Adaptive Neuron Adjustment for Transient Accuracy Enhancement in FPGA-Based Power Electronics Simulation

An adaptive neuron adjustment strategy is proposed for the real-time simulation of transient processes in power electronic devices on FPGA platforms, aimed at enhancing model accuracy within the constraints of existing computational hardware resources. By dynamically reallocating neuron resources based on the computational complexity of different transient states, the proposed method optimizes both resource utilization and model precision. The strategy has been implemented on a self-developed FPGA-based real-time simulation platform, using a bidirectional buck-boost converter as a case study. Comparative analysis between real-time and offline simulation results demonstrates the effectiveness and general applicability of the proposed algorithm, highlighting its potential for broader applications in power electronics simulations.