The efficient operation of power electronic converters (PECs) is crucial for various industrial applications, such as renewable power generation. System Parameter Identification (SPI) plays an important role in maintaining PEC performance by indirectly assessing the conditions of critical components through available physical signals, without additional hardware installation or operation interruptions. This study focuses on the estimation of the phase inductances and the DC-link capacitance of a three-phase AC–DC converter using ensemble models with Gated Recurrent Unit (GRU) neural networks, leveraging permutation feature importance and delta parameters for dimensionality reduction. Experimental results demonstrate that the proposed approach achieves high performance scores with reduced computational costs. The outcome of this work is the first successful SPI step in a wider prognostics process, which will be suitable for practical applications.