Sensor technologies empower Industry 4.0 by enabling integration of in-field and real-time raw data into digital twins. However, sensors might be unreliable due to inherent issues and/or environmental conditions. This paper aims at detecting anomalies in measurements from sensors, identifying the faulty ones and accommodating them with appropriate estimated data, thus paving the way to reliable digital twins. More specifically, we propose a general machine-learning-based architecture for sensor validation built upon a series of neural-network estimators and a classifier. Estimators correspond to virtual sensors of all unreliable sensors (to reconstruct normal behaviour and replace the isolated faulty sensor within the system), whereas the classifier is used for detection and isolation tasks. A comprehensive statistical analysis on three different real-world data-sets is conducted and the performance of the proposed architecture validated under hard and soft synthetically-generated faults.
Sensor technologies empower Industry 4.0 by enabling integration of in-field and real-time raw data into digital twins. However, sensors might be unreliable due to inherent issues and/or environmental conditions. This paper aims at detecting anomalies instantaneously in measurements from sensors, identifying the faulty ones and accommodating them with appropriate estimated data, thus paving the way to reliable digital twins. More specifically, a real-time general machine-learning-based architecture for sensor validation is proposed, built upon a series of neural-network estimators and a classifier. Estimators correspond to virtual sensors of all unreliable sensors (to reconstruct normal behaviour and replace the isolated faulty sensor within the system), whereas the classifier is used for detection and isolation tasks. A comprehensive statistical analysis on three different real-world data-sets is conducted and the performance of the proposed architecture is validated under hard and soft synthetically-generated faults.
The rapid adoption of Internet-of-Things (IoT) and digital twins (DTs) technologies within industrial environments has highlighted diverse critical issues related to safety and security. Sensor failure is one of the major threats compromising DTs operations. In this paper, for the first time, we address the problem of sensor fault detection, isolation and accommodation (SFDIA) in large-size networked systems. Current available machine-learning solutions are either based on shallow networks unable to capture complex features from input graph data or on deep networks with overshooting complexity in the case of large number of sensors. To overcome these challenges, we propose a new framework for sensor validation based on a deep recurrent graph convolutional architecture which jointly learns a graph structure and models spatio-temporal inter-dependencies. More specifically, the proposed two-block architecture (i) constructs the virtual sensors in the first block to refurbish anomalous (i.e. faulty) behaviour of unreliable sensors and to accommodate the isolated faulty sensors and (ii) performs the detection and isolation tasks in the second block by means of a classifier. Extensive analysis on two publicly-available datasets demonstrates the superiority of the proposed architecture over existing state-of-the-art solutions.
1 December 2022 in IEEE Internet of Things Magazine
Carbon capture and storage (CCS) is critical for climate-change policies and strategies targeting global warming within the Paris Agreement. The overarching technological requirements are well described in the strategic plans, yet several barriers exist to the technology's wide-spread deployment, including improved cost-effectiveness and enhanced process integration. For the safe and reliable operation of large-scale CCS systems, the development of effective Internet of things (IoT)-based monitoring tools to ensure flow assurance of CO 2 throughout the CCS value chain is crucial. Further, reliable sensor measurements related to different transport parameters such as temperature, pressure, and flow across the process are essential to develop these methods. However, sensors are prone to errors due to inherent issues or environmental conditions which result in performance degradation of the overall monitoring system. Developing techniques for detecting anomalies in the measurements, identifying the faulty sensors and accommodating them with appropriately estimated data is one of the paramount requirements for the reliable operation of the CCS systems. In such context, the present article provides an overview of CCS's monitoring and control requirements, emphasizing data-fusion synergies. This work investigates the state-of-the-art methods for sensor validation and proposes a roadmap to further deploy metering technologies for industrial needs.
In this paper, we propose a novel sensor validation architecture, which performs sensor fault detection, isolation and accommodation (SFDIA). More specifically, a machine-learning based architecture is presented to detect faults in sensors measurements within the system, identify the faulty ones and replace them with estimated values. In our proposed architecture, sensor estimators based on neural networks are constructed for each sensor node in order to accommodate faulty measurements along with a classifier to determine the failure detection and isolation. Finally, numerical results are presented to confirm the effectiveness of the proposed architecture on a publicly-available air quality (AQ) chemical multi-sensor data-set.
The development of Digital Twins (DTs) has bloomed significantly in last years and related use cases are now pervading several application domains. DTs are built upon Internet of Things (IoT) and Industrial IoT platforms and critically rely on the availability of reliable sensor data. To this aim, in this article, we propose a sensor fault detection, isolation and accommodation (SFDIA) architecture based on machine-learning methodologies. Specifically, our architecture exploits the available spatio-temporal correlation in the sensory data in order to detect, isolate and accommodate faulty data via a bank of estimators, a bank of predictors and one classifier, all implemented via multi-layer perceptrons (MLPs). Faulty data are detected and isolated using the classifier, while isolated sensors are accommodated using the estimators. Performance evaluation confirms the effectiveness of the proposed SFDIA architecture to detect, isolate and accommodate faulty data injected into a (real) wireless sensor network (WSN) dataset.