摘要
城市需水预测是城市水资源规划、管理、决策的重要依据。为提升需水预测的准确性,提出一种基于最小二乘支持向量机(LSSVM)的非线性预测模型。针对灰狼优化算法(GWO)寻优过程易陷入局部最优的问题,采用一般性反向学习与非线性控制参数对其进行改进,以提升全局优化性能。改进后的GWO算法更适合应用于非线性模型参数的拟合,因此将其引入LSSVM需水预测模型,用来拟合模型的超参数。采用上海市近年来城市供水数据来检验模型,仿真结果表明提出的城市需水预测模型平均相对误差为0.78%,最大相对误差为1.37%,具有良好的泛化能力,可作为需水预测的一种有效方法。
Urban water demand prediction is an important basis for urban water resources planning, management and decision-making. In order to improve the accuracy of water demand prediction, a nonlinear prediction model based on least squares support vector machine(LSSVM) was proposed. Aiming at the issue that the optimization process of Grey Wolf Optimization Algorithm(GWO) was easy to fall into local optimization, general opposition-based learning and nonlinear control parameters were used to improve the global optimization performance. The improved GWO algorithm was more suitable for the fitting of nonlinear model parameters. Therefore, IGWO was introduced into LSSVM water demand prediction model to fit the super parameters of the model. The model was tested by using the urban water supply data of Shanghai in recent years. The simulation results show that the average relative error of the urban water demand prediction model proposed in this paper is 0.78%, and the maximum relative error is 1.37%. It has good generalization ability and can be used as an effective method of water demand prediction.
作者
陈永政
CHEN Yongzheng(School of Applied Technology,Chongqing College of Finance and Economics,Chongqing 402160,China)
出处
《人民黄河》
CAS
北大核心
2023年第2期97-100,126,共5页
Yellow River
基金
重庆市教育委员会科学技术研究项目(KJQN202101906)。
关键词
水资源
灰狼优化算法
城市需水预测
最小二乘支持向量机
water resources
Gray Wolf Optimization Algorithm
urban water demand prediction
least squares support vector machine
