Information Technology Reference
In-Depth Information
pesticide utilization ratio is very low [2]. Therefore, there is an urgent need to
research and apply the variable spraying technology to agricultural production.
The so-called variable spraying technology refers to the technology of adjust-
ing the injection quantity with the change of the target information. Various
variable spraying systems have been designed by scholars both at home and
abroad for different demands. In [3], a multi-nozzle air blast spray was designed
with three electromagnetic valves and three ultrasonic sensors, which implements
the variable spraying according to the area change of grape leaves, but it lacks the
ability of disease information acquisition. A fuzzy method was proposed in [4], for
the lack of learning and self-adaption, the method wasnt applied to greenhouse
with harsh environment. An adaptive neural decision method was put forward
in [5], setting weed area and motion speed as inputs quantity and spray vol-
ume as output, and using the measured data to train the generated initial fuzzy
system, however, the changes of the speed will lead to the changes of spraying
system pressure, it is dicult to control the spraying performance. It is hard for
the existing decision-making methods to get a desirable effect in the complicated
uncertain greenhouse. Consequently, in this paper, an intelligent decision-making
method combining artificial neural network with fuzzy control was put forward.
An intelligent decision-making system of off-line training fuzzy neural network
was constructed, and the spraying quantity was calculated according to the ob-
tained information of target features. It offers a new decision-making method
for the variable spraying researches of greenhouse mobile robots.
2 System Compositions
Greenhouse is a complex system. There are many factors influencing pesticide
spraying, such as crop area, distance, the level of plant diseases and insect pests,
temperature, humidity and so on. Theoretically, the more impact factors as
inputs of the decision maker, the better effect of decision will be got. However,
with the inputs increasing, the structure of decision will be more complex, and
the fuzzy rule will be made more dicultly. Among the numerous factors, crop
area, distance and the level of plant diseases and insect pests are the most
influential factors. According to the cultivation method and growth features
of band crop in modern greenhouse, combined with the spraying principle of
mobile robots, the spraying quantity mainly decided by the crop area, distance
and damage degree to implement the rapid and effective decision-making.
The variable spraying system of mobile robot is consisted of an image collector,
an ultrasonic sensor, a decision maker, a controller, an electromagnetic valve, a
flow sensor, a sprayer, etc. The system structure is shown in Fig. 1.
The spraying quantity is calculated by the decision maker according to the
got target information, and is compared with the actual flow measured by the
flow sensor. According to results of the comparison, the PWM (Pulse Width
Modulation) [6] duty cycle is adjusted real-timely to drive the electromagnetic
valve and realize the variable spraying.
Search WWH ::
Custom Search