Cognitive Insights into Metaheuristic Digital Twin based Health Monitoring of DC-DC Converters

Reliability of components has always been a major concern to the performance and stability of DC-DC converters. After long-term operation, these passive components and switching devices start to degrade and become weak to withstand normal electrical and thermal stresses. An insightful digital interface to the physical layer known as Digital Twin (DT) can be a sustainable solution for ensuring reliability. This paper extends the DT concept to component level health monitoring in DC-DC converters. The proposed concept is noninvasive and does not require additional sensors. The working principle is to minimize the weighted least squared error between the digital twin output and the measured data of state variables through metaheuristic optimization. An application for Two-Phase Interleaved Boost Converter with reverse coupled inductor is considered and Hardware-in-the-loop (HIL) platform is used for sensitivity analysis for component degradation. Further, the optimization problem is solved using the following two popular metaheuristic optimization methods: Particle Swarm Optimization (PSO) and Genetic Algorithm (GA). Further, the performance of both methods for 20 executions in terms of computational time; convergence rate and dispersion are compared. It is evident from the results that GA outperforms PSO with 50 % less execution time and better accuracy> 95 %.