HUANG Bifei，FENG Zhimin*，ZHANG Gang，LI Hongwei（.Maritime College，Ningbo Universty，Ningbo Zhejiang 35，China；.Ningbo Shangong Center of Structural Monitoring and Control Engineering Co，Ltd，Ningbo Zhejiang 3500，China）



Research on Algorithm of Piezo-Film Vehicle Weigh-in-Motion System*

HUANG Bifei1，FENG Zhimin1*，ZHANG Gang1，LI Hongwei2
（1.Maritime College，Ningbo Universty，Ningbo Zhejiang 315211，China；2.Ningbo Shangong Center of Structural Monitoring and Control Engineering Co，Ltd，Ningbo Zhejiang 315100，China）

A vehicle weigh-in-motion system based on piezo-film sensors is used，which has been tested by three kinds of vehicles at different ambient temperature and speed.The singular spectrum algorithm is adopted to reduce the noise in testing axle-load signal.A novel reconstructed order selecting method is presented，which simplifies the judgment criteria for singular spectrum algorithm.The impact of ambient temperature and speed on weighing result is analyzed in detail.The axle-load signal area is compensated in the improved weighting algorithm.The test plat⁃form of piezo-film vehicle weigh-in-motion system based on double sensors is established.The test shows that the av⁃erage vehicle-weight error measured by double sensors is 22.4%lower than the single sensor.The proposed weighin-motion system meets the practical application requirement at speed within 50 km/h and the average error less than 5%，justifying the validity of the proposed SSA algorithm.

piezo-film；weigh-in-motion；ssa algorithm；double sensors；vehicle-weight calculation

随着我国公路交通运输业的快速发展，运输车辆超限超载现象愈加普遍，这严重影响了公路桥梁的寿命，也给交通安全带来了巨大危害。车辆动态称重系统的发展不仅可有效治理车辆超限超载，并且随着动态称重技术的不断提高，将逐步取代传统的人工及静态称重收费模式。目前，称重精度差仍是制约动态称重系统发展的主要因素。称重精度的影响因素是测得的轴重信号中混有许多干扰信号，包括车速、车辆振动、轮胎驱动力、路面激励以及系统自身产生的测量干扰等［1］，如何利用算法从干扰信号中提取真实的轴重信号，并对称重数据进行处理以求得车重值，这是动态称重研究的核心问题之一。徐志玲［2］提出了一种改进的算术平均测量法，即以信号中相对平稳信号区间的平均值计算车重，以减少上、下台冲击对称重结果的影响，但该方法并不适合于轴重称及高速称重。潘昊［3］采用改进的神经网络算法处理采集信号，误差控制在5%以内，但要获得大量样本是十分困难的。周志峰［4］采用相关系数法判断虚假模态，端点延拓法抑制端点效应，适合车速小于20 km/h时，最大轴重误差为4.34%。潘若禹［5］采用SSA算法较传统算法能有效提高车辆称重精度，但其在重构阶次的选择上计算量较大。

因SSA算法不受波形信号正弦性的假定约束，其对信号的识别和描述采用时域性的频域特征分析方法，可更好的对时序信号进行去噪和特征提取处理［6］。……