Research And Development Of Adaptive Gear Selection System For Automatic Transmission

Along with the popularization of family cars, non-professional driversconstantly emerge in large numbers, and automatic transmission is providingthe important guarantee for their driving convenience and safety. Thedevelopment of control system is the nuclear problem in the research anddevelopment of automatic transmission, while the gear selection subsystem isone of the most important issues in control system, which is playing a keyrole in drivability, economic fuel, comfort, safety and emission ofauto-shifting cars. Thus the research on the gear selection system of automatictransmission has great practical significance.Regarding the Mitsubishi F4A42hydraulic automatic transmission as theresearch model, this paper develops an adaptive gear selection system withrapid prototyping method and tools. Firstly, based on the vehicle longitudinaldriving equation, the general gradient force is evaluated to identify slope;secondly, a method is proposed by which acceleration load, brake load,sportiness and hill-climbing load are used to evaluate the drivingrequirements and continuous real-time compensation is made to theconventional2-parameter shift lines; thirdly, a3-layer neural network is builtup to calculate the engine brake necessity and the downhill driving shiftfunctionality is realized by method of finite state machine (FSM); fourthly, aself-learning method based on the driver’s actions feedback is proposed tomake the downhill driving shift strategy more adaptive; fifthly, a rapidprototyping development platform is established with Bypass tools, on whichapplication software for the functions above is developed, integrated andcompiled for the real car test; finally, a series of trial and validation of software are made on the Hafei Saima1.6AT demo car.The test results show that compared with the conventional2-parametershift schedule, the new developed adaptive gear selection system shows muchbetter performance in drivability. Firstly, the adaptive system improves thepower performance while accelerating and climbing hills, and satisfies thepower requirements from drivers in different situations; secondly, while heavybraking and downhill driving, it can properly shift down to use engine braketo relieve the load of braking system and guarantee driving power forre-acceleration; finally, its self-learning feature can satisfy the differentdrivers with different preference of downhill engine brake.The research result shows that by using Bypass rapid prototyping system,the efficiency of software development can be greatly improved, withshortened R&D cycle and reduced development cost, implying the extensiveapplication prospect of this system in automotive electronics control systemarea. Besides its good performances, the adaptive gear selection systemdesigned in this paper has relatively simple software structure and algorithmsso that it will not occupy too much CPU resource, and it also providesadjustable parameters for the ease of the software applications in differentcars, showing the high practical value and promising application prospect ofthe adaptive gear selection system.