改进粒子群算法优化支持向量机的短期负荷预测

针对支持向量回归机在预测建模中的参数选取问题,提出一种基于混沌自适应策略的粒子群优化支持向量回归机参数的方法.采用混沌映射算法和聚合度自适应判断策略,增强种群的全局寻优性能,提升粒子的多样性,从而避免种群过早收敛.充分考虑天气、节假日、居民消费等因素的影响,提出一种改进的支持向量回归机预测模型并与粒子群算法的支持向量回归机模型进行对比分析.分析结果表明,该预测模型可将预测的均方根误差降低约40％,绝对值误差降低约42％,相对误差降低约46％,仿真结果验证了所提方法优化了支持向量回归机参数,改善了预测效果.

Power Short-Term Load Prediction Based on Support Vector Machine Optimized by Improved Particle Swarm Optimization Algorithm

For the parameter selection of support vector regression (SVR) in predicting modeling,a particle swarm optimization(PSO) method based on chaos adaptive strategy is proposed to optimize the parameter of SVR.The chaotic mapping algorithm and adaptive degree of aggregation judgment strategy are used to enhance the global optimization performance of the population and the diversity of particles,so the premature convergence of the population is avoided.Considering the influence of weather,holiday factors and local consumption on the prediction result,a SVR prediction model based on the proposed PSO method is developed,and compared with the SVR model of particle swarm.The analysis results show that the prediction model can reduce the estimated root mean square error by about 40%,the absolute value error by about 42%,and the relative error by about 46%.The simulation results verify that the proposed method optimized SVR parameters and improved the prediction effect.