed based on chaos theory, and then the support vector machine SVM parameters are taken as a particle location string, and catfish effect is introduced to overcome the shortcomings of particle swarm algorithm to find the optimal parameters of sup- port vector machine through the particle interactions, short-term load forecasting model is built according to the optimum para- meters and the model performance is tested by simulation experiment. The simulation results show that, compared with other prediction models, BFPSO-SVM accelerates the parameters optimizing speed of support vector machine and improves the pre- diction precision of short term load, and it is more suitable for short-term load prediction needs.