Electr Eng
DOI 10.1007/s00202-016-0436-8
ORIGINAL PAPER
Optimization of soil moisture sensor placement for a PV-powered
drip irrigation system using a genetic algorithm and artificial
neural network
Mahir Dursun
1
· Semih Özden
2
Received: 5 January 2016 / Accepted: 12 September 2016
© Springer-Verlag Berlin Heidelberg 2016
Abstract The efficiency and installation costs of solar-
powered drip irrigation systems depend on not only the
efficiencies of the electrical motor, its driver, and the pump,
but also the efficient usage of irrigation water. In this study,
the initial installation costs and energy consumption of pho-
tovoltaic irrigation systems were decreased by obtaining the
soil moisture level as a reference for optimizing energy and
water consumption in a solar-powered drip irrigation system.
The data from 15 moisture sensors placed in the area cov-
ered by the system were collected by a central unit using
radio transmission. The soil moisture was estimated via an
artificial neural network with the data obtained for 6 m × 6m
micro-regions. Next, the locations of the moisture sensors in
the area were optimized using a genetic algorithm to provide
the optimum energy and water consumption in the system.
Subsequently, the drip irrigation was controlled using mois-
ture data from only five sensors located at the best points, as
determined by the genetic algorithm. The obtained experi-
mental results indicated that the moisture rate at the end of
the period of irrigation using the system developed was more
homogeneous than that of traditional irrigation systems for
each micro-region using only five soil moisture sensors in a
non-homogeneous area. Thus, daily energy and water con-
sumption were decreased by 32 %, while the moisture rate
in the soil was maintained within the desired range.
B Mahir Dursun
mdursun@gazi.edu.tr
Semih Özden
sozden@gazi.edu.tr
1
Department of Electrical-Electronics Engineering,
Faculty of Technology, Gazi University, 06500 Ankara,
Turkey
2
Department of Electrical and Energy, Technical Sciences
Vocational College, Gazi University, 06374 Ankara, Turkey
Keywords Artificial neural network · Genetic algorithm ·
Drip irrigation · Soil moisture · Photovoltaic irrigation
system
1 Introduction
The depletion of energy resources and the increasing scarcity
of clean water represent two of today’s most pressing global
problems. For this reason, it is extremely important to use
existing resources efficiently. The efficient use of available
water and energy resources is particularly crucial when one
considers that 75 % of clean water resources are used for agri-
cultural irrigation in countries such as Turkey, where large
portions of the land and economy are involved in agricultural
production [1]. The use of solar-powered drip irrigation sys-
tems has many advantages, such as the efficient use of both
energy and water resources in agricultural irrigation areas; as
a result, such systems have become increasingly widespread.
The most important factor in the spread of these systems is
their cost. Although prices are decreasing daily, solar cells
still constitute a serious additional cost for farmers [2, 3].
To increase the applicability of solar-powered drip irriga-
tion systems and provide efficient use of resources, costs must
be reduced as much as possible. Solar panels, batteries, and
pump motors constitute a large portion of the initial setup
costs of photovoltaic-based irrigation systems. This makes
it critical to select the appropriate settings for these devices.
The power ratings of the solar panel, battery and pump motor
hardware are determined by the amount of water required
by the planted product. In this case, sufficiency and high
efficiency in irrigation are important. In current practice, irri-
gation is applied to the land evenly, depending on the previous
experiences of farmers. However, the irrigation quantity is
influenced by the following factors: the crop type and its
123