城镇居民生活用水量预测及用水需求量影响因素分析

Prediction of Domestic Water Consumption of Urban Residents and Analysis of Influencing Factors of Water Demand

揭示城镇居民生活用水量的变化规律及其影响因素,对于改善城市供水调度和控制居民用水量的增长具有重要意义。为克服BP神经网络学习收敛速度慢、易陷入局部极小值、网络结构难以确定的缺点,提出将遗传算法(GA)与BP神经网络组合形成GA-BP混合训练网络,应用于居民生活用水量预测当中,预测结果显示,该方法与单一BP神经网络相比,具有更好的预测精度和适用性。其次,通过修正后的模型分析水价、人均可支配收入及人均住房面积三个用水需求影响因素,估计其价格弹性、收入弹性和人均住房需求弹性。模型估计结果表明,水价的上调,会抑制居民用水量的增长;居民用水量明显随人均可支配收入的增加而提升;人均住房面积与居民用水量之间的关系未得到较好体现。

It is of particular importance to reveal the variation law of urban residents’domestic water consumption and its influencing factors for improving urban water supply dispatching and controlling the growth of residents’water consumption.In order to overcome the shortcomings of BP neural network like converges slowly,quickly falling into a local minimum and challenging to determine the network structure,a GA-BP hybrid algorithm model of combining genetic algorithm(GA)and BP neural network was proposed,which is applied to the prediction of domestic water consumption.The prediction results show that the GA-BP algorithm has better accuracy and applicability than the single BP neural network.Secondly,three influencing factors of water demand including water price,per capita disposable income and per capita housing area analyzed using the modified model and the price elasticity,income elasticity and per capita housing demand elasticity were estimated.The results show that the increase of water price will inhibit the growth of residents’water consumption;The relationship between per capita housing area and residents’water consumption is not significant.