F2008-05-018
Automatic Calibration Method of Servo Gain Curve on Electronically Controlled Brake System with Genetic Algorithm
Electronically Controlled Brake System (ECB/ Fig 1) is realizing the high performance control functions contributing fuel-efficient and vehicle safety such as Regenerative-Friction Brake Coordination, Vehicle Dynamics Integrated Management (VDIM), and ABS etc. Currently this system has been applied to various cars, for example, hybrids car, "Prius". In order to install ECB to other worldwide cars, the calibration process has been required to adjust each car's demands similar to other vehicle systems like power train. This tuning procedure, which has been carried out manually, requires repetitions of trial and error. This is because there exist lots of hardware specification of calipers, or specific demands of each car or target destination, and there are lots of evaluation points and modified parameters. Therefore, in order to keep high quality and to reduce the expense of tuning, there has been the demand of automatic calibration. This proposed method based on Genetic Algorithm (GA) has been constructed following the ECB´s calibration procedure. In order to apply GA, which is usually adopted on discrete off-line processing, to continuous on-line fluid system like ECB there are three points should be embedded with the algorithm. 1. Tuned parameters should be presented with genetic pattern (Fig 2). 2. Controllability should be defined as a scalar quantity according to the demands of the target vehicle, considering particular behaviors of fluid system 3. Effective modification of tuned parameters should be carried out according to the defined controllability Furthermore, in order to verify the validity of the proposed algorithm, following four points should be examined. 1. Convergence towards the desired area of target modified parameters within the finite span (Fig 3) 2. Comparison of controllability between automatic tuning and manual tuning 3. Repeatable result of this algorithm 4. Implementation with different calipers, which have various characters. Finally, considering above requests and confirming the above results, it has been cleared that this proposed algorithm has been able to compute the desired gain curve automatically with very brief time.
This abstract is supplemented by a PDF, which can be viewed here.
Session: Process