CAJ | 학술논문

通过对车辆转向机构的尺寸和定位参数进行优化,能有效减小车辆转向机构的实际运动轨迹和理论运动轨迹间误差,进而有效改善车辆的操纵性能和提高转向安全性.该文研究了转向梯形机构的工作原理及其对车辆转向性能的影响,建立了转向梯形机构的非线性优化模型;然后引入越界检测函数改进传统粒子群优化算法,并给出了求解转向梯形机构非线性优化模型的方法;编制了改进粒子群算法的实现程序,并对3款不同车型的转向梯形机构进行了优化设计;最后选取3种不同智能算法分别对途乐 GRX 转向梯形机构进行多组优化试验.试验结果表明,改进粒子群算法的收敛速度快于传统粒子群算法和基于模拟退火的粒子群算法,求解精度略逊于基于模拟退火的粒子群算法,但仍能保证求解精度.
통과대차량전향궤구적척촌화정위삼수진행우화,능유효감소차량전향궤구적실제운동궤적화이론운동궤적간오차,진이유효개선차량적조종성능화제고전향안전성.해문연구료전향제형궤구적공작원리급기대차량전향성능적영향,건립료전향제형궤구적비선성우화모형;연후인입월계검측함수개진전통입자군우화산법,병급출료구해전향제형궤구비선성우화모형적방법;편제료개진입자군산법적실현정서,병대3관불동차형적전향제형궤구진행료우화설계;최후선취3충불동지능산법분별대도악 GRX 전향제형궤구진행다조우화시험.시험결과표명,개진입자군산법적수렴속도쾌우전통입자군산법화기우모의퇴화적입자군산법,구해정도략손우기우모의퇴화적입자군산법,단잉능보증구해정도.
Errors exist between actual trajectories and theoretic trajectories of vehicle steering trapezoid mechanisms in the process of steering, which leads to shorter service life of wheels and worse vehicle handling, stability, and safety. Parameter optimization of steering trapezoid mechanisms can efficiently reduce these errors and improve the safety of these vehicles. The principal purpose of this paper is to develop an improved particle swarm optimization for an optimal design of steering trapezoid mechanisms. First, a nonlinear optimization model of the steering trapezoid mechanism is established by investigating how they work and how they influence the stability of vehicle maneuvering characteristics. The sum of the absolute value of difference between actual rotational angle of anterolateral steering wheel and theoretical rotational angle of anterolateral steering wheel is taken as the objective function of the nonlinear optimization model, while the bottom angle and steering arm length of steering trapezoid mechanisms are selected to be design variables. After that, an improved particle swarm optimization algorithm (IPSO) is proposed based on the traditional particle swarm optimization by introducing over-flow dealing functions to deal with complicated nonlinear constraints. The core idea of IPSO can be described as follows: complex nonlinear constraints are regarded as over-flow dealing functions, check whether over-flow dealing functions meet restricting condition at each iteration, if not, initialize design variables in proper ranges and then repeat the check, otherwise, go to next iteration. Finally, codes for IPSO are programmed and parameters of steering trapezoid mechanisms for three different models are optimized. To test the accuracy of the IPSO algorithm as proposed above, the nonlinear optimization problems for three different models (Nissan Duke, Patrol GRX and Patrol GL) are given;numerical results show that errors of the objective function's actual values and objective function's optimization values are less than 0.1%, which means that IPSO possesses high accuracy in solving nonlinear optimization problems, and that IPSO is a promising method for solving complicated constraint optimization problems. To verify effectiveness and efficiency of the IPSO algorithm, performance comparison experiments of three intelligent algorithms were analyzed. The problem of the steering trapezoid mechanism of Patrol GRX was carried out, with the Improved particle swarm optimization algorithm (IPSO), traditional particle swarm optimization (TPSO), and particle swarm algorithms based on simulated annealing (SA-PSO) being used as the optimizing parameters. For Patrol GR with the same initialization parameters and error percentage of objective function's actual values and objective function's optimization values (Percentage of Error), the minimum number of iterations to get the objective function's optimum solution, the minimum number of iterations to obtain the objective function's optimum solution (Min-iterationNum), the average number of iterations to obtain the objective function's optimum solution (Ave-iterationNum), and the total number of times to obtain the objective function's optimum solution (Total-Times) were selected as key performance comparison indicators of three intelligent algorithms of performance comparison experiments. The performance comparison experiment results indicates that the proposed new algorithm is superior to the particle swarm algorithm based on simulated annealing and traditional particle swarm optimization in fast convergence and small calculating quantity, but a little inferior to particle swarm algorithm based on simulated annealing in calculation accuracy in the process of optimization.