Error Storage Based Online Linearization of the Nonlinear Transfer Function of a High Power Dual Active Bridge

This paper introduces an error storage based online linearization method for the reduction of the nonlinearity inherent in the current transfer function of a Dual Active Bridge in order to increase the possible dynamics of current and voltage control. The proposed algorithm autonomously learns deviations from the ideal current transfer function based on previous operational data of the current controller. This approach enables real-time compensation through the implementation of an error storage system, resulting in an effective online compensation method. Furthermore, this method is not limited to Dual Active Bridges but can also be used for any other time invariant system with repeating operation points and nonlinear transfer characteristics. The algorithm is implemented on a Field Programmable Gate Array and experimentally verified on a 40 kW test bench. Experimental measurements affirm the practical feasibility and effectiveness of this algorithm.