Experimental validation of real-time energy management for hybrid energy storage systems based on predictive wavelet transforms

The wavelet transform (Wav), due to its capability of effectively allocating high- and low-frequency energy, has been widely applied in real-time energy management strategy (EMS) for hybrid energy storage systems (HESS). However, the time delay generated by Wav during signal processing can reduce the overall effectiveness of HESS. In this paper, a real-time hardware platform for HESS is built to study the effect of Wav on the performance of batteries (Bat) and supercapacitors (SC) with and without neural networks for delay compensation. Furthermore, a current-prediction-based predictive adaptive wavelet transform method is proposed to enhance system real-time performance and examine the potential risks of speed prediction. Rapid control prototype experimental results show that, compared with conventional Wav, the predictive wavelet transform can reduce SC peak current by approximately 7.6 % and 15.6 % in urban and highway conditions, respectively, and decrease energy cycling losses by about 5 %. The proposed predictive adaptive wavelet transform further reduces peak current by approximately 12.6 % and 20 %, decreases energy cycling losses by 14 % and 20 %, and extends battery lifetime by 1.5 %.