| Medium and heavy duty trucks consume a lot of oil,which are one of the largest CO2 emission sources in transportation.Energy saving and emission reduction of heavy-duty commercial vehicles is very important.Hybrid system is one of the important means to achieve energy saving and emission reduction.Foreign countries have done a lot of research on heavy-duty hybrid electric commercial vehicles,and have launched mature products.China’s research on hybrid heavy-duty commercial vehicles is still in its infancy,so it is urgent to study and formulate a reasonable energy management strategy for hybrid electric heavy-duty commercial vehicles.The advantages and disadvantages of energy management strategy are closely related to driving conditions.In this paper,a single axle parallel hybrid truck is taken as the research object,aiming at the identification of driving pattern recognition and the optimization strategy of energy management.Based on the driving conditions of domestic and foreign heavy-duty commercial vehicles,the typical driving conditions database for heavy-duty hybrid electric trucks is established by using hierarchical clustering and comparative analysis.Using the method of "Kruskal-Wallis Test ",this paper analyzes the distribution difference of characteristic parameters in different conditions,and selects 12 characteristic parameters to represent the driving information.In this paper,BP neural network algorithm is used as the pattern recognizer to train the typical driving conditions offline.The relationship between the slide window time,the prediction time and the correction of pattern recognition is studied.After full training,the recognition accuracy for single scene is as high as 99.6%,and for composite driving cycle is as high as 87.5%.According to the principle of equivalent fuel consumption,a mathematical model of energy management strategy based on the equivalent consumption minimum strategy(ECMS)is established.Using LMS Amesim to build the vehicle dynamics model,establishing the control strategy model in Simulink,and building the cosimulation platform through the software interface.The best equivalent factor of each typical driving cycle is obtained by the method of experimental design.On this basis,the dynamic programming method is used to obtain the global optimal energy distribution method for each typical driving cycle under different initial SOC values,and the optimal SOC distribution path is extracted as the reference path of the penalty function in the ECMS,which enhances the adaptability of the original control strategy to SOC.Combined with the pattern recognition and the adaptive ECMS,an energy management strategy based on pattern identification is constructed.A composite driving cycles conditions covering six typical patterns are used to simulate and analyze the energy management strategy.The results show that the fuel economy of the energy management strategy based on pattern recognition is improved by 3.37%.In this paper,the combination of pattern recognition and optimal control enhances the adaptability of energy management strategy to different driving cycles,reduces fuel consumption,and has certain guiding significance for the development of energy management strategy of hybrid electric heavy-duty truck. |
