FPGA-Based Online Parameter Identification for Active Power Filter

In active power filter (APF), the output filter inductance mostly-changed with temperature variation and insulation aging. In order to suppress the influence of inductance parameter variation on the current compensation performance, online identification of the actual inductance is required. However, the traditional field-programmable gate array (FPGA)-based online parameter identification implementation requires lots of resources without reducing the identification accuracy. In order to reduce resource occupation while maintaining identification accuracy, an FPGA-based implementation of parameter identification with forgetting factor is proposed for APF. And time-sharing multiplexing method is adopted in matrix multiplication. The effectivity of the proposed implementation method is validated by simulation and experimental results. It shows that the error between the identification and the actual inductance decreases to 6%. When the output filter inductance changes, the THD of the grid current before identification is 2% to 10%, and the THD of the grid current after identification is always less than 3%.