Optimal Design and Performance Investigation of Artificial Neural Network Controller for Solar- and Battery-Connected Unified Power Quality Conditioner

Nowadays, integration of renewable sources into the local distribution system and nonlinear behavior advanced power electronic equipment have made a large impact on quality (PQ). The unified conditioner (UPQC) is multifunctional FACTS device, which combination both shunt active filter series filters via common DC link. Presently, artificial intelligence playing vital role in development intelligent control methods. Traditional training methods neural network (ANN) like back propagation Levenberg-Marquardt may get stuck optimal solution leads to invention ANN trained optimally by metaheuristic algorithms. This paper develops firefly algorithm-trained (FF-ANNC) controller for proportional integral (PI-C) UPQC integrated with solar energy battery storage boost converter (B-C) buck converters (B-B-C). main aim proposed FF-ANNC reduce mean square error (MSE) thereby achieving constant link capacitor voltage (DLCV) during load irradiation variations, reduction imperfections current waveforms, improvement factor (PF), mitigation sag, swell, disturbances, unbalances grid voltage. working developed was tested five test studies different types loads source balancing/unbalancing conditions. However, demonstrate supremacy suggested FF-ANNC, comparative study genetic algorithm (GA) ant colony optimization (AC-O) also other that exist literature PI-C, fuzzy logic (FL-C), neuro interface (ANFI-S) conducted. method reduces total harmonic distortion 2.39%, 2.32%, 2.27%, 2.45%, 2.66% are lower than existing available literature. shows an excellent performance reducing fluctuations (THD) successfully improving PF.