Detection and Monitoring of Bottom-Up Cracks in Road Pavement Using a Machine-Learning Approach

Abstract

The current methods that aim at monitoring the structural health status (SHS) of road pavements allow detecting surface defects and failures. This notwithstanding, there is a lack of methods and systems that are able to identify concealed cracks (particularly, bottom-up cracks) and monitor their growth over time. For this reason, the objective of this study is to set up a supervised machine learning (ML)-based method for the identification and classification of the SHS of a differently cracked road pavement based on its vibro-acoustic signature. The method aims at collecting these signatures (using acoustic-sensors, located at the roadside) and classifying the pavement’s SHS through ML models. Different ML classifiers (i.e., multilayer perceptron, MLP, convolutional neural network, CNN, random forest classifier, RFC, and support vector classifier, SVC) were used and compared. Results show the possibility of associating with great accuracy (i.e., MLP = 91.8%, CNN = 95.6%, RFC = 91.0%, and SVC = 99.1%) a specific vibro-acoustic signature to a differently cracked road pavement. These results are encouraging and represent the bases for the application of the proposed method in real contexts, such as monitoring roads and bridges using wireless sensor networks, which is the target of future studies.

An IoT System for Social Distancing and Emergency Management in Smart Cities Using Multi-Sensor Data

“…Note that, to obtain a more detailed environmental and structural monitoring, the IoT board was also equipped with a smoke sensor (to detect carbon monoxide, liquid petroleum gas, and smoke), a flame sensor, and an additional microphone. The latter was added to receive the vibro-acoustic response [64][65][66] of the structure on which the units were installed. In more detail, the additional microphone was isolated from the airborne noise through a cover (inside the box) and isolating material (between the box and the structure), and was able to receive the acoustic signals (a.k.a., vibro-acoustic signature of the structure) that travelled into the structure.…”

“…In more detail, the additional microphone was isolated from the airborne noise through a cover (inside the box) and isolating material (between the box and the structure), and was able to receive the acoustic signals (a.k.a., vibro-acoustic signature of the structure) that travelled into the structure. Proper analyses of these acoustic signals (i.e., feature extraction, multidomain analysis for cracks identification and monitoring, structural health status classification; [64][65][66]) allow one to carry out the Structural Health Monitoring (SHM) of the structures where the sensing node is attached, which is also applicable in the case of road monitoring, where the use of self-powered and smart sensors is envisioned [67].…”

A Real-Time Decision Platform for the Management of Structures and Infrastructures

“…As an example, one of the CNNs mentioned above is presented in this paper (see Figure 8). This CNN was specifically designed to classify the features extracted from the acoustic signals gathered from the road pavement and to recognize variation in the structural health status of the road [44]. (see, e.g., [16]), but in this application they were used to (i) take as an input the feature sets extracted;…”

“…As an example, one of the CNNs mentioned above is presented in this paper (see Figure 8). This CNN was specifically designed to classify the features extracted from the acoustic signals gathered from the road pavement and to recognize variation in the structural health status of the road [44]. The abovementioned CNN was built while paying attention to the recommendations and the recurrent problems described in the literature [45,46], and it has the following characteristics: (1) two fully connected layers (with 70 and 30 hidden nodes, respectively), which carry out pattern recognition using the activation function ReLu (relu(x) := max(0, x), i.e., this function, f (z), is zero The abovementioned CNN was built while paying attention to the recommendations and the recurrent problems described in the literature [45,46], and it has the following characteristics: (1) two fully connected layers (with 70 and 30 hidden nodes, respectively), which carry out pattern recognition using the activation function ReLu (relu(x): = max(0, x), i.e., this function, f (z), is zero when z < 0, and it is equal to z when z 0); (2) one convolutional layer, which automatically extracts additional features from the input); (3) one pooling layer, which carries out the average pooling of the features extracted while applying the valid padding; (4) Adadelta Optimized was selected as the optimizer function (for the adjustment of weights and biases); (5) the activation function Softmax cross entropy measures the probability error in discrete classification tasks, and confusion matrices were used to show and analyze the results of the classification.…”

Sensor-based pavement diagnostic using acoustic signature for moduli estimation

“…1. embedded [5][6][7] and non-embedded sensor-based systems [8,9]; 2. mobile [10][11][12] and stationary systems [13][14]; 3. wireless [15,16], wired [17,18] and self-powered systems [19,20]; 4. traditional [21,22] and smart data management [23][24][25]; Despite the promising advantages of the sensor-based solutions, and the growing need of infrastructures in the Internet of Things (IoT) world, it must be underlined that these solutions are sometimes in an early stage of investigation, and that there is a lack of applications in real contexts. Based on the above, the main objective of the presented study is to validate the results of an innovative road pavement monitoring solution with data derived using traditional methods (i.e., GPR, FWD).…”