Artificial Intelligence Based Performance Enhancement of AC-DC Power Converters for Smart Energy Systems

Abstract

With such operation mode, the smart energy systems require a high-power quality and fast regulation of electrical signals, making artificial intelligence (AI)guided control increasingly important for AC-DC power converters under continuously changing operating conditions. To bridge the conventional fixed-gain controller with rapidly evolving operation settings, such as renewables or smart loads, this work proposes to augment a single-phase AC-DC converter with machine learning (ML)-based duty cycle correction, online monitoring, and maintenance-based diagnostics and optimization functions. Enhancements to output-voltage regulation, input power quality, and diagnostic capability with stable conversion from utility alternating current (AC) to regulated direct current (DC) are the primary goals. The description of this converter model is given in terms of source-voltage, source-current, inductor, capacitor, duty-cycle, power factor (PF) and total harmonic distortion (THD) relations, wherein the results presented proved successful regulation at a 230-volt root mean square input level supply, giving a resultant 325volt peak input wherein the average must be 206.9 volts at the rectified level, which shall provide up to a 320-volt regulated DC output along with a steady-state duty cycle. As depicted in comparative operating cases, it can output a power of 971.1 watts with a fault detection rate of up to 97.5%, and prediction accuracy reaches as high as 96.6% while maintaining reduced current distortion and controlled voltage ripple. The performance is compatible with renewable energy interfaces, smart loads and energy storage systems.