Automatic Crack Classification by Exploiting Statistical Event Descriptors for Deep Learning

Abstract

In modern building infrastructures, the chance to devise adaptive and unsupervised data-driven structural health monitoring (SHM) systems is gaining in popularity. This is due to the large availability of big data from low-cost sensors with communication capabilities and advanced modeling tools such as deep learning. A promising method suitable for smart SHM is the analysis of acoustic emissions (AEs), i.e., ultrasonic waves generated by internal ruptures of the concrete when it is stressed. The advantage in respect to traditional ultrasonic measurement methods is the absence of the emitter and the suitability to implement continuous monitoring. The main purpose of this paper is to combine deep neural networks with bidirectional long short term memory and advanced statistical analysis involving instantaneous frequency and spectral kurtosis to develop an accurate classification tool for tensile, shear and mixed modes originated from AE events (cracks). We investigated effective event descriptors to capture the unique characteristics from the different types of modes. Tests on experimental results confirm that this method achieves promising classification among different crack events and can impact on the design of the future of SHM technologies. This approach is effective to classify incipient damages with 92% of accuracy, which is advantageous to plan maintenance.

Detection and Classification of Crack for Heritage Building

“…For crack detection, a metaheuristic edge detection model is proposed in [30], [31]. In the paper, a comparison of the performance of the Roberts, Prewitt, Canny, and Sobel approaches is conducted.…”

Deep learning-based damage estimation in concrete structures using acoustic emission data

“…Some of them are Deep Neural Network (DNN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Deep Belief Network (DBN), Auto-encoders, and Generative Adversarial Networks (GANs), etc. [3]. In this study, concerning the classification of damage states, DNN based architecture using a multilayer perceptron (MLP) neural network was used to classify the damage states of the reinforced concrete beam which is discussed in section 4.…”

“…Aside from material design, the structural integrity of reinforced concrete structures is continuously monitored using AE, but one of the most potent approaches for identifying damage in reinforced concrete structures is using machine learning and deep learning. In the present era, deep learning has attracted a lot of interest among researchers since it was first introduced in civil engineering and researchers are focusing on developing automated monitoring techniques for structural responses [3]. Unlike the traditional parameter-based models, data-driven models offer bottom-up solutions that include the identification of damage/cracks and life estimation of structures [4].…”

“…In different fields, crack detection studies are carried out by using multiple modules related to deep learning [5], [6]. In AE, which is used for structural health monitoring in both infrastructure and superstructures is highly accurate in damage estimation and can be used with deep learning techniques [7], [3]. Although AE characteristic parameters can be used to identify damage modes in concrete structures utilizing a parameter-based approach, a data-driven approach shouldn't be ignored as well [2].…”

Low-Cost Smart Living Environment Monitoring Systems from a Metrological Point of View

“…For every Smart Environmental Monitoring Systems (SEMS), in order to work properly, they must be able to accomplish three main actions: (i) acquire the data from the sensors, (ii) processing it to obtain the desired measurement information, and (iii) communicating the measurement information to the user (human or automatic systems) to take the required actions. SEMSs allow for a continuous and real-time monitoring of the potentially hazardous entity of the environment [5], as well as the ability to rise an alarm (or take countermeasures if possible) so that end users can avoid being in dangerous or uncomfortable environments [6].…”