Vehicle Handling Targets Mining And Consistency Of Subjective And Objective Evaluation

Compared with other vehicles, the automobile is used most widely, and the human-vehicle interaction is the most directly and every user has the right to choose or evaluate an automobile. Hence, the future automobile will become more and more personalized and humanized on the promise of safety, comfort, energy saving, environmental protection, which requires the OEM predict the developing automobile’s performance accurately during the automotive design and development, and realize the real forward automotive design with modular, digital and intelligent technology.As one of the key factors influence on vehicle active safety, the handling not only affects the convenience of vehicle driving, but also decides the safety of high-speed cars. However, with the limit of research contents, driving conditions, subjective feeling and surroundings, there has not been a recognized vehicle handling objective evaluation system so far.The establishment of vehicle handling objective evaluation system is essentially a non-linear classification problem with big data, many-to-many and tight coupling. Aimed at the problems existing in the subjective and objective consistency research, this dissertation is intended to explore a relatively simple objective-target-mining method; establish a good-consistency handling evaluation system; design a optimization and decomposition scheme for suspension characteristics from vehicle performance. In this study, theoretical analysis, numerical simulation and experimental method are combined for the main investigations and conclusions shown as follows:1. Objective targets mining. Continuous targets turn to discretization by the algorithm of information entropy. With the target-selection model based on discerni-bility matrix of rough set and the limited tolerance relation improved, attribute reduction of the objective evaluation system get rid of redundancy and 12 key objective targets are extracted.2. Objective targets weights multiple-attribute decision making(MADM). Integrated objective and subjective weights, an experience factor model is proposed. With the combination of information entropy and rough set, the importance of 4-level subjective targets is calculated based on MADM by CWAA operators, which is used to acquire the relative sequence of 12 key objective targets in evaluation system.3. Establishment of objective evaluation system. Selected as membership function, kernel function is used to combine objective targets and their weights, which form the comprehensive evaluation result with high accuracy of 80%, strong generalization and satisfactory consistency. Aiming at the model structural optimization, on one hand, coupled system reduction based on normalization and structure-preserving is used to reduce system dimension, improve operation efficiency and maintain system stability; on the other hand, 4 comprehensive evaluation targets are extracted to reduce attributes and computation. The two methods both have feasibility with high accuracy of 70%.4. Optimization and decomposition for suspension characteristics from vehicle performance. A full vehicle Car Sim model is proposed and validated, and the suspension characteristics are parameterized with a 12-parameter-dimensionless evaluation method for vehicle handling. And then, a suspension K&C characteristics optimization platform is developed via co-simulation with Car Sim, Matlab and i SIGHT. An optimization model for the initial suspension alignment parameters and the suspension K&C characteristics are proposed. The suspension K&C characteristics are optimized and decomposed from the full vehicle targets via an improved multi-objective genetic algorithm.This dissertation gives practical approaches for the consistency of vehicle handling subjective and objective evaluation, provides a basis for suspension optimization, and is also a foundation of the modular, digital, intelligent development technology.