Online rolling parameter identification of fuel cells in unmanned aerial vehicle applications via a convex optimization-wavelet transform method

Accurate parameter identification of fuel cells is critical for ensuring the stable and reliable flight of unmanned aerial vehicles. This study proposes a combined convex optimization-wavelet transform (COWT) method for online parameter identification of fuel cells. The proposed method integrates rolling convex optimization and wavelet transform to achieve high-accuracy parameter identification under strong measurement noise. The convex optimization method is proposed to solve the parameter identification problem, and an online rolling framework is developed to improve the real-time capability and optimality. In addition, wavelet transform is introduced to separate high-frequency noise and enhance the dynamic characteristics of the measured data. The parameter identification experiments are conducted to demonstrate the effectiveness and practicability of COWT. The experimental results show that COWT exhibits excellent performance for the parameter identification problem. Compared with recursive least squares and sequential quadratic programming, COWT reduces the mean absolute error between the model output voltage and the experimental measurement voltage by 19.4% and 16.7%, respectively. It also decreases the root mean square error by 16% and 19.3%, respectively. This study is conducive to real-time fuel cell performance management under rapidly varying UAV flight conditions. It also enables a reliable and efficient power supply and enhances flight safety for UAV missions.