一种基于智能手机的道路质量检测方法

A Road Quality Detection Method Based on Smartphone

目前,道路质量检测主要通过专业的道路质量检测车完成,精度高,但成本昂贵、检测周期长、普适性差。智能手机内置了大量传感器,能够收集用户及其周围的数据信息,具有成本低廉、普适性好等优点,在许多道路检测研究中被广泛应用。因此,提出了一种基于智能手机的道路质量检测方法。首先,以道路质量检测车为承载平台,利用智能手机采集路面的加速度和角速度等数据,利用检测车采集路面质量的真值数据;然后,提取手机加速度和角速度等数据的特征值并标注真值标签;最后,使用机器学习方法对道路质量进行分类并构建合适的模型。以湖北省武汉市和四川省达州市共计173 km的道路的数据为例,利用机器学习方法实现了道路质量的有效分类,其中,每隔10 m记录一次国际平整度指数(international roughness index,IRI)时,支持向量机模型对城市道路和高速公路数据的分类平均准确率分别达到89.4%和88.9%。结果表明,该方法在不同的道路类别下均能有效检测道路质量。

At present,road quality detection is mainly done by professional road quality inspection vehicles,which is of high accuracy but costly,and has a long detection period and poor universality,while smartphones,having a large number of built-in sensors that can collect data information from users and their surroundings,are widely used in many road detection studies because of their advantages of low cost and good universality.Therefore,we propose a smartphone-based road quality detection method.Firstly,a road quality inspection vehicle is used as the carrier platform to collect data such as acceleration and angular velocity of the road surface with the smartphone,and the inspection vehicle is used to collect the true value data of the road quality.Then,the feature values of the data such as acceleration and angular velocity of the mobile phone are extracted and labelled with the true value labels.Finally,machine learning methods are used to classify the road quality and construct a suitable model.Taking the data of the road with a total length of 173 km in Wuhan,Hubei Province and Dazhou,Sichuan Province as an example,we use machine learning methods to classify the road quality.The average accuracy of support vector machine model for urban road and highway data classification is 89.4%and 88.9%,respectively,when the international roughness index(IRI)is recorded every 10 meters.The results show that the proposed method is effective in detecting road quality under different road categories.