Scientific African (Jun 2024)
Qualitative performance improvement of a hybrid power supply at the DC common coupling point using a neuro-fuzzy method
Abstract
Continuity of power supply service in production industries is of crucial importance, given the severity imposed in these sectors. A poor power supply leads to underproduction or degradation of equipment sensitive to electrical variations. A direct consequence of this is loss of earnings for industry. To ensure continuity of service for industrial power supplies, at least one secondary power source must be available. However, the switching system between sources may not be spontaneous due to the presence of electromechanical switches. The solution to this problem is to use electronic converters for control. The aim of this work is to enhance the parameters of an industrial photovoltaic-grid hybrid power supply to ensure uninterrupted power supply and balance between energy supply and demand. The sources are coupled to the DC bus via electronic converters. The photovoltaic system and the inverter are assumed to be unknown, and only the load terminal voltage is accessible and measurable. To address the lack of mathematical model of the system, we propose a mechanism that combines a fuzzy logic (FLC) centred mechanism with a radial basis function (RBF) neural network for the studied case. The FLC self-determines the image of the compensation power to be extracted through the current to the secondary energy source, ensuring continuity of service and balance between supply and demand. The FLC mechanism self-adjust the parameters of the RBF neural network and coordinates the energy at the DC link. The results obtained in the Matlab/Simulink environment are compared with the fuzzy logic and proportional integral (PI) control method. The quality of the results show that the proposed method is excellent compared with FLC and PI control. It improves response time between 24.223% and 82.9% for PI and between 22.292% and 63.694% for FLC. It guarantees an improvement in the mean square error on PI of between 5.865% and 57.994%, and on FLC of between 1.582% and 29.683%. The proposed method guarantees a high rationing coefficient, helping to maintain the balance between supply and demand, ensure continuity of service for the power supply, and improve the accuracy and self-compensation of the energy to be supplied, both qualitatively and quantitatively.
