Neuro-controller implementation for the embedded control system for mini-greenhouse

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Embedded Systems

Background:

• Traditional control systems face limitations in embedded applications.
• Neural networks offer advanced control capabilities but pose implementation challenges.
• Existing solutions often lack modularity and cost-effectiveness.

Purpose of the Study:

• To propose a novel neuro-controller for embedded systems.
• To design a modular neuro-controller structure for rapid development.
• To demonstrate a cost-effective implementation using STM32 microcontrollers.

Main Methods:

• Development of a modular neuro-controller architecture.
• Design of a functioning algorithm and data processing model using artificial neural networks.
• Hardware implementation utilizing STM32 microcontroller, sensors, and actuators.

Main Results:

• A functional neuro-controller enabling processing of technological data.
• A modular design facilitating system improvement during development.
• A low-cost hardware implementation suitable for embedded applications.

Conclusions:

• The proposed neuro-controller effectively integrates artificial neural networks into embedded control systems.
• The modular design and software-based neural network implementation allow for rapid adaptation and improvement.
• The STM32-based system provides a practical and economical solution for intelligent control, exemplified by a mini-greenhouse application.