【作者】 郭龙；

【导师】 王洪礼；

【作者基本信息】 天津大学 ， 工程力学， 2009， 博士

【摘要】 本文针对当前汽车底盘测功机测试实验中存在的参数标定偶然性大、过程控制缺失、测试效率低下、试验成本高,以及数据并发式输入空间的非线性因素导致的系统偏差较大等传统控制模式无法解决的问题,将人工智能领域的最新成果──智能主体技术应用于汽车底盘测功机系统的智能化建设中,构建了基于π演算的智能主体模型,并重点研究了系统中的智能控制、阻力模拟等问题,从而提升了汽车底盘测功机的制造水平和多工况排放测试的实验水平,具体如下:首先,针对汽车底盘室内检测试验中阻力模拟试验的“偶然、多次、全过程人工操作”等特点,利用智能主体控制技术,首次提出了适合于汽车底盘测功机排放测试的π演算模型,从而保证了在不同转速工况下的精度。由于汽车悬架与底盘测功机系统是一个由较多非线性因素共同作用的复杂系统,因此借鉴免疫反馈原理,结合积分控制的规律,提出了一种模糊免疫PID控制方法,并利用免疫进化算法进行参数优化设计。仿真结果表明,它优于常规的PID控制器,对测功机恒扭矩和恒转速的控制取得了良好的控制效果。利用ADAMS建立了轮胎道路模型和测功机道路行驶阻力的模型,并通过调用上位机阻力计算模型,实时控制测功机转鼓转矩的加载,从而实现了车辆道路行驶阻力的精确模拟。针对传统底盘测功机在条件发生变化时,无法自动进行参数调节的局限性,采用核PCA分析方法,对数据记录集进行特征提取,这样就能较好地满足输入空间的非线性特性,以适应由于测量值波动而引起的参数变化。本文还提出了一种适合于汽车底盘测功机阻力模拟的高精度快速自补偿法,该方法可达到较快的响应时间和较高的模拟精度。最后,利用面向对象的技术构造了测功机台架与道路实体模型,采用VS.NET2003构架的WEB服务为基础平台,集成了ADAMS、VRML、JAVA以及SQL数据访问技术,建立了多路并发的消息同步通讯机制,从而实现了MCG1000型汽车底盘测功机VR三维同步仿真平台。更多还原

【Abstract】 There are many problems in process of detction nowadays, such as random parameter confirms, the loss of process control, the low efficiency of detection, the high experimental cost, nonlinear influence of input space. This article apply the most advanced technology of artificial intelligence—intelligent agent system to intelligence realizing of automobile chassis dynamometer, the model of detection system is created byπcalculus model. It is the most important problem to research the intelligent control and resistance simulation of dynamic detection system.The first, in order to solve the problems in automobile chassis dynamometer detection, we put forword aπcalculus model for emission detection firstly, the control accuracy can be improved in multi work condition.Second, In the process of automotive suspension control, nonlinear characteristics could not be expressed accurately after linearization transaction. With digital approximation, nonlinear influence factor had been put into system, such as spring and damp. A quarter nonlinear dynamic model of automotive suspension was get. According to the immune feedback principle and the nonlinear change law of the integral gain in the response process of the system, a fuzzy immune PID(Proportion Integral Differentiation) control method was proposed. The simulation results show the feasibility and effectiveness of the method in nonlinear automotive suspension control system and the method is superior to PID method on the response characteristic. Ride comfort and maneuvering performance are improved greatly.Third, According to the otherness analysis between road test and bench test, resistance simulation mode of vehicle chassis dynamometer was constructed by ADAMS, With visual programming technique, resistance simulation control system based on industrial computer was completed. Torque on runner boss can be given accurately and load-on as soon as possible by using the mode of resistance calculation, also reach-ability, security and stability of the system can be seen from the result of imitation research.Then, as we know, parameters will become weakly adjust when the test condition of dynamometer detection changes, it is means that input space is an nonlinear distinction, so, by use of prime PCA method, we got the distinction of input data, then to analyze the influence quantity from fluctuation of measured value. According to parameter regression analysis, the response and the accuracy are all improved.At last, taking MCG1000 automobile chassis dynamometer as a subject investigated, bench and road entity model have been created by the technique of OO, intelligent detection platform is based on VS.NET2003 with WEB service. In the web platform, we take the technique integration of ADAMS, VRML, JAVA and SQL. Finally, the tridimenstional synchronous simulation VR system of MCG1000 dynamonmeter was realized.更多还原