Optimal FPGA Implementation of Dense Extended Kalman Filter for Simultaneous Cell State Estimation

This work considers hybrid embedded implemen-tations of the dense extended Kalman filter (DEKF) - a novel state estimation technique which reduces the traditional Kalman filter's time complexity from <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{O}(N^{3})$</tex> to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{O}(N)$</tex> - to estimate the state-of-charge (SOC) of 16 serially-connected lithium-ion battery cells. Design space exploration (DSE) is used to recommend an optimal hybrid binding considering execution time and energy as the primary objectives. High-level synthesis (HLS) tools are used to inform the DSE and to aid in implementation. Furthermore, each implementation's performance is gauged and compared to our MATLAB model of the DEKF and our DSE. Our results indicate close precision and accuracy to MATLAB with RMS near 0.2% with execution times as low as 52 <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mu \mathrm{s}$</tex>.