Evaluation of Single-Phase DC-AC Converters with Condition Monitoring Algorithm of Aluminum Electrolytic Capacitors Using Artificial Learnings with Various Circuit Signals and Filtering Combinations

Capacitors are essential parts of power converters since the cost, size, and performance of converters are mainly dependent on them. Nevertheless, the capacitor is the most degeneration device among all converter parts owing to its aging failures and little lifetime. Thus, the monitoring process is an essential route for valuing health status and gives predictive maintenance to ensure steadiness in electric converter. The equivalent series resistance and the capacitance are commonly indexes employed for estimating the condition grade of capacitors. In this research, six artificial intelligence (AI) algorithms are adopted to estimate the aluminum capacitor (Al-Cap) parameters in the single-phase inverter system. Various circuit signals, such as load voltage and current, capacitor voltage and current, are examined by utilizing the discrete wavelet transform (DWT) analysis and the combinations of fast Fourier transform with various filters. The considered signals are handled as AI model's inputs to guesstimate the health status of the Al-cap. In addition, the root-mean-square value is employed as an index to compare the accuracy with the analyzed signals. Furthermore, several indicators are mixed to acquire the best recipes for capacitor health evaluation.