http://www.paper.edu.cn
- 1 -
1
Research on Forecasting Method of Urban Water Demand
Based on Fuzzy Theory
Liu Hongbo
1
，
2
，Deng Tegang
1
，Zhang Hongwei
1
1. School of Environmental Science and Technology, Tianjin University, Tianjin (300072)
2. Tianjin Polytechnic University, Tianjin (300160)
E-mail :hwzhang@tju.edu.cn
Abstract
Water demand forecasting is of great importance in urban water supply management. A large
number of researches have been done on water demand forecasting method, which requires easy
operation and high accuracy. To meet the requirements, combining advantages of fuzzy inference
system and neural network, a new forecasting method, adaptive neuro-fuzzy inference
system(ANFIS), based on fuzzy theory was presented. In ANFIS, fuzzy rules are obtained from the
data sample, and appropriate membership function parameters are gained through hybrid- learning
algorithm, which contributes to effective and quick forecasting. Along with the properties of
ANFIS, four input variables were brought up, which coupled all main impact factors together in the
model. In the procedure of modeling and forecasting, the software, Matlab, can assist the design of
ANFIS, which makes this method easy to master and operate for most users. Furthermore, a case
was studied, and the results indicated that the predicting precision of this method can satisfy the
engineering request.
Keywords:
water demand forecasting; fuzzy theory; adaptive neuro-fuzzy inference system; fuzzy
inference
1. Introduction
Our country's urbanization process is moving forward rapidly. By the end of 2006, the total
urban population is 577 million, in other word, the level of urbanization amounts to 43.9%, with
annual growth rate of 0.9 percentage during the years 1980 to 2006. And it is expected that in the
next 15 to 20 years, the level of urbanization will still keep an average growth rate of about 0.8
percentage [1]. Nevertheless, water as necessities of industrial and agricultural production and
peoples’ daily life, its shortage limits and affects the steady and sustainable development of cities
to a certain extent. Thus, many concerned management sectors are seeking to effectively manage
water supply system to serve residents better urban infrastructure. And an important thing is to
balance the supply and demand of water. This needs scientific water demand forecasting, which is
also key to urban planning. The degree of forecasting precision has direct impact on water supply
system’s reliability, security and economy. So a lot of researches have been done on water demand
forecasting methods. Water demand forecasting includes short-term forecasting and middle and
long-term forecasting with different purposes. Short-term forecasting is for online real time
simulation and optimization of water delivery. While middle and long-term forecasting are for
water resources planning and city design & planning.
At present there are kinds of water demand forecasting methods. Based on quantity of
information about causal relationship between impact factors and water demand, forecasting
methods can be divided into white-box model, grey-box model and black-box model. The above
three categories have their own typical forecasting methods, as follows, time series forecasting,
the grey forecasting model, Artificial Neural Network(ANN), respectively.
Time series forecasting is an important area of forecasting in which past observations of the
same variable are collected and analyzed to develop a model describing the underlying
1
Supported by NSFC（50278062，50578108），Specialized Research Fund for the Doctoral Program of Higher
Education（20050056016）.
剩余10页未读，继续阅读
资源评论
weixin_38686153
- 粉丝: 11
- 资源: 887
