摘要
在对天津市需水量现状进行调查的基础上,分析需水量与相关因素的变化规律,建立天津市需水量预测模型.应用粒子群优化算法(PSO)对神经网络权值进行优化,建立PSO-BP神经网络,应用于需水量预测模型的求解.将PSO-BP法与传统的BP神经网络法的计算结果进行对比,前者的预测平均相对误差比后者低500/.结果证明,该预测模型能够较好地拟合天津市需水量变化趋势,PSO-BP方法比BP方法具有更高的收敛速度和精度.
Based on the investigation of the present data of Tianjin, forecast model for municipal water resource demand in Tianjin was set up through analyzing water demand and its influencing factors. The power value of artificial neural network is optimized by particle swarm optimization（PSO） to set up PSO-BP network to find the solution between the model. The results between the PSO-BP and classical BP were compared, and the average prediction error was reduced by 5%. The example showed that PSO-BP network was more fit for urban water demand prediction in Tianjin and had higher forecasting precision than BP method.
出处
《天津大学学报》
EI
CAS
CSCD
北大核心
2007年第6期742-746,共5页
Journal of Tianjin University(Science and Technology)
基金
国家自然科学基金资助项目(50278062
50578108)
关键词
城市需水量
粒子群优化算法
人工神经网络
预测模型
urban water demand
particle swarm optimization
artificial neural network
prediction model
