Code and data from an ADALINE network trained with the RTRL and LMS algorithms for an MPPT controller in a...

Area of Science:

• Electrical Engineering
• Renewable Energy Systems
• Artificial Intelligence in Control

Background:

• Off-grid photovoltaic systems require efficient Maximum Power Point Tracking (MPPT) to maximize energy harvest.
• Traditional MPPT algorithms like Perturb and Observe (P&O) have limitations in dynamic conditions.
• Artificial neural networks offer a promising alternative for adaptive and robust MPPT control.

Purpose of the Study:

• To present comprehensive data from computational simulations and experimental tests of an ADALINE neural network-based MPPT controller.
• To compare the performance of the neural network MPPT controller against the traditional P&O algorithm.
• To provide accessible data and code for further research and optimization of MPPT control in photovoltaic systems.

Main Methods:

• Development and simulation of an MPPT controller using an ADALINE artificial neural network with FIR architecture in MATLAB/Simulink.
• Training the neural network controller with Recurrent Random Topology Learning (RTRL) and Least Mean Squares (LMS) algorithms.
• Experimental validation of the neurocontroller in an open-space, off-grid photovoltaic system under varying environmental conditions.
• Implementation of the MPPT neurocontroller in C for the PIC18F2550 microcontroller.

Main Results:

• The ADALINE neural network MPPT controller demonstrated effective performance in simulations and experiments.
• Comparative data show the behavior of the neural control method against the P&O algorithm under various test cases.
• MATLAB scripts for neural training algorithms and C codes for microcontroller implementation are provided.

Conclusions:

• The ADALINE neural network with FIR architecture, trained via RTRL and LMS, is a viable and effective control mechanism for MPPT in off-grid photovoltaic systems.
• The presented data and codes facilitate the evaluation and optimization of neural network-based MPPT control strategies.
• This research contributes to improving the efficiency and reliability of renewable energy systems through advanced control techniques.