Research on Brake Energy Recovery Strategy Based on Working Condition Identification

Abstract

As a transitional vehicle between fuel and electric vehicles, hybrid vehicles achieve energy savings and emission reductions without range anxiety. Regenerative braking has a direct impact on the fuel consumption of the whole vehicle; however, the current regenerative braking strategy for commercial vehicles is not yet perfect and has a poor adaptability in terms of working conditions and whole-vehicle load changes. Therefore, this paper proposes a regenerative braking strategy based on the identification of working conditions, by considering the influence of the vehicle load state and driving conditions on braking. Firstly, historical driving data of commercial vehicles were obtained from GPS data, driving conditions were classified using principal component analysis (PCA) and K-means, and a working condition recogniser was constructed using a back propagation neural network (BPNN) optimised with the Coati optimisation algorithm (COA). The recognition accuracy of the COA-BPNN was 7.6% better than that of the BPNN. Secondly, front and rear axle braking force distribution strategies are proposed, according to the braking intensity magnitude and load state under empty-, half-, and full-load conditions. Finally, a genetic algorithm (GA) was used to find the optimal control parameters for each category of working conditions, and the COA-BPNN condition recogniser identified the current category of working conditions needed to retrieve the corresponding optimal control parameters in the offline parameter library. The simulation results under C-WTVC and synthetic conditions show that the energy recovery rate of the proposed strategy in this paper reached up to 69.65%, which is at most 206.3% higher than that of the fixed-ratio strategy and at most 37.4% higher than that of the fuzzy control strategy.