AI-Driven Optimization Approach for Enhanced Performance of Power Converters through Efficient Controller Tuning

The electric power sector is continuously embracing new approaches to improve the reliability and efficiency of the energy system while addressing the growing energy demand and associated technical problems. Recently, the development of artificial intelligence (AI) has provided researchers with powerful tools to deal with various challenges in the power system. One significant advancement in this field is the Voltage Source Converter (VSC), which leverages advancements in power electronics and semiconductor technology. VSC holds immense potential for realizing smart grids, integrating renewable energy, and enabling HVDC transmission systems. Traditionally, the manual tuning of Proportional-Integral (PI) controllers for VSCs depends on a trial-and-error approach or the experience of design engineers, which does not show optimal performance. This process becomes even more complex when dealing with multiple grids, such as VSC-based Multi-Terminal DC (MTDC) grids. In this research article, a Bayesian optimization (BO) based artificial neural network (ANN) technique is employed to tune the VSC controllers, leading to significant improvements in dynamic response under fault and unbalanced load scenarios. To evaluate the proposed method, multi-terminal MTDC network is built in PSCAD/EMTDC software. The suggested technique is compared with the classical technique under fault and unbalanced load conditions.