Integrated Water Resources Management using Artificial Intelligence
Author: Devrishi Bharadwaj
“Water is the livelihood of our bodies, our economy, our nation
and our well-being.” –Stephen Johnson
Water is an indispensable part of life. Proper scientific management of water
resources is a must to meet the demands of the rising population and to
ensure its availability for our future generations. Hydrological processes have
significant influence on the ecology of earth. Development of water resources
can greatly benefit communities, ensure sustainable development and
significantly contribute to the economy of a region, especially at river basins
and coastal plains.
Management of water resources is a complex process because of great
variations in rainfall patterns; flow of rivers, weather and climate, and the
whole hydrological process is a dynamic phenomenon. It is also influenced by
socio-economic factors such as consumption levels, location of industries and
townships, festivals etc. With many parts of India facing severe water scarcity,
the alarm bell has rang for planning out efficient, scientific and a real time
water resources management system. It should take into account all the
factors of water resources management, to meet the rising demand of India’s
increasing population and economy but at the same time ensure no
disturbance in the ecological cycle. Water is life, so availability of freshwater is
a must for the continuity of the human civilization.
This article explains only the applications of ANN, GA and GIS in the theoretical
way, not the mathematical formulas, equations and computer programming
methods involved. Also this article specially focusses on use of AI in planning &
managing Water Distribution Systems.
Conventional mathematical and statistical models have failed to give optimal
solutions, predictions and simulations to aid in the management of water
resources. Although there has been a great amount of research being carried
out in India, on the branches of Artificial Intelligence (AI) such as Artificial
Neural Networks (ANN) and Genetic Algorithms (GA), but its application has
remained very limited. ANN and GA can be applied to solve, simulate and
predict almost any problem but this article focuses on the use of ANN and GA
with GIS on the management and development of water resources in India,
which will have revolutionary scientific, social and economic benefits for the
nation.
Artificial Neural Networks - The ANN is a computational approach to complex
problems inspired by the biological nervous system. It has adaptive learning
capabilities, can recognize patterns and learn from their interactions. Both
historical data and real time data can be integrated in ANN. ANN give the best
optimal solution, simulation or prediction of a given problem.
Genetic Algorithms- An algorithm is any procedure that takes in data and
modifies it according to a step-by-step set of instructions. GAs simulate the
logic of the Darwinian theory of Natural Selection to solve complex problems.
Holland gave the concept of GAs in 1970s. GAs can also be called self-learning
algorithms.
Geographic Information System- A geographic information system (GIS) is a
computer system for capturing, storing, checking, and displaying data related
to positions on Earth’s surface. By relating seemingly unrelated data, GIS can
help individuals and organizations better understand spatial patterns and
relationships.
Due to stochastic parameters and uncertainties in the water resources system,
GA and ANN will be helpful as they have adaptive learning capabilities. They
are useful for complex optimization problems, where the number of
parameters is large and analytical solutions are difficult to obtain. Integrated
use of surface and ground water is essential for optimum utilization of water
resources. In most models, the actual interaction between surface and
groundwater is often neglected. Conjunctive use of surface and groundwater is
important. ANN based simulation models can be used in GA based
optimization models to provide sector wise optimal water allocation. This data
can be presented in GIS for easy understanding of decision makers, and also
automated systems aided by human interaction can be developed. ANN has
greater advantage in case of assessing and simulation ground water resources
than software such as PMWIN (Processing MODFLOW for Windows). ANN &
GA can be used to optimize hydropower generation and at the same time
develop an efficient irrigation system from a multipurpose dam. There are a
number of important areas that have to be studied in detail to develop an
Integrated Water Resources Management System such as Water Distribution
Systems, Sewage Management, Irrigation, Natural & Artificial Floods, Dams &
Reservoirs, Ground & Surface Water, Rainfall & Climatic Factors, Local
Hydrology of a particular area, etc.
From a hydraulic point of view, water supply system is a complex system,
which needs to meet the demand continuously. Since a modification of a single
parameter like pipe diameter, water pressure, pipe layout can have an impact
on the entire system. Therefore a thorough estimation is required. The
situation in case of firefighting, low rainfall, demand during peak hours, pipe
leakage, damaged pumps etc. are important factors. These all parameters can
be taken care by ANN and GA. Water at a required pressure have to be
supplied continuously but water supplying authorities also want to save costs
and make the system more efficient. ANN and GA make harmonisation of
interests and requests possible.
GA can solve combinatorial optimisation problems, which cannot be solved by
conventional operational research methods. The GA provides results which are
within reasonable limits of deviation from the real situation, unlike the
conventional methods whose results do not match the actual situation. Also a
single GA can be used for a range of desired objectives, because of its multiobjective
function capability. Through the integrated system, pumping cost can
be reduced and ground water table variations can be controlled. In some
areas, pipes may not be necessary (which can be planned using GA) which will
reduce the construction and maintenance cost. Using GIS is effective to
determine maintenance works in water supply systems, which is essential to
ensure provide sufficient water to the citizens and to maintain longer lifeline of
the system. Water Distribution Systems are comprised of interconnected
sources such as pipes and hydraulic control elements (pumps, valves,
regulators, tanks etc). Using this integrated system, water distribution/supply
systems can be designed to deliver sufficient quantity of water to consumers at
required pressure, quality (safe) and in a continuous, cost effective and
sustainable manner. Therefore using ANN and GA based models coupled with
GIS is important for optimal design of a water distribution network to minimize
construction costs while ensuring adequate system performance under
specified design criteria.
Water Resources management and design problems often involve political,
social, and other subjective goals that cannot be represented mathematically.
But by coupling an ANN simulation model with a GA based optimisation
approach, a mathematically optimal solution may be identified. GAs can even
help in identification of important factors which are not clearly determined.
GAs are flexible for the ever changing dynamics of water resources
management. There is a developed methodology for a GA model calibration
with interactive evolution to incorporate un-modelled objectives (or objectives
that remain unknown to the policy makers) in the search procedure, which is
not possible in conventional methods.
Interactive evolution aims to utilize subjective responses from human users to
drive the evolutionary search. Human responses are given numerical ranks and
then used as fitness function for the GA. The computational limits posed by
fitness evaluation of GA have paved the way for development of
approximation frameworks, where by the process of ANN coupled evolution
and online training of an ANN (means the ANN is learning during the
operational process of the GA) to adaptively classify the trustworthiness of the
approximation model and enhance its performance.
Development of a an Integrated Water Resources System for India linking
multiple ANNs, GAs (sector & purpose wise) presenting it in a GIS or other
suitable software will not only help policy makers to make optimum and
sustainable use of the precious water resources, but will also lead to automatic
cost-effective water distribution systems and aid in the overall socio-economic
growth of the nation.
(The author is the Chairman of Bharat Innovates Universals)
Published by Department of Research & Publications, Bharat Innovates
Universals
Read our articles at www.biutransformingindia.blogspot.com
Artificial Neural Networks - The ANN is a computational approach to complex
problems inspired by the biological nervous system. It has adaptive learning
capabilities, can recognize patterns and learn from their interactions. Both
historical data and real time data can be integrated in ANN. ANN give the best
optimal solution, simulation or prediction of a given problem.
Genetic Algorithms- An algorithm is any procedure that takes in data and
modifies it according to a step-by-step set of instructions. GAs simulate the
logic of the Darwinian theory of Natural Selection to solve complex problems.
Holland gave the concept of GAs in 1970s. GAs can also be called self-learning
algorithms.
Geographic Information System- A geographic information system (GIS) is a
computer system for capturing, storing, checking, and displaying data related
to positions on Earth’s surface. By relating seemingly unrelated data, GIS can
help individuals and organizations better understand spatial patterns and
relationships.
Due to stochastic parameters and uncertainties in the water resources system,
GA and ANN will be helpful as they have adaptive learning capabilities. They
are useful for complex optimization problems, where the number of
parameters is large and analytical solutions are difficult to obtain. Integrated
use of surface and ground water is essential for optimum utilization of water
resources. In most models, the actual interaction between surface and
groundwater is often neglected. Conjunctive use of surface and groundwater is
important. ANN based simulation models can be used in GA based
optimization models to provide sector wise optimal water allocation. This data
can be presented in GIS for easy understanding of decision makers, and also
automated systems aided by human interaction can be developed. ANN has
greater advantage in case of assessing and simulation ground water resources
than software such as PMWIN (Processing MODFLOW for Windows). ANN &
GA can be used to optimize hydropower generation and at the same time
develop an efficient irrigation system from a multipurpose dam. There are a
number of important areas that have to be studied in detail to develop an
Integrated Water Resources Management System such as Water Distribution
Systems, Sewage Management, Irrigation, Natural & Artificial Floods, Dams &
Reservoirs, Ground & Surface Water, Rainfall & Climatic Factors, Local
Hydrology of a particular area, etc.
From a hydraulic point of view, water supply system is a complex system,
which needs to meet the demand continuously. Since a modification of a single
parameter like pipe diameter, water pressure, pipe layout can have an impact
on the entire system. Therefore a thorough estimation is required. The
situation in case of firefighting, low rainfall, demand during peak hours, pipe
leakage, damaged pumps etc. are important factors. These all parameters can
be taken care by ANN and GA. Water at a required pressure have to be
supplied continuously but water supplying authorities also want to save costs
and make the system more efficient. ANN and GA make harmonisation of
interests and requests possible.
GA can solve combinatorial optimisation problems, which cannot be solved by
conventional operational research methods. The GA provides results which are
within reasonable limits of deviation from the real situation, unlike the
conventional methods whose results do not match the actual situation. Also a
single GA can be used for a range of desired objectives, because of its multiobjective
function capability. Through the integrated system, pumping cost can
be reduced and ground water table variations can be controlled. In some
areas, pipes may not be necessary (which can be planned using GA) which will
reduce the construction and maintenance cost. Using GIS is effective to
determine maintenance works in water supply systems, which is essential to
ensure provide sufficient water to the citizens and to maintain longer lifeline of
the system. Water Distribution Systems are comprised of interconnected
sources such as pipes and hydraulic control elements (pumps, valves,
regulators, tanks etc). Using this integrated system, water distribution/supply
systems can be designed to deliver sufficient quantity of water to consumers at
required pressure, quality (safe) and in a continuous, cost effective and
sustainable manner. Therefore using ANN and GA based models coupled with
GIS is important for optimal design of a water distribution network to minimize
construction costs while ensuring adequate system performance under
specified design criteria.
Water Resources management and design problems often involve political,
social, and other subjective goals that cannot be represented mathematically.
But by coupling an ANN simulation model with a GA based optimisation
approach, a mathematically optimal solution may be identified. GAs can even
help in identification of important factors which are not clearly determined.
GAs are flexible for the ever changing dynamics of water resources
management. There is a developed methodology for a GA model calibration
with interactive evolution to incorporate un-modelled objectives (or objectives
that remain unknown to the policy makers) in the search procedure, which is
not possible in conventional methods.
Interactive evolution aims to utilize subjective responses from human users to
drive the evolutionary search. Human responses are given numerical ranks and
then used as fitness function for the GA. The computational limits posed by
fitness evaluation of GA have paved the way for development of
approximation frameworks, where by the process of ANN coupled evolution
and online training of an ANN (means the ANN is learning during the
operational process of the GA) to adaptively classify the trustworthiness of the
approximation model and enhance its performance.
Development of a an Integrated Water Resources System for India linking
multiple ANNs, GAs (sector & purpose wise) presenting it in a GIS or other
suitable software will not only help policy makers to make optimum and
sustainable use of the precious water resources, but will also lead to automatic
cost-effective water distribution systems and aid in the overall socio-economic
growth of the nation.
(The author is the Chairman of Bharat Innovates Universals)
Published by Department of Research & Publications, Bharat Innovates
Universals
Read our articles at www.biutransformingindia.blogspot.com
Yorumlarınızı Bizimle Paylaşın
Sadece üyelerimiz yorum yapabilir, hemen ücretsiz üye olmak için Tıklayın