Design of the Feeding System Automatic Koi Fish Based on Internet

of Things Using the Fuzzy Logic Controller Method

M. Udin Harun Al Rasyid, Nur Rosyid Mubtadai and Aditya Yogi Dwi Nugraha

Informatic Engineering, Politeknik Elektronika Negeri Surabaya (PENS), Sukolilo, Surabaya, Indonesia

Keywords: Automatic Feeder, Internet of Things, Fuzzy.

Abstract: Feeding is important in fish farming. However, in general, feeding is still done manually, which is oriented

to human resources. This method has drawbacks that also affect fish growth, such as scheduling errors and

uncontrolled feeding doses that can cause overfeeding. This research aims to create a system for feeding and

scheduling feeds and checking feed remotely using Internet of Things (IoT) technology which is equipped

with several types of sensors, namely temperature sensors, humidity sensors, and rain sensors that function to

monitor weather conditions around the pond as input. fuzzy to determine the amount of feed issued, while the

ultrasonic sensor functions to monitor the availability of feed in the container. The sensor takes the data and

then it is received by the Arduino Uno microcontroller. Furthermore, the data will be sent to the ESP8266 and

there is data processing with the fuzzy method as an output for how long the servo motor is open. The data

that is ready will be passed to the database server (Firebase) using the ESP module with a Wi-fi network, then

from the Firebase the data will be passed to the Android application via the API. That way users can see the

results of monitoring feed conditions, automatic feeding scheduling, monitoring the weather around the pond

in real time through an Android-based application. Feeding monitoring trials and application functionality

have been successfully carried out with scenarios testing applications, devices, and sensors that are connected

to each other that are attached to the feed container.

1 INTRODUCTION

Feeding is important in fish farming. However, in

general, feeding is still done manually, which is

oriented to human resources. This method has

drawbacks that also affect fish growth, such as

scheduling errors and uncontrolled feed dosages.

Feeding koi fish is done 3-4 times a day with the right

dose and time. Feeding that is too frequent and

excessive will affect the health of the fish, because the

leftover food will mix with feces so that it becomes

ammonia and decomposes into nitrite which is

harmful to fish health.

In addition to cultivation ponds, small koi ponds

owned by many homeowners are often left to go out

of town or are not at home, this is also a problem if

the fish are not fed at the right time or even not fed at

all. Outside temperature conditions greatly affect the

amount of feed given to fish, if during the rainy

season the dose of feed given is not as much as during

the dry season or when it is not raining, the

temperature also affects feeding (Rasyid, 2021).

The problem with koi fish farmers today is

feeding fish in ponds. In traditional koi fish

cultivators, human error often occurs with feeding

hours and feed dosages in koi ponds, especially for

small ponds if the homeowner is not at home for a

long time. This system causes farmers to be unable to

monitor feed availability and check feeding at any

time or in real time in koi ponds. So that the

management of koi pond feed with this system is still

not optimal.

To solve the above mentioned problem, this

research proposes feeding system automatic for Koi

fish based on internet of things (IoT) using the fuzzy

logic controller method. The development of feed

management is needed to increase automation,

intelligence, productivity, and expand the aquaculture

industry.

Several studies have been carried out to solve the

above-mentioned problems to build an automatic feed

system. Sousa et al. proposed an integrated IoT

platform for aquaculture environmental monitoring

and environmental data collection. A mobile

unmanned surface vehicle and buoy equipped with

Al Rasyid, M., Mubtadai, N. and Nugraha, A.

Design of the Feeding System Automatic Koi Fish Based on Internet of Things Using the Fuzzy Logic Controller Method.

DOI: 10.5220/0011957300003575

In Proceedings of the 5th International Conference on Applied Science and Technology on Engineering Science (iCAST-ES 2022), pages 921-930

ISBN: 978-989-758-619-4; ISSN: 2975-8246

921

environmental sensors collects data on water

conditions (aquatic), and sends it to a data centre for

further data storage and processing (Sousa, 2019).

Sun et al. proposed an integrated water quality

monitoring system with GIS and IoT technology. The

system performs remote monitoring, management

and control of water temperature, dissolved oxygen,

pH, and water level. The system performs packet loss

rate analysis using a Wi-Fi network and analyses

dissolved oxygen conditions (Sun, 2019).

Lee et al. proposed an IoT-based urban aquaponic

system. The system consists of a fish tank,

hydroponic tank, IoT monitor (pH), and a water test

kit (pH, NO2, NO3). Case study using hydroponics

plant lettuces and breed goldfish (Lee, 2019). Ismail

et al. proposed a model for direct measurement and

IoT-based fishpond water parameter monitoring

system. The system consists of a Raspberry Pi

microcontroller, DO sensor, temperature sensor, pH

sensor. Data from the sensor is sent to the

microcontroller, and then sent to the database server

(Ismail, 2020).

Uddin et al. proposed a real-time freshwater

shrimp monitoring system. The system detects the

sensor values for temperature, pH, dissolved oxygen,

salinity level, and turbidity. The system will give an

alert if it is out of the range value that has been set.

The system analyses the size, weight, and percentage

of live shrimp (Uddin, 2020). Ouyang et al. proposed

a monitoring framework for aquaculture farms. The

framework consists of a robotic sensing platform

equipped with water quality sensors, an automatic

charging system and sensor cleaning, a data

processing system using machine learning, and an

aquaculture farm control centre (Ouyang, 2021).

2 SYSTEM DESIGN

Figure 1 shows an overview of the proposed system

design. In the design of this system, the data to be

processed will be obtained directly from the data on

the feed container in the fish pond. Where there are

three parts contained in the design of this system,

namely, block input, process, and output.

In the input block section there are users who can

monitor the status of feed conditions and feeding

schedules in real time with the Android application.

And also, users can receive notifications of feeding

schedules.

In the process block, there is an Arduino Uno as

the sensor control centre. The sensor used is an

ultrasonic sensor, which functions to measure the

availability of feed. The DHT sensor and Rain

detector for monitoring weather conditions as an

output of the amount of feed measured using the

fuzzy method. NodeMCU will receive the data value

sent from Arudino Uno. The data values are sent by

nodeMCU to Firebase.

At the output, nodeMCU gives commands to the

servo motor to move the feed container valve to open

it according to the specified time and the amount of

feed depends on the fuzzy calculation.

After the data is processed, the data will be sent

and stored on the Firebase real-time database server.

The android application will display the data in the

database in real-time to the application user via

internet communication. The main features of the

Android application as data visualization are

monitoring feeder, control schedule, and monitoring

weather.

Figure 1: Design of an automatic feeder system in a koi

fishpond.

Figure 2: Design of an automatic feeder system in a koi

fishpond.

Fuzzy logic control is applied to the servo motor.

The function of this control is to control the length of

time the servo motor opens the valve. The input of the

fuzzy logic control system is data from temperature

and humidity sensors and rain sensors. Both sensors

are used to determine the outdoor weather conditions

in the koi fish pond. The final desired result from this

fuzzy logic control application is that fish feed can be

distributed on a scheduled basis and according to the

iCAST-ES 2022 - International Conference on Applied Science and Technology on Engineering Science

922

expected dose. The fuzzy logic control block diagram

can be seen in figure 2.

This process aims to create a membership function

from real numbers. The output is a linguistic variable

that distinguishes each condition based on the range

value.

a) Temperature Variable

Air temperature will be grouped into three

different variables. The variables used include cold,

normal, and hot. Then the values and variables are

described in the fuzzy set of air temperature into the

membership function in figure 3.

Figure 3: Temperature member.

For the temperature membership function as

follows:



















(1)



































(2)



















(3)

b) Rain Variable

Rain Status will be grouped into two different

variables. The variables used include rain and no rain.

Then the values and variables are described in the

fuzzy set of rain status into the membership function

in figure 4.

Figure 4: Rain member.

For the rain membership function as follows:

























(4)



















(5)

Fuzzy Rule

R1) IF Temperature = Cold AND Rain Status = Rain

THEN Valve Condition = Fast

R2) IF Temperature = Cold AND Raining Status =

No Rain THEN Valve Condition = Medium

R3) IF Temperature = Normal AND Rain Status =

Rain THEN Valve Condition = Fast

R4) IF Temperature = Normal AND Rain Status = No

Rain THEN Valve Condition = Long

R5) IF Temperature = Hot AND Rain Status = Rain

THEN Valve Condition = Medium

R6) IF Temperature = Hot AND Raining Status = No

Rain THEN Valve Condition = Old

Defuzzification

Defuzzification in the Sugeno method is to calculate

the centre of single-ton or the centre point of the crisp

value using the weight average method which is

described in the following formula.



    

     



(6)

Description:

Z = fuzzy calculation result

an = value for the nth combination (singleton value)

zn = fuzzy membership rule value for the nth

combination

Design of the Feeding System Automatic Koi Fish Based on Internet of Things Using the Fuzzy Logic Controller Method

923

The value of zn is obtained from the implication

function that uses the minimum function, namely by

looking for the i-th rule and can be expressed by the

following equation:











 

















(7)

Description:





= minimum value of fuzzy sets A and B in the i-th

rule







= degree of member x of the fuzzy set A in the

i-th rule





= degree of member x of the fuzzy set B in the

i-th rule

So, the value of the servo motor will open

according to the z value obtained and according to the

following z function.

















(8)

Where if it's fast then the servo motor opens for 2

seconds, medium 4 seconds, and long 6 seconds.

2.1 Hardware

The components are assembled on a breadboard with

the aim of making the components neat so that there

are no many wires and no need for soldering. Arduino

as the main microcontroller which is connected to

each DHT sensor, ultrasonic, and rain sensor. While

the NodeMCU is connected to the servo motor and

RTC. Each component is connected by a jumper

cable, for the voltage to the sensor is taken from the

5v pin available on the Arduino Uno, while for the

servo and rtc motors the voltage is taken from the 3.3v

pin available on the NodeMCU. The results design is

as in figure 5.

The detailed sections of each piece of hardware

are described as follows:

Main Component

The detailed sections of each piece of hardware are

described as follows:

a. Arduino Uno

Arduino Uno using the ATmega328 chip

functions to control sensors including dht11

sensors, rain sensors and ultrasonic sensors. This

device requires 5v power.

Figure 5: Design hardware.

b. Nodemcu v3 ESP8266

The NodeMCU with a power of 5v serves as a

liaison between the hardware and the cloud server

or database that is connected via Wi-Fi technology

using the internet network.

c. DHT11 sensor(Temperature and humidity sensor)

The DHT11 sensor is a temperature and humidity

measurement sensor that is directly related to the

Arduino Uno on the digital pin (pin 2) with 5v

input. °C and humidity from 20% to 90% with an

accuracy of ±1 °C and ±1%.

d. Y-11 Rain Sensor

This Rain Sensor or Rain Sensor uses a 5v voltage

which is directly related to the Arduino Uno on the

analog pin, namely pin A0.

e. Ultrasonic Sensor

This ultrasonic sensor serves to measure the

distance from the sensor to the fish feed, so that

it can detect the availability of existing feed. This

sensor uses a 5v voltage which is directly related

to Arduino Uno on digital pins (pin 11 and pin 12).

f. Servo Motor

Servo motor as an automatic feed valve controller

which is ordered by the NodeMCU to a certain

angle according to the conditions and

opening time from the fuzzy calculation results.

The servo motor is connected to the NodeMCU on

digital pin D4 with 3.3v input.

g. RTC DS3231

This DS3231 RTC is used for timing controllers.

Directly connected to NodeMCU on digital pins

D1 and D2 with 3.3v input.

Supporting Component

The following components function to support the

main components so that the system and hardware

design can work optimally:

a. Jumper Cable

iCAST-ES 2022 - International Conference on Applied Science and Technology on Engineering Science

924

Jumper cable serves as a connector between

the hardware. In this study, 3 types of jumper

cables were used, including: male to male, female

to female, and male to female.

b. Micro USB Cable

This cable serves to connect the NodeMCU

with the power adapter so that the NodeMCU can

be turned on.

c. Breadboard

Breadboard is used to make temporary

electronic circuits for the purpose of testing or

prototyping without having to solder. This avoids

components from being damaged if incorrectly

soldered.

Hardware Assembly

The components that have been previously designed

are then assembled into one according to the

electronic component design. The results can be seen

in figure 6.

The components are assembled on a breadboard

with the aim of making the components neat so there

are no many wires and no soldering required. Arduino

as the main microcontroller which is connected to

each sensor. While the NodeMCU is connected to the

servo motor and RTC. Each component is connected

by a jumper cable, for the voltage to the sensor is

taken from the 5v pin available on the Arduino Uno,

while for the servo and rtc motors the voltage is taken

from the 3.3v pin available on the NodeMCU.

For the implementation of the manufacture of feed

containers, acrylic is used as the main material in the

manufacture of containers. Acrylic material is

considered to have a lighter and more durable

material quality, because later this container is placed

in an outdoor environment. The following is the result

of the implementation of the container which can be

seen in figure 7. Pre-installed hardware will be

installed or integrated with the finished receptacle.

The hardware is stored on the back of the case

consisting of arduino uno, esp8266, beardboard, rtc,

etc. Meanwhile, the sensors are mounted on the

outside of the top of the container in order to take

maximum sensor data. There is a valve that is

connected to a servo motor as the driving motor of the

valve, so that the feed can come out through the hole

in the front of the container.

Figure 6: Assembled Hardware Components.

Figure 7: Implementation of feed containers.

2.2 Database Server (Firebase)

To accommodate sensor data from the

microcontroller, a real-time database system is

designed. Realtime database created using Firebase.

To create a Firebase database project, you must first

create a Google Firebase account. Then create a

database and select the Realtime Database type.

In this study, there are three children in one

database. The database that has been created is named

feeder. Each database that has been created will

automatically degenerate an account to perform

remote data. This remote data function is to send data

and retrieve data. The account that must be owned to

connect firebase with the server and gateway is the url

name and secret key which is automatically

degenerated according to the database name created.

Initialize firebase_host as account url and

firebase_auth as authentication account in the form of

a unique secret key. This secret key can be

regenerated by requesting a request to firebase. So in

connecting nodeMCU to firebase, it is enough to enter

the url and secret key.

Design of the Feeding System Automatic Koi Fish Based on Internet of Things Using the Fuzzy Logic Controller Method

925

Figure 8: Design Database.

Figure 8 displays a database containing data that

accommodates sensor data. The data type used in

firebase is Float. The children used in this study

include Weather, Fuzzy, and Feed. Weather is a child

that is used to store weather sensor data outside the

pool, while the reference child is rain, humidity, and

temperature. Fuzzy Child is a child that is used to

accommodate data on the value of defuzzification in

this research system, which is used as a determinant

of how long the feed valve opens. Feed Child is a

child that is used to accommodate feed data obtained

from ultrasonic sensors, there are reference children

namely distance, condition, and percentage of feed

availability. All children in this database have the

Float data type. Each data sent will have a unique id

generated by Firebase.

2.3 Software Application

Application made for visualization of data sent from

a microcontroller based on android. The android

application in this research uses the native java

language that is built using the android studio tool.

There are several features in the mobile-based

(android) application in this research, namely:

1) Feed Monitoring Features.

Users of this application will get information

about feed availability status data, information on

remaining feed, and information about the feed itself.

The sensor data displayed will be automatically

updated at any time (realtime) using the firebase

database.

2) Auto Feed Feature.

Users of this application will get information about

the feeding schedule every day at previously set

hours.

3) Weather Monitoring Features Around the Pool

This feature will display weather sensor data

obtained in real time and users can see the results of

fuzzy calculations as well as how long the feed valve

has been open.

4) Real-time Notification Feature.

Users of this application will get notifications in

real time if the feed valve opens according to a

predetermined schedule. In the notification there is

information on the schedule of feeding hours and how

long the valve opens at that hour.

3 EXPERIMENTAL STUDY

This section describes the performance of measuring

weather conditions and feed distances with each of

the available sensors. We compare dynamic

responses to get the weather state in seconds. We

analyzed the results of this comparison experiment to

find out how efficiently these monitoring devices

work.

The first is hardware testing and sensor

performance in detecting the weather.

3.1 Rain Sensor Testing

This test is carried out to determine whether the rain

sensor is able to read rain conditions through water

droplets. In this rain sensor, the test is carried out by

dripping water on the sensor panel using a tissue

moistened with water as shown in figure 9 below.

This test is a simulation as if it were raining in the

original conditions outside the room.

We observed the interaction of changes in the rain

sensor value whose sensor value was received and

displayed by Arduino Uno. Table 1 shows data from

the rain sensor that has been tested. It can be seen in

output number 1 when the rain senor is still dry and

has not been wetted by water and shows the value is

1022. Then at output number 2 the sensor starts to be

wetted by water droplets, and the value drops to 414,

as well as at output number 3, which are 439. The

sensor value means that the more puddles or water

intensity on the rain sensor panel, the smaller the

sensor value obtained. It can be seen that the more

water droplets are given, the smaller the value of the

rain sensor will be. The average delay obtained is 2

seconds. So it can be concluded that the rain sensor

data that has been tested has a change in value and the

rain sensor test has been successful.

iCAST-ES 2022 - International Conference on Applied Science and Technology on Engineering Science

926

Figure 9: Rain sensor testing.

Table 1: Rain Sensor Test Results.

Time

Analog

Value

Panel Condition