Research On Muti-mode Control Strategy Of Start-Stop System Based On Driving Cycle Recognition

The start-stop system is the key technology of micro hybrid vehicle. It can reduce fuel consumption and emissions in engine idle speed, so it has been widely used and has become the current hot problem in the study on vehicle energy conservation and emissions reduction. However, two major problems facing the start-stop system was invalid idle downtime and frequent start-stop, what resulting in a loss of fuel economy and the wastage of the auto parts. The driving condition is the main cause of invalid idle downtime and the frequent start-stop. In order to make the start-stop system more effective, respectively on different kinds of driving condition to design reasonable start-stop control strategy. In order to make the start-stop system more intelligent, to automatically identify the driving condition.This paper uses the reasonable control parameters(delay time and speed threshold) of start-stop system control strategy for research based on driving cycle recognition. In this paper, the main research contents include:(1) Analyzing the start-stop system structure, working principle and the basic control strategy. To avoid the invalid idle downtime and frequent start-stop, the effective method is choose the reasonable delay time and speed threshold. Expounds the relationship between driving condition and these two factors.(2) Given the driving condition of the impact on the start-stop system. In order to avoid invalid idle downtime and frequent start-stop, design reasonable control strategy under different driving modes. This article selects four kinds of typical driving cycle. The four kinds of typical driving modes of idle distribution characteristics were analyzed. Select the average speed during every idle stop block as the speed of threshold. The control parameter optimization method is put forward based on nonlinear constrained optimization. Take the original problem into nonlinear constraint programming problem. Among the problem, the objective function is save the maximum idle time, the control variable is delay time and speed threshold. So that it can respectively obtain the best control parameters under four kinds of typical driving conditions, namely the multi-mode control strategy.(3) The driving cycle recognition is the premise to realize multiple model control strategy. This paper introduces the research status of driving cycle recognition. To improve the accuracy in driving cycle recognition, using feature selection method based on the framework of Wrapper to select the optimal characteristic parameters subset.(Including breadth-first search for complete characteristic parameters and SVM as evaluation function). Take advantage of neural network, using the LVQ neural network to identify the driving cycle.The start-stop system multiple model control strategy based on the driving condition recognition is simulated in this paper. First, using LVQ neural network of the optimal characteristic parameters subset to identify four kinds of typical driving cycle, the recognition accuracy of 95.02% can well satisfy the identification requirements. Second, the optimization control strategy is used for the simulation analysis under four kinds of typical driving conditions. Simulation results show that the optimal control strategy can achieve better fuel economy and effectively avoid invalid idle downtime.