摘要
城市供水时用水量预测精度对城市供水系统具有重要影响.传统的反向传播(back-propaganda,BP)神经网络预测方法容易陷入局部解,并且需要大量的训练数据.人工鱼群算法具有较优的全局收敛能力及较快的寻优速度.为此,利用人工鱼群算法对BP神经网络的初始权值和阈值进行优化,建立了一种新的时用水量预测模型.将该模型应用到华北某市时用水量的预测中,预测结果表明人工鱼群神经网络算法的均方差比BP神经网络算法的均方差小5%.实例证明,人工鱼群神经网络比BP神经网络的预测精度更高,收敛速度更快.人工鱼群神经网络算法可用于短期水量预测.
The forecast precision of city water consumption per hour has great effect on the city water supply system. The traditional forecast method of back-propagation（BP）neural network tends to offer local values and requires a lot of data training. The artificial fish-swarm algorithm（AFSA）has better global convergence ability and higher optimiza-tion speed. AFSA was adopted to optimize the initial setting weights and thresholds of BP neural network. Then anew forecast model of water consumption per hour was built and was applied to forecast the water consumption per hour of a city in North China. Results show that the mean square error of the artificial fish-swarm neural network algorithm is lower than that of BP neural network algorithm by 5%. It has been verified by instances that the artificial fish-swarm neural network has better forecast precision and higher convergence speed than BP neural network. Artificial fish-swarm neural network algorithm can be used to forecast the short-term water consumption.
出处
《天津大学学报（自然科学与工程技术版）》
EI
CAS
CSCD
北大核心
2015年第4期373-378,共6页
Journal of Tianjin University：Science and Technology
基金
国家社会科学基金重点资助项目(13AZD011)
关键词
城市用水
人工鱼群算法
水量预测
city water consumption
artificial fish-swarm algorithm
water consumption forecast
