差分进化最小二乘支持向量机法预测日用水量

为解决最小二乘支持向量机的参数确定问题,提出采用自适应差分进化最小二乘支持向量机法预测日用水量.引入改进粗糙集算法分析日用水量主要影响因素,利用自相关系数法确定序列的相关性,并将自适应差分进化算法(SADE)用于优化最小二乘支持向量机(LSSVM)的参数,建立了基于SADELSSVM的预测模型.结果表明,与传统差分进化算法(DE)和自适应遗传算法(SAGA)相比,SADE具有更快的最优个体搜索速度和群体进化速度,与基于SAGALSSVM和基于DELSSVM的模型相比,本文提出模型的预测能力更强.

Application of differential evolution and least squares support vector machine method in daily water demand prediction

To find the optimal parameters of least squares vector machine(LSSVM), the daily water demand forecasting method based on self-adaptive differential evolution(SADE) and LSSVM was proposed. The main influencing factors of daily water consumption were determined using improved rough set algorithm, and the correlation analysis on daily water consumption series was conducted. SADE was applied to optimize the parameters of LSSVM to build SADELSSVM-based forecasting model. The case study shows that compared with self-adaptive GA(SAGA) and differential evolution(DE), SADE has stronger global search ability and faster evolution speed, and the proposed model has better prediction performance than SAGALSSVM-based model and DELSSVM-based model.