基于四阶段法的大型车超速辨识模型及其优化

Large vehicle overspeed identification model based on four-stage method and its optimization

为了准确判定垂直侧碰电动自行车的大型车是否超速,结合骑车人、电动自行车的抛距和大型车碰撞前后制动距离,将碰撞过程逆向还原成四个阶段,提出基于四阶段法的大型车超速辨识模型(OILV)。分析OILV误差产生的原因,设置大型车碰撞前制动距离调节因子,并利用BP神经网络确定调节因子值,提出基于BP神经网络的大型车超速辨识模型(BP-OILV)。验证结果表明,相比于OILV,BP-OILV最高可将单个验证样本辨识误差由3.17%减小至0.01%,可将多个样本的误差均值由1.99%减小至0.29%,误差标准差由0.011减小至0.002,可精确稳定地辨识大型车是否超速。

In order to accurately determine whether the large vehicle is overspeed in vertical side collision with the electronic bicycle,the collision process is reversely recovered into four stages to propose the four-stage method based overspeed identification model for large vehicles(OILV)by combining the thrown distance of rider and electronic bicycle,and large vehicle′s braking distance before and after collision.The reason causing OILV error is analyzed,and the regulatory factor of the large vehicle′s braking distance before collision is set.The BP neural network is utilized to determine the regulatory factor,and the optimized overspeed identification model for large vehicle based on BP neural network(BP-OILV)is proposed.The verification results indicate that,in comparison with OILV,the BP-OILV can reduce the single sample′s identification error from 3.17%to 0.01%,reduce the multiple samples′average error from 1.99%to 0.29%,and reduce the error deviation from 0.011 to 0.002,which can identify whether the large vehicle is overspeed accurately and stably.