Research On Predictive Shift Decision For Automated Manual Transmission Of Heavy Commercial Vehicles

As an important part of road transportation,the fuel economy and driving maneuverability of heavy commercial vehicles have received extensive attention from manufacturers and users.The application of automatic transmission on the one hand reduces the driver’s driving labor intensity and improves maneuverability through automatic gear shifting,and on the other hand,improves the utilization rate of the engine’s high efficiency zone and improves fuel economy through reasonable speed ratio matching and gear utilization.Automated manual transmission(AMT)has become the preferred solution for domestic and foreign heavy commercial vehicles with its cost advantage,high transmission efficiency,high technical maturity,suitability for multiple gears flexible arrangement of speed ratios.In 2022,the AMT penetration rate of domestic heavy-duty trucks on the road exceeded 20%,which is expected to exceed 50% by 2025.Gear decision(also known as shift schedule)to solve the problem of when to shift,that is,according to people,vehicles,roads,and environment to choose a reasonable gear,is the core technology of automatic transmission theory,but also one of the key technologies of AMT product development.Gear decision-making technology has gone through four stages: single-parameter,two-parameter,dynamic three-parameter,and intelligent gear decision-making.Intelligent network technology empowers vehicle powertrain control,bringing new opportunities for the development of gear decision-making technology.Combined with predictive road information such as road gradient and road curvature,as well as location information such as vehicle position and vehicle positioning,vehicle speed,and gear position are reasonably planned to adapt to changes in the driving environment,taking into account fuel economy and transportation efficiency.This thesis takes heavy commercial vehicle AMT as the research object.To improve vehicle fuel economy,this thesis combines the gear decision-making technology with electronic map and vehicle state prediction technology,to research anticipatory gear decision-making for heavy commercial vehicles.The main research content includes:(1)Based on Truck Sim and MATLAB/Simulink,a joint simulation platform is established,which includes the engine model,transmission system model,gear decision model,and longitudinal dynamics model of the whole vehicle,and the accuracy of the simulation platform is verified through the real-vehicle test,which provides the simulation platform basis for the subsequent development of gear decision.On this basis,the types and characteristics of the basic shift strategies are analyzed,and the basic power and economy gear decision-making method is designed using vehicle speed and throttle opening as the control parameters by using the analytical method.The evaluation indexes of AMT gear decision-making for commercial vehicles are constructed from the aspects of power,economy,and drivability,which provide the basis for gear decision optimization and comparative analysis.(2)To improve the accuracy and stability of vehicle mass estimation for heavy commercial vehicles,a data-driven vehicle mass estimation method is proposed.Firstly,the vehicle networking platform is utilized to obtain predictive road information such as road gradient and road curvature,and the road information is applied to the vehicle mass estimation and gear position decision-making system.The RBF neural network is used to learn the vehicle driving history data,and the confidence factor is calculated to characterize the vehicle driving state by comparing the learned data with the actual collected data.Based on the least squares method with the forgetting factor,the vehicle mass is estimated,and the confidence factor is used to correct the control gain,to obtain a more accurate estimate.The effectiveness of the proposed data-driven vehicle mass estimation method is verified through real vehicle data simulation and real vehicle tests,and the error of the proposed algorithm for vehicle mass estimation is about 2%.(3)In order to solve the problem of optimizing the gearshift strategy under curves and ramps,the architecture of the gearshift strategy considering the influence of multidimensional factors is established.We analyze the influencing factors of gear decision-making,and through the four steps of inputting the vehicle state and foreseeable road state,calculating the powertrain limitation,selecting the shift strategy,and calculating the final gear decision-making,we complete the shift strategy considering the influences of multi-dimensional factors and propose the corresponding shift strategy for the on-ramp and curved road working conditions.The effectiveness of the proposed shift strategy for ramp and curve conditions is verified by simulation analysis.(4)To improve the adaptability of the gear decision system to future uncertain driving conditions and reduce vehicle fuel consumption,a multi-timescale predictive gear decision method based on a two-layer control architecture is proposed.A speed prediction model based on the long and short-term memory neural network model is constructed to predict the speed for future working conditions.The predictive road information obtained through the vehicle networking platform and the prediction of the vehicle state in the predictive time domain are used as inputs to construct the predictive gear decision-making method based on a two-layer control architecture with multiple time scales.Based on Dynamic Programming(DP),the upper layer establishes a gear decision optimization problem that takes into account both economy and drivability,adopts the DP algorithm to optimize the gear sequences in the long-time domain,and uses the optimized gear sequences as the basis for the lower controller to make gear decisions in the predictive time domain.The lower controller combines the predicted operating condition information of the vehicle in the short-time domain,establishes the gear decision optimization problem considering the economy,and carries out the real-time optimization of gears in the short-time domain based on the Levenberg-Marquarelt algorithm.The simulation verifies that the proposed predictive gear decision method can further reduce fuel consumption compared with the economic gear shift strategy.(5)To verify the effectiveness of the proposed anticipatory gear decision-making method.A domestic heavy commercial vehicle equipped with a 12-speed AMT is selected as the test platform,and the self-developed AMT controller and hardware system are used for the test verification.For the vehicle uphill,downhill,and curved road shifting strategy for the corresponding conditions of the test verification,to verify the effectiveness of different conditions of the shifting strategy through the real-vehicle road test,the predictive gear decision-making method of the real-vehicle test verification,compared with the original economic shifting strategy,the predictive gear decision-making method can effectively reduce the fuel consumption of about 3.3%.