登录 EN

添加临时用户

汽车驾驶员舒适位置优化及自适应干预方法研究

Research on the Car Seat Optimal Adjustment and Adaptive Interventions

作者:王波
  • 学号
    2008******
  • 学位
    博士
  • 电子邮箱
    thu******com
  • 答辩日期
    2013.06.07
  • 导师
    成波
  • 学科名
    机械工程
  • 页码
    137
  • 保密级别
    公开
  • 培养单位
    015 汽车系
  • 中文关键词
    车辆,座椅,舒适性,优化,干预
  • 英文关键词
    Car,seat,comfort,optimization,intervention

摘要

舒适性是汽车的主要性能之一,座椅系统是影响驾驶员舒适性的最主要因素。目前的座椅系统设计尚未充分解决舒适性与安全性的耦合问题及驾驶员差异性问题,也未考虑到驾驶员生理状态迁移对乘坐环境的不同要求,因而容易让驾驶员感到不适,引起疲劳甚至病变。针对上述问题,以乘用车为研究对象,本文研究提取了体现个体差异性的舒适性评价指标,提出了考虑舒适性与安全性耦合的双层优化结构,设计了面向多峰问题的狮群进化算法,进一步根据驾驶员状态迁移机制设计了自适应干预方法,课题研究为解决驾驶员在驾驶时间轴上的舒适优化及干预问题奠定了基础。针对中国驾驶员群体,研究提取了体现个体差异性的舒适性评价指标。对驾驶员舒适位置的研究发现,身高、性别对舒适位置有显著影响,且中西方人群间存在明显差异性。进一步研究显示,与座椅系统位置及姿势关节角度相关性最显著的多是表征人体各部分比例信息的局部人体参数,坐垫前后位置等四个参数与人体参数相关性显著,进而提取了座椅系统位置及姿势关节角两类舒适性评价指标,同时研究建立了基于人体参数主成分分析的评价指标回归预测模型。针对座椅系统优化这一多目标优化问题,提出了一种优先眼点定位的双层优化结构。首先通过坐垫前后位置等参数定位眼点,然后采用权函数法以四个姿势关节角为指标优化其他座椅系统参数。该方法能够在准确优化眼点位置的基础上,保证H点及全身姿势的优化精度。验证结果显示,该方法能够为驾驶员优化出舒适位置,且优于现有的其他方法。为解决以座椅系统优化为例的多峰优化问题,本文进一步设计了一种内外交互式狮群进化算法。依据狮群激烈个体竞争、频繁群体更新特性设计的狮群进化算法具有间断进化特征,在多峰优化问题上的求解精度显著优于遗传算法等多种算法,且具有较高的计算效率、鲁棒性、抗居中倾向性。根据长时间驾驶过程中,驾驶员“抗争运动”模式以及“顺从运动”模式的内在机制,提出了基于座椅系统几何形态调节及表面特性调节的两种自适应姿势干预方法。前者能够改善驾驶姿势,调节脊柱形态;后者能够改善腰部支撑特性,减轻腰部压力,从而有效延缓疲劳进程。

Drivers’ sitting comfort plays an important role in the perception of a vehicle’s overall quality, and seat system is the most important factor that affects the sitting comfort. However, the interconnection of comfort and safety, and individual difference are still problems faced in the existing seat system design, and moreover the different demands for seat system due to drivers’ status transfer have not been considered up to now. Therefore, drivers may easily become tired when driving. To address these issues in sedans, the evaluation criteria that can reflect individual difference were studied firstly in this thesis, and then a double-layer optimization framework was proposed to solve the interconnection of comfort and safety. A novel lion pride optimization algorithm was designed for multimodal optimization problems. In addition, drivers’ postures may change with the increase of driving time. Two adaptive intervention methods corresponding to the posture movements were proposed to release the discomfort for drivers during overall driving process.The evaluation criteria for Chinese drivers were studied in this thesis. The results indicate that stature and weight have significant effects on drivers’ preferred seat positions, and there is significant difference between Chinese and Western drivers. The further study indicates that the local dimension variables have the most significant correlation with preferred postures or seat positions. Two seat position variables and two posture angles were found to have significant correlation with body dimensions, and the evaluation criteria were obtained from seat position variables and posture angles. The multiple linear regression models were built to predict the evaluation criteria by using the factors obtained from the principal component analysis of above-mentioned dimension variables.A double-layer optimization framework with priority on eyepoint-positioning was proposed to fulfill seat optimal adjustment, i.e. first to optimize the eyepoint by predicting the cushion fore-aft position, cushion rear height and backrest angle, and then to optimize other parameters by four posture angles. This framework outperforms other optimization frameworks in terms of both eyepoint-positioning and other components’ positioning, and it has the merit of more accurately predicting comfort positions for drivers.The car seat optimal adjustment is a multimodal optimization problem. This thesis designed a novel optimization algorithm, Lion Pride Optimizer (LPO), to solve this problem. The LPO was mainly designed according to the sharp competition of individuals and frequent pride updating of the lion pride. The evolution process of the LPO is discontinuous, and there is strong interaction inside and outside of the pride. The LPO outperforms other seven algorithms on commonly used multimodal functions. Simulation indicates that the LPO has strong robustness, and no significant central tendency was found for LPO.As the increase of driving time, drivers have two kinds of typical posture movements. This paper proposed two adaptive interventions according to their intrinsic difference: one is to adjust the geometry of the seat system; the other is to adjust its surface. The former improves the driving posture and adjusts the spine shape to improve driving comfort. The latter improves the support to the lower back and releases its pressure to reduce driving discomfort.