Deep Learning Based Suggestive Mechanism for Aquaponics
Smart Aquaponics is an automated farming system that provides environment and health conscious consumers with organic food while minimizing water consumption and maximizing yield by using a closed loop farming system. Aquaponics combines aquaculture ?fish farmin
2025-06-28 16:26:05 - Adil Khan
Deep Learning Based Suggestive Mechanism for Aquaponics
Project Area of Specialization Artificial IntelligenceProject SummarySmart Aquaponics is an automated farming system that provides environment and health conscious consumers with organic food while minimizing water consumption and maximizing yield by using a closed loop farming system.
Aquaponics combines aquaculture –fish farming, and hydroponics –growing plants in a soilless medium (Figure 1), in an aim to provide an alternative to food production on an industrial, as well as, private scale. The project has been designed by considering urban areas with low agricultural-land/per capita due to over-population and urbanization.
Once fully functional, the project will require minimal human interference during the cultivation process with an increase in food output compared to traditional farming techniques. The final product will be organic – free from contaminants such as pesticides – and cheaper than traditionally grown crops. The system will also provide users with useful suggestions regarding the next cultivation process and the feasibility of the current crop.
Apart from all this, the project also incorporates a Machine Learning Model which will help the farm owners decide about the crop yield, the feasibility of the project at different locations and lastly the acceptable crops that can be grown.
Project Objectives-
To design efficient systems in terms of water usage, and a higher crop yield.
-
To design a sensor mesh, generic in nature; so that it may be used in any other related art.
-
To develop power efficient and eco-friendly prototypes.
-
To develop a Machine learning based mechanism that predicts the feasibility according to location and the crop yield.
-
To increase awareness relating to the benefits of aquaponics.
-
To install portable, lightweight and sturdy systems; compatible for use in urban environments.
The block diagram depicts the proposed system. Starting from aqua-culture (1), where fishes are fed, and their waste is deposited at the bottom of the tank. The waste water from the aqua-culture is pumped out using a motor (2). The water reaches plants (3), is checked for its pH, Electrical Conductivity, Total Dissolved Solids and temperature readings. This is done at (4) using the sensor mesh, which communicates with the main server. The main server (5) collects the data from the mesh, and runs it through the Machine Learning algorithm (6) for monitoring purposes and provide valuable feedback to the user about the feasibility of the current crop in the given conditions.
The water travels through (7), where plants are grown inside a PVC structure. There ammonia in the water is reduced to nitrates, absorbed by the plants. The purified water is fed back to the aquaculture, as it is now harmless to the fish.
-
Less land and water required
-
Eco-friendly, oxygen is the by-product
-
Home grown vegetables, no pesticides
-
Larger yield per area
-
Cheap alternative to market products (tomatoes, cucumbers, bell peppers)
-
Locality
-
Self Sufficiency in Cities
The sensors are interfaced with the Micro Controller Unit which communicates with the main server using BLE and sends sensor values (of the respective sensors shown below) which are stored on My SQL server and also used in ML algorithm to provide with a prediction (Figure 9). The figure 10, below shows the specific details of the sensors used, i.e. their accuracy, working conditions, limitations etc.
The microcontroller will be communicating with the main server (central device) which is based on Raspberry Pi 3b+. The main server based on Raspberry Pi uses BLE for communication, it receives data through UUIDs of TX and RX, and uploads data to a MySQL database for storage. It also runs the data through a pre-trained Machine Algorithm.
The collected data of the sensors is stored in an XLS file. The data points of each sensor, are extracted and saved in a numpy-array (array like structures in python). Curve fitting is applied to each attribute, i.e. we will have a curve for pH, EC, temp etc.
Finally, this data is used for the machine learning model. The ML approach is as under:
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 63990 | |||
| Red Bear BLE NANO E | Equipment | 1 | 7100 | 7100 |
| Temperature Sensor | Equipment | 1 | 890 | 890 |
| pH Sensor | Equipment | 1 | 6700 | 6700 |
| EC Sensor | Equipment | 1 | 8200 | 8200 |
| PVC Structure Pipes | Equipment | 1 | 12800 | 12800 |
| Water Pump | Equipment | 1 | 1800 | 1800 |
| Air pump | Equipment | 1 | 3300 | 3300 |
| Aquarium | Equipment | 1 | 3500 | 3500 |
| Tilapia fish | Miscellaneous | 30 | 60 | 1800 |
| Server Cost | Miscellaneous | 1 | 3500 | 3500 |
| Website Domain Cost | Miscellaneous | 1 | 2500 | 2500 |
| Ammonia Strips | Equipment | 3 | 1600 | 4800 |
| Travel Cost | Miscellaneous | 1 | 2000 | 2000 |
| Plant Seed packets | Equipment | 30 | 50 | 1500 |
| PCB Printing | Equipment | 1 | 400 | 400 |
| Axino Module | Equipment | 1 | 3200 | 3200 |