遗传算法在汽车惯性参数估计中的应用

需积分: 10 101 浏览量更新于2024-07-23 收藏 503KB PDF 举报

"车辆惯性参数的估计方法" 在车辆工程领域，参数识别与估计是一项至关重要的技术，特别是在车辆动力学研究中。本资源详细介绍了如何利用遗传算法来估计汽车的惯性参数，这是一种非常有效的参数估计方法。文章由Matthew Rozyna和Nong Zhang共同撰写，发表在2010年的《Vehicle System Dynamics: International Journal of Vehicle Mechanics and Mobility》上，该期刊由Taylor & Francis出版，是车辆力学和移动性研究的重要参考资料。遗传算法是一种基于生物进化原理的优化方法，它模拟了自然选择和遗传过程，用于在复杂问题中寻找最优解。在汽车参数估计中，遗传算法可以处理多目标优化问题，如车辆的质量、转动惯量和重心位置等惯性参数的识别。这些参数对于理解和模拟车辆的行为至关重要，它们直接影响到车辆的动态响应，包括操控性、稳定性和安全性。文章中提到，估计车辆的惯性参数通常是一项挑战，因为这需要处理大量的数据和复杂的非线性关系。传统的估计方法可能效率低下，而遗传算法则提供了一种更高效且鲁棒的解决方案。通过将车辆的实际运行数据（如加速度、速度和转向角等）作为输入，遗传算法可以逐步优化参数设置，从而得到最接近实际值的估计结果。车辆的惯性参数不仅影响车辆的动力学模型，还对控制系统的设计有着直接影响。例如，在自动驾驶系统或高级驾驶员辅助系统（ADAS）的开发中，精确的车辆模型参数是必不可少的。此外，这些参数还可以用于车辆性能的改进，包括悬挂系统的设计、轮胎选择以及电子稳定程序的校准等。 "车辆惯性参数的估计方法"这篇论文提供了关于如何运用遗传算法进行参数识别的详细步骤，这对于车辆工程领域的研究人员和工程师来说是一个宝贵的工具，有助于提高车辆性能分析和控制策略设计的准确性。通过深入理解和应用这些方法，可以进一步提升车辆的安全性、舒适性和能效。

Vehicle System Dynamics 549

the double beneﬁt of decreasing cost and reducing the estimator’s reliance on measurement of

the input forces. Another beneﬁt of the proposed methodology is once the inertial parameters

have beenmeasuredan estimate of thenon-lineartyre forces is possible[42],further increasing

the accuracy of information provided to active vehicle controllers.

Theproposedmethodisbasedonmodalparameterestimationusingthefree-decayresponses

of the vehicle, estimation of the system characteristic matrix using state space methods, least

squares analysis, combined withasimpliﬁedvehicle model. Knowledge of the vehicle’s equiv-

alent linear spring stiffness is also required. This is a valid assumption in the opinion of the

authors; this information is well known by the vehicle manufacturers and can be measured,

is reasonably linear for vehicles such as passenger cars, SUVs and pickups and has much

less parameter variation than damping coefﬁcients, aerodynamic drag, road friction forces or

varying gravity forces due to changes in road grade.

2.1. Measurement sensors

The number, location and type of sensors mounted onto the vehicle’s sprung chassis mass

will determine the maximum degrees of freedom, what parameters can be measured and the

mathematical basis for the estimation model. An overview of sensors and sensor location is

presented in this section.

The ﬁrst step is to choose the type of sensors, linear or angular, to be mounted onto the

vehicle chassis. For simplicity and proof of concept, the examples presented in this paper

uses three accelerometers mounted to the vehicle chassis, allowing for estimates of the sprung

mass, pitch and roll mass moments of inertia as well as lateral and longitudinal centre of

gravity locations of the vehicle.With this conﬁguration, measurement of the yaw inertia is not

strictly possible; however, it is well documented in the literature [24,25] that the yaw and pitch

inertia values are within a few percentage points of each other, and therefore a measurement

of the pitch inertia will yield a reasonable estimate of the current yaw mass moment of inertia.

Other sensors such as gyroscopes or Linear Variable Differential Transformers (LVDTs) can

also be used.

The location of the sensors is also very important. If only accelerometers are used then they

should be spaced as far apart as possible to maximise the signal-to-noise ratio of the phase and

amplitude information. Another important consideration for positioning the accelerometers

is related to the wheelbase ﬁltering effect. Most frequency components that excite vehicle

vibrations come from the road surface, but the geometric properties of the vehicle can result in

an ampliﬁcation or attenuation of certain frequencies due to the front and rear axles following

the same proﬁle delayed in time. This phenomenon is known as wheelbase ﬁltering and is

dependent on the speed of the vehicle and the wheelbase. Best [43] shows that accelerometers

placed near the centre of the wheelbase will display null points and false resonance peaks.

Accelerometers placed further away from the centre of gravity position, for example above the

wheelstations,will detectboththe bounceand pitchmodesand theeffect ofwheelbaseﬁltering

is minimised.Therefore, in the opinionofthe authors, the optimum locationforaccelerometers

to minimise wheelbase ﬁltering effects and to maximise the signal-to-noise ratio for phase

information is at the outer edges of the vehicle. For simplicity, in this paper the sensors are

placed directly above the unsprung masses, but this does not preclude other locations.

2.2. Extraction of free-decay response

The method relies on the successful extraction of the dynamic system’s free-decay response.

The method for extracting the free-decay response is dependent on the type of input to which

Downloaded by [Nanjing University of Aeronautics & Astronautics] at 08:05 13 April 2014

剩余19页未读，继续阅读

zhaofenglailema

粉丝: 0
资源: 1

遗传算法在汽车惯性参数估计中的应用

模式识别非参数估计方法PPT

模式识别_参数估计

基于改进聚类的输电线路参数多智能体识别与估计

扩频信号参数估计-信号检测-识别-参数估计

手写数字概率模型参数估计与识别（贝叶斯估计）.rar

脉压雷达信号的识别与参数估计算法研究

数字通信信号的自动识别与参数估计研究

模式识别课件解析：参数估计与非参数估计

北京邮电大学模式识别课件：参数估计与非参数估计详解

模式识别：参数估计与统计推断

最新资源