【作者】 李志杰；

【导师】 许镇琳；

【作者基本信息】 天津大学 ， 电力电子与电力传动， 2005， 硕士

【摘要】 电动助力转向系统具有机械和液压转向系统无可比拟的优势,是汽车助力转向机构发展的必然趋势。本课题在理论研究的基础上,吸取和借鉴国外研究的一些概念与思想,进行电动转向系统的深入研究和系统软件的具体开发,最终实现了台架试验性能优良的数字化EPS系统。本课题对电动助力转向系统的物理模型和数学模型进行了分析,并针对实际的应用情况简化数学模型,对电动助力转向系统的助力控制、回正控制和阻尼控制作了深入研究,提出了切实可行的实现方法。在Motorola DSP56800芯片的编程环境和本课题独立开发的上位机综合控制软件基础上,应用离散助力特性曲线成功的实现了EPS数字化系统。经实验台架测试,该系统克服了助力时抖动的问题,低速时助力电流最大达到17A,可以满足低速时转向的轻便性;高速时死区明显、助力电流小,可以满足高速稳定的控制要求。实验台架上打完方向盘后能够快速回轮,回正性能也基本达到了要求。该系统手感柔和、阻尼明显,可以满足实车的高低速要求,经简单改动就可以应用到实车中,具有非常高的实际应用价值。本课题在采用离散助力特性曲线实现助力特性后,进一步应用神经网络拟合助力特性曲线,经离线训练取得良好结果,并将离线训练结果成功应用到实际系统中。应用神经网络后死区和饱和区的设置更加灵活,可以改善低速的轻便性和高速的稳定性,实现了全车速范围助力,同时系统运行的快速性也没有受到影响。该实验将高层的理论真正应用到实际系统中并取得了传统控制难以达到的效果,验证了高层理论在实际系统中的实用性。嵌入式实时操作系统在嵌入式应用中具有很高的理论意义和研究价值,本课题在对μC/OS-II实时系统的理论研究的基础上,完成了该实时系统在MotorolaDSP56800编程环境下的移植,并成功实现了EPS数字系统在该实时系统下的正常运行。经转向台架实验验证,应用μC/OS-II实时操作系统后EPS系统的实时性能得到了进一步的提高。本课题通过对μC/OS-II实时系统的移植和具体应用,为进一步研究和应用该实时系统打下良好的基础。更多还原

【Abstract】 Electric power steering system is the inevitable developing direction forautomobile power steering, which is much superior to hydraulic power steeringsystem. The digital electric power steering system is accomplished, which meet withhigher stability, reliability, and excellent dynamic and stationary performance.Physical and mathematics model of electric power steering system is analyzedand studied, and three types of controls is presented, which is assist control, returncontrol and damp control. Further more, the feasible method of realization of threetypes of control is designed.Based on Motorola DSP56F805 as the central control unit and the integratedcontrol software on PC as the communication software between PC and ECU, thedigital electric power steering system is successfully accomplished, in which discretenonlinear assist character curve is used. Assist control is used at low speed and dampcontrol is used at high speed. Applying this digital system, steering is comfortable,steady and reliable in full speed range.Neural network is put forward in this subject to describe assist characteristic,while assist characteristic is realized based on discrete nonlinear assist character curve.Neural network is trained off line, and the training result accord with the target value.Making use of the training result, it is realized in the practical system. The experimentresult in experimentation platform which applied neural network assist characteristicis superior to the experiment result that used discrete nonlinear assist character curve.The ability of advanced theory in the practical system is perfectly validated in thissubject.Finally, the theory of MicroC/OS-II is analyzed and studied in this paper, whichis an embedded real time operating system(RTOS). It is perfectly transplanted toMotorola DSP56F805, and the digital electric power steering system is able to runsuccessfully in this real time operating system. The return ability using RTOS issuperior to the ability without it.更多还原