Research On Control Strategy Of Idling Start-stop System Based On Multi-modal Information

The original intention of the idling start-stop system design is to save energy and reduce emission.However,due to various factors such as traffic control signals and traffic jams during the practical application process,vehicles with start-stop system often experience short-term idle and frequent start-stop,which could increase equipment wear and reduce driving comfort,rather than save energy.So many drivers often turn the system off.This study comprehensively analyzes various factors such as traffic control signals,traffic environment and driving conditions,and proposes a method by predicting idle conditions to control engine start and stop to effectively reduce the number of invalid idling start-stop and improve the fuel economy of idling start-stop system.The purpose of this study is to improve the fuel economy of the idling start-stop system,and to conduct a study around the prediction of idle conditions and the establishment of start and stop control strategies.The main research contents are as follows:1)Idle condition analysis of actual urban roadsIn order to fully consider the influence of road environment,driving conditions and other factors on the start-stop of vehicles,Shanghai,Wuhan,Shenyang and Changchun are selected as experimental sites,and a large number of driving data are collected for systematic analysis.The time threshold for invalid idle speed is determined by summarizing the relevant research and the actual measurement of the vehicle.According to the invalid idle time threshold,the distribution rules of idle speed and invalid idle speed in different cities and different time periods and the distribution rules of velocity under different idle conditions are analyzed,which provides basic analysis based on the actual data for the improvement of the subsequent algorithm and data support for model verification.2)Start-stop control method based on traffic control signalsTaking the effect of traffic control signals on the start-stop system into comprehensive consideration,a start-stop control method based on traffic control signal acquisition is proposed and a traffic control signal recognition system based on the vehicle-mounted monocular camera is constructed in this paper which can effectively identify the lights and countdown information in front of the vehicle during driving,and calculate the distance between the vehicle and the signal lights in real time according to the distance estimation model,so as to predict the idle condition to control the start and stop of the vehicle.The appropriate color space is selected in the detection of signal lights,a variety of visual features are analyzed and the appropriate feature vectors are determined,and the WKNN model is proposed to recognize signal lights and countdown.According to the results of image recognition,the start-stop control strategy is developed.The experimental results show that the improved start-stop control strategy can effectively improve the fuel economy of the start-stop system,with an average increase of 11.64% during the peak period and an average increase of 9.63% during the off-peak period.3)Method for predicting the idle conditions based on spatiotemporal learningBased on the analysis of actual traffic data,the paper proposes a method to control the start and stop of vehicles according to the idle condition prediction result and a method for predicting idle conditions based on spatiotemporal learning.In order to determine the basic model suitable for idle prediction,the generative model Marcov model and the discriminative model LS-SVM model are used for prediction.When using the Marcov model for prediction,the prediction effects of different time scales and different state classifications are compared,and the optimal parameters of the Marcov model are determined;when using the LS-SVM model for prediction,the impact of short-term driving conditions,nuclear functions,penalty parameters,and nuclear parameters on the prediction are analyzed to determine the optimal parameters of the LS-SVM model.The two methods are compared and the LS-SVM is used as the basic model for predicting idle conditions,in order to improve the prediction model for the characteristics of invalid idle speed distribution,a LS-SVM idle condition prediction method based on spatiotemporal learning is proposed and a start-stop control strategy is developed according to the prediction results.The verification experiment shows that the improved start-stop system control strategy is significantly more fuel-efficient than that of the traditional start-stop system,with an average increase of 17.09% during peak periods and an average increase of 5.40% during off-peak periods;it also effectively reduces the number of invalid idle start-stop,with an average reduction of 74.43% during peak periods and an average reduction of 53.33% during off-peak periods.4)Weight-adaptive idle condition prediction method based on similarities of driving conditionsIn order to further improve the accuracy of the idle prediction,a weight-adaptive idle stop time prediction model based on the similarity of driving conditions is proposed.The maximum expectation algorithm is used to cluster the driving conditions based on the characteristics of the idel-related interval driving conditions and six categories of driving conditions are determined.When the prediction of idle condition is performed,the characteristics of the conditions to be predicted in the same interval are extracted to determine the similarities between the conditions to be predicted and the driving conditions in each category,the driving conditions in each category are completed separately,and the final prediction results are determined based on the weight of similarity.Taking various information such as the traffic control signals and the idle conditions into comprehensive consideration,an idle start-stop control strategy based on multi-modal information is proposed.The experiment shows that this start-stop control strategy is significantly more efficient than that of traditional start-stop system and the number of invalid idle is significantly reduced with the average fuel economy increasing by 23.17% and the average invalid idle stop-start decreasing by 93.59% during the peak period,the average fuel economy increasing by 8.79% and the average invalid idle stop-start decreasing by 74.17% during the off-peak periods.The paper proposes an idle start-stop control strategy based on multi-modal information.It comprehensively considers the traffic control signal information,short-term driving conditions,interval driving conditions,time,space,and other modal information to predict the idle conditions of the vehicle.The start-stop of the vehicle is controlled according to the prediction results,the number of invalid start-stop is effectively reduced and the fuel economy of the start-stop system is improved.