Sensor Failures Research Articles

Multimodal Deep Neural Network-Based Sensor Data Anomaly Diagnosis Method for Structural Health Monitoring

Due to sensor failure, noise interference and other factors, the data collected in the structural health monitoring (SHM) system will show a variety of abnormal patterns, which will bring great uncertainty to the structural safety assessment. This paper proposes an automatic data anomaly diagnosis method for SHM based on a multimodal deep neural network. In order to improve the detection accuracy, both two-dimensional and one-dimensional features of the sensor data are fused in the multimodal deep neural network. The network consists of two convolutional neural network (CNN) channels, one a 2D-CNN channel for extracting time–frequency features of sensor data and the other a 1D-CNN channel for extracting raw one-dimensional features of sensor data. After convolution and pooling operations for the sensor data by the 2D channel and 1D channel separately, the two types of extracted features are flattened into one-dimensional vectors and concatenated at the concatenation layer. The concatenated vector is then fed into fully connected layers for final SHM data anomaly classification. In order to evaluate the reliability of the proposed method, the monitored data lasting for one month of a long-span cable-stayed bridge were used for training, validation, and testing. Six types of training conditions (missing, minor, outlier, over-range oscillation, trend, and drift) are studied and analyzed to address the issue of imbalanced training data. With an accuracy rate of 95.10%, the optimal model demonstrates the effectiveness and capability of the proposed method. The proposed method shows a promising future as a reliable AI-assisted digital tool for safety assessment in structural health monitoring systems.

Model predictive control of nonlinear processes using neural ordinary differential equation models

Neural Ordinary Differential Equation (NODE) is a recently proposed family of deep learning models that can perform a continuous approximation of a linear/nonlinear dynamic system using time-series data by integrating the neural network model with classical ordinary differential equation solvers. Modeling nonlinear dynamic processes using data has historically been a critical challenge in the field of chemical engineering research. With the development of computer science and neural network technology, recurrent neural networks (RNN) have become a popular black-box approach to accomplish this task and have been utilized to design model predictive control (MPC) systems. However, as a discrete-time approximation model, RNN requires strictly uniform step sequence data to operate, which makes it less robust to an irregular sampling scenario, such as missing data points during operation due to sensor failure or other types of random errors. This paper aims to develop a NODE model and construct an MPC based on this novel continuous-time neural network model. In this context, closed-loop stability is established and robustness to noise is addressed using a variety of data analysis techniques. An example of a chemical process is utilized to evaluate the performance of NODE-based MPC. Furthermore, the performance of the NODE-based MPC under Gaussian and non-Gaussian noise is investigated, and the subsampling method is found to be effective in building models for MPC that are suitable for handling the presence of non-Gaussian noise in the data.

Multi-dimensional spatial–temporal graph convolution for urban sensors imputation and enhancement

Spatio-temporal data are critical for intelligent systems, such as smart transportation and smart cities. However, due to sensor failure or power failure, the spatiotemporal data missing tends to have a big impact on downstream tasks. Meanwhile, if sensors are scarce, some spatial positions without sensors need data enhancement through intelligent methods. Existing workarounds focus on modeling temporal information (such as time series), often ignoring spatial dependency, or modeling the spatial and temporal domain separately for imputation. In this paper, we propose a Long-term Multi-dimensional Spatial–Temporal Graph Convolution Network (LMSTGCN), which not only inductively estimates some missing data, but also achieves data augmentation of target locations. It contains a periodic temporal encoding mechanism, a gated temporal capture module, and a multi-dimensional spatial–temporal GCN module. The long-term temporal dependencies are captured by the periodic temporal encoding mechanism. The spatial and extra-short-term temporal dependencies are simultaneously modeled by the multi-dimensional GCN module, which can achieve exponential growth in the range of receptive fields. Corresponding to this module, we designed a spatiotemporal adjacency matrix construction method. It generates spatiotemporal adjacency matrices of corresponding time length as needed. The short-term dependencies in sequences are captured by the gated temporal capture module. In experimental analysis, results demonstrate that the proposed model outperforms the state-of-the-art baselines on real-world data sets.

Assessment of Oil Chromatography Sensor Failure for Transformer Based on Multi-dimensional Data-driven

Online monitoring and analysis of transformer oil chromatography is of great significance for the assessment of internal insulation with transformer operation. The existence of numerous invalid monitoring sensors in the field makes the quality of online monitoring data degraded, which makes the online monitoring system unable to monitor the operating status of transformers in a timely and accurate manner. This paper propose a data-driven method for evaluating the effectiveness of oil chromatography monitoring sensors by selecting a fixed-length feature data set from online data for evaluation, and then using the distribution of abnormal values, the distribution of consecutive identical values, the variation of coefficient of variation and the variation of gas production rate in the feature data set as criteria to judge the feature data set and obtain the corresponding discriminant values. Afterwards, the discriminant values are assigned weights according to the focus of each criterion to obtain the sensor status values, which are compared with the preset tolerance level to obtain the sensor evaluation results. The method proposed evaluates the effectiveness of online monitoring sensors from several aspects, which can detect faulty sensors in time and is of great significance for improving the accuracy of online monitoring data and the reliability of monitoring device.

Investigation of denoising autoencoder-based deep learning model in noise-riding experimental data for reliable state-of-charge estimation

State-of-charge (SOC) estimation plays a crucial role in battery management systems to ensure safe and reliable operation. However, SOC estimation remains challenging due to the dynamic nature of battery systems and varying ambient conditions. Data-driven methods have emerged as effective tools for analyzing nonlinear dynamical systems, but their performance heavily relies on data quality. In actual applications, data susceptible to distortions caused by external factors such as sensor failure, circuitry, and temperature variations, leading to degraded model performance. To address the performance degradation resulting from data quality deterioration, this paper introduces a denoising autoencoder is implemented as a stacked multi-layer perceptron, which learns to reconstruct distorted data. Furthermore, we propose the ensemble method that combines the autoencoder with an estimation model for SOC estimation in lithium-ion batteries. The effectiveness of the proposed model is demonstrated through tests conducted on a dataset comprising drive cycle profile of Panasonic 18650PF cells. The model validated under two ambient temperatures scenarios: identical and different, using a distorted dataset with added randomly added noise and dropout. The experimental results reveal that the proposed model achieved a 3 % error in training the drive profile relative to the actual values at different ambient temperatures. When compared to the plain model, the proposed ensemble model showed an increased RMSE of 4 %. Additionally, the performance of different estimation models was compared, with the LSTM model achieving an RMSE 0.67 at different ambient temperatures, outperforming the Support Vector Regression (SVR) with an RMSE 1.35 and the Extended Kalman Filter (EKF) with an RMSE of 0.87.

A novel method for the recovery of continuous missing data using multivariate variational mode decomposition and fully convolutional networks

The sensor failure is the frequently-occurring event in engineering activities related to the measurement, which could yield a serious problem of continuous data loss. This special issue will affect the accuracy of structural damage identification and condition assessment. However, traditional methods (e.g., deep learning methods) are hard to settle the recovery of the data with complex characteristics. To this end, this study develops an innovative framework based on the combination of multivariate variational mode decomposition (MVMD), fully convolutional networks (FCN), and a systematic evaluation pattern. In this framework, MVMD is first used to simultaneously decompose the multichannel data into a set of intrinsic mode functions (IMFs) with a modal alignment attribute. Then, FCN is designed to construct a mapping relationship among different IMFs having the same level number in an end-to-end manner, by which the continuous data recovery at the location of interest can be realized. Obviously, this method can not only make the utmost of the spatial information, but also involve an excellent data mining ability of FCN. Further, a systematic pattern from the perspective of time and frequency domains is defined to evaluate the recovery ability of the concerned models. Finally, case study based on the measured data of a steel-frame reinforced arch rib is used to exhibit the performance of the proposed method. Additionally, the parameter analysis and practical application are respectively performed to examine its robustness and practicality. All these results demonstrate the superiority of this method, and its performance is closely related to the location of interest, the number of training data channels and their corresponding location.

A Systematic Review of Data Quality in CPS and IoT for Industry 4.0

The Internet of Things (IoT) and Cyber-Physical Systems (CPS) are the backbones of Industry 4.0, where data quality is crucial for decision support. Data quality in these systems can deteriorate due to sensor failures or uncertain operating environments. Our objective is to summarize and assess the research efforts that address data quality in data-centric CPS/IoT industrial applications. We systematically review the state-of-the-art data quality techniques for CPS and IoT in Industry 4.0 through a systematic literature review (SLR) study. We pose three research questions, define selection and exclusion criteria for primary studies, and extract and synthesize data from these studies to answer our research questions. Our most significant results are (i) the list of data quality issues, their sources, and application domains, (ii) the best practices and metrics for managing data quality, (iii) the software engineering solutions employed to manage data quality, and (iv) the state of the data quality techniques (data repair, cleaning, and monitoring) in the application domains. The results of our SLR can help researchers obtain an overview of existing data quality issues, techniques, metrics, and best practices. We suggest research directions that require attention from the research community for follow-up work.

A human-centered framework to update digital twins

Developing a digital twin (DT) involves establishing (1) a predictive capability (a model) relevant to the application, (2) means to collect data from the physical counterpart, and (3) means to apply the collected data to the model. Ideally, with these three goals achieved, long periods of steady-state use of the DT might be interrupted only by failure of the sensors used to collect data from the physical counterpart. In reality, however, it can be difficult to confirm that the DT system occupies this comfortable steady-state position. Assessing uncertainty in the predictive model, and ensuring the relevance of data collected from the physical counterpart are design-time activities with unclear termination points. Distinguishing sensed change in the physical counterpart from sensor failure is a persistent challenge. In this short paper we describe early work towards a human-centered framework to establish, refine, and update digital twins. Condition-based maintenance and gear backlash in production equipment are used as examples.

Image similarity-based gap filling method can effectively enrich surface water mapping information

Satellite monitoring is an effective way to obtain the spatiotemporal information for surface water resources. However, due to problems such as clouds, cloud shadows, and sensor failures, there many gaps in satellite images, posing challenges for the accurate monitoring of surface water changes. Here, a new image-similarity-based gap-filling method using Landsat images, to obtain gapless surface water images, is proposed. The new method, based on a database composed of gapless images, uses the symmetrical difference method to determine the gapless image whose water pixels distribution is most similar to that of the gap image, finally replacing the gapped image pixels with gapless image pixels, to obtain a gap-filled water image. The performance of the new method is evaluated for arid and humid regions via numerical experiments and case verification. Then new method is compared with the existing gap-filling method, using the Aral Sea as an example. The new method’s average overall accuracy for surface water in arid areas is about 0.99, and that of surface water in humid areas is about 0.98. It predicted surface water more accurately than other methods, and with high accuracy (R2 = 0.92). This proposed gap-filling method provides an important technical solution to obtaining long-term high-quality observations of surface water. It can also provide a database of image for accurate water resource management.

Tensorized topological graph learning for generalized incomplete multi-view clustering

The success of the current multi-view clustering lies in the default assumption of completeness on each view, while it is hardly satisfied for real-world applications. Incomplete multi-view clustering (IMC) studies clustering multi-view instances that are partially observed due to data corruption or sensor failure. Although making some progress, the current research suffers from the following obstacles: (1) Existing works are always built on pairwise similarity measurement, which cannot fully capture the topological structure of incomplete multi-view samples for precise clustering; (2) Due to the absence of partially-observed samples across multiple views, how could we recover and enhance the high-order similarities across different views becomes one of the main bottlenecks of IMC; (3) Current research mainly aims to handle specific dual-view IMC, which greatly limits the deployment in real-world applications. In this paper, we propose a novel Tensorized Topological Graph Learning (TTGL) for the generalized incomplete multi-view clustering, which jointly considers the topological graph construction, missing feature completion, and high-order uncertain correlation enhancement. Specifically, instead of using pairwise similarity measurement, we formulate a consensus topological graph construction module, which can coalesce affinity matrices to analyze the topological structure of incomplete multi-view data. Moreover, we build an iterative missing similarity imputation scheme for feature completion, which ensures the intrinsic similarity preservation on observed instances as well as an imputation on unobserved ones. Furthermore, a low-rank tensor structure is imposed to capture the high-order similarities via both feature and similarity enhancement and completion across different views. Our method can cluster incomplete data with an arbitrary number of views and any missing statuses. Extensive experiments on several benchmark datasets validate the effectiveness of our method when compared with a series of state-of-the-art IMC clustering algorithms. The source code of our TTGL is available at: https://github.com/DarrenZZhang/TTGL.

ABTCN: an efficient hybrid deep learning approach for atmospheric temperature prediction.

Temperature prediction is an important and significant step for monitoring global warming and the environment to save and protect human lives. The climatology parameters such as temperature, pressure, and wind speed are time-series data and are well predicted with data driven models. However, data-driven models have certain constraints, due to which these models are unable to predict the missing values and erroneous data caused by factors like sensor failure and natural disasters. In order to solve this issue, an efficient hybrid model, i.e., attention-based bidirectional long short term memory temporal convolution network (ABTCN) architecture is proposed. ABTCN uses k-nearest neighbor (KNN) imputation method for handling the missing data. A bidirectional long short term memory (Bi-LSTM) network with self-attention mechanism and temporal convolutional network (TCN) model that aids in the extraction of features from complex data and prediction of long data sequence. The performance of the proposed model is evaluated in comparison to various state-of-the-art deep learning models using error metrics such as MAE, MSE, RMSE, and R2 score. It is observed that our proposed model is superior over other models with high accuracy.

Fault-Tolerant Control Strategy for Hall Sensors in BLDC Motor Drive for Electric Vehicle Applications

The adoption of the brushless DC motor in the electric drive vehicle industry continues to grow due to its robustness and ability to meet torque–speed requirements. This work presents the implementation of a fault-tolerant control (FTC) for a BLDC motor designed for electric vehicles. This paper focuses on studying the defect in the Ha sensor and its signal reconstruction, assuming possible cases, but the same principle is applied to the other two sensors (Hb and Hc ). In this case, the fault diagnosis allows for the correction and reconstruction of the signal in order to compel the system to work despite the presence of a fault. Indeed, several robust control systems are used within the work to regulate the speed of the motor properly, such as control via fuzzy logic and control via a neural network. This paper presents three BLDC control configurations for EVs, PID, fuzzy logic (FL), and an artificial neural network (ANN), discusses the pros and cons, and develops corresponding mathematical models to enhance a fault-tolerant control strategy which is analyzed and studied using MATLAB-based simulations (by discussing the two cases, the steady state and the transient state), allowing for a novel design based on the analytical models developed. The results obtained from the simulation of this system improved the speed controlled by the neural network compared to the fuzzy logic controller. At the same time, the sensor failure had no effect on the system’s operation due to the efficiency of the FTC control.

Speed Estimation of Multiphase Induction Motor Using Rotor Slot Harmonics With Limited SNR and Dynamic Load Conditions

Accurate and reliable speed feedback is vital for closed-loop control of multiphase induction motors (IMs), especially in harsh operating scenes like ship propulsion. High performance speed estimation is effective in getting rid of usual sensor failures, through the rotor slot harmonic (RSH) based method that featured for parameter mismatch robustness. However, the accurate and real-time low speed estimation through RSHs under limited signal noise ratio (SNR) and dynamic load conditions is still a challenge, mainly due to the extremely weak and nonstationary amplitude and closely distributed frequency properties of RSHs. In this paper, we investigate the performance limitations of low speed estimation using primary rotor slot harmonics (PSH) of multiphase IMs, whose detectability is ensured through an improved interior detection coil, under limited SNR and dynamic load conditions. A frequency shifting and an amplitude limiting zero-crossing detection algorithm are proposed for efficient extraction and processing of PSHs in full speed range, which effectively improve the estimation accuracy with reduced complexity. Finally, the proposed method is investigated on a nine-phase IM experimental platform within the whole speed range, limiting the steady and transient error respectively under 4r/min and 20r/min for closed-loop control.

A Markov Model of Gap Occurrence in Continuous Glucose Monitoring Data for Realistic in Silico Clinical Trials

Background and objectiveContinuous glucose monitoring (CGM) sensors measure interstitial glucose concentration every 1-5 min for days or weeks. New CGM-based diabetes therapies are often tested in in silico clinical trials (ISCTs) using diabetes simulators. Accurate models of CGM sensor inaccuracies and failures could help improve the realism of ISCTs. However, the modeling of CGM failures has not yet been fully addressed in the literature. This work aims to develop a mathematical model of CGM gaps, i.e., occasional portions of missing data generated by temporary sensor errors (e.g., excessive noise or artifacts). MethodsTwo datasets containing CGM traces collected in 167 adults and 205 children, respectively, using the Dexcom G6 sensor (Dexcom Inc., San Diego, CA) were used. Four Markov models, of increasing complexity, were designed to describe three main characteristics: number of gaps for each sensor, gap distribution in the monitoring days, and gap duration. Each model was identified on a portion of each dataset (training set). The remaining portion of each dataset (real test set) was used to evaluate model performance through a Monte Carlo simulation approach. Each model was used to generate 100 simulated test sets with the same size as the real test set. The distributions of gap characteristics on the simulated test sets were compared with those observed on the real test set, using the two-sample Kolmogorov-Smirnov test and the Jensen-Shannon divergence. ResultsA six-state Markov model, having two states to describe normal sensor operation and four states to describe gap occurrence, achieved the best results. For this model, the Kolmogorov-Smirnov test found no significant differences between the distribution of simulated and real gap characteristics. Moreover, this model obtained significantly lower Jensen-Shannon divergence values than the other models. ConclusionsA Markov model describing CGM gaps was developed and validated on two real datasets. The model describes well the number of gaps for each sensor, the gap distribution over monitoring days, and the gap durations. Such a model can be integrated into existing diabetes simulators to realistically simulate CGM gaps in ISCTs and thus enable the development of more effective and robust diabetes management strategies.

Moving-Landmark Assisted Distributed Learning Based Decentralized Cooperative Localization (DL-DCL) with Fault Tolerance

This paper considers the problem of cooperative localization of multiple robots under uncertainty, communicating over a partially connected, dynamic communication network and assisted by an agile landmark. Each robot owns an IMU and a relative pose sensing suite, which can get faulty due to system or environmental uncertainty, and therefore exhibit large bias in their estimation output. For the robots to localize accurately under sensor failure and system or environmental uncertainty, a novel Distributed Learning based Decentralized Cooperative Localization (DL-DCL) algorithm is proposed that involves real-time learning of an information fusion strategy by each robot for combining pose estimates from its own sensors as well as from those of its neighboring robots, and utilizing the moving landmark's pose information as a feedback to the learning process. Convergence analysis shows that the learning process converges exponentially under certain reasonable assumptions. Simulations involving sensor failures inducing around 40-60 times increase in the nominal bias show DL-DCL's estimation performance to be approximately 40% better than the well-known covariance-based estimate fusion methods. For the evaluation of DL-DCL's implementability and fault-tolerance capability in practice, a high-fidelity simulation is carried out in Gazebo with ROS2.

CLUE-AD: A Context-Based Method for Labeling Unobserved Entities in Autonomous Driving Data

Generating high-quality annotations for object detection and recognition is a challenging and important task, especially in relation to safety-critical applications such as autonomous driving (AD). Due to the difficulty of perception in challenging situations such as occlusion, degraded weather, and sensor failure, objects can go unobserved and unlabeled. In this paper, we present CLUE-AD, a general-purpose method for detecting and labeling unobserved entities by leveraging the object continuity assumption within the context of a scene. This method is dataset-agnostic, supporting any existing and future AD datasets. Using a real-world dataset representing complex urban driving scenes, we demonstrate the applicability of CLUE-AD for detecting unobserved entities and augmenting the scene data with new labels.

Fault‐tolerant control of multi‐agent systems with input delay and sensor failure

AbstractThe problem of fault‐tolerant control (FTC) for a multi‐agent system (MAS) with input delay and sensor failures is addressed in this study. The topology of communication is an undirected subgraph having directed connections between the followers and leader. The Lyapunov approach is used to identify sufficient criteria for consensus control. We propose a fault‐tolerant controller based on observed descriptors. On the basis of proposed fault‐tolerant control, actuator problems, sensor faults, and state estimations can be solved. To minimize the size of the data, the Artstein–Kwon–Pearson reduction method is applied. Based on a Lyapunov functional predictive descriptor observer architecture, sufficient conditions are derived to guarantee the asymptotic stability. To demonstrate the theoretical results, numerical examples are provided.

A data-driven persistence test for robust (probabilistic) quality control of measured environmental time series: constant value episodes

Abstract. Robust quality control is a prerequisite and an essential component in any data application. That is especially important for time series of environmental observations such as air quality due to their dynamic and irreversible nature. One of the common issues in these data is constant value episodes (CVEs), where a set of consecutive data values remains constant over a given period. Although CVEs are often considered to be an indicator of sensor failure or other measurement errors and are removed during quality control procedures, there are situations when CVEs reflect natural environmental phenomena, and they should not be removed from the data or analysis. Assessing whether the CVEs are erroneous data or valid observations is a challenge. As there are no formal procedures established for this, their classification is based on subjective judgment and is therefore uncertain and irreproducible. This paper presents a novel test procedure, i.e., constant value test, to estimate the probability of CVEs being valid data. The theoretical foundation of this test is based on statistical characteristics and probability theory and takes into account the numerical precision of the data values. The test is a data-driven (parametric) approach, which makes it usable for time series analysis in different environmental research domains, as long as serial dependency is given and the data distribution is not too different from Gaussian. The robustness of the test was demonstrated with sensitivity studies using synthetic data with different distributions. Example applications to measured air temperature and ozone mixing ratio data confirm the versatility of the test.

Methods for Monitoring and Diagnosing Space Communication and Orientation Subsystems

Monitoring of the state and behavior of spacecraft subsystems based on telemetry data is designed to continuously perform the tasks of controlling, controlling and maintaining the technical characteristics of a spacecraft. Increasing requirements for the characteristics of monitoring methods inevitably leads to a revision of the control technologies used and the need to create a scientific and technical reserve in the form of approaches, methods and technologies for building promising competitive space technology, including: technologies for building integrated information support for solving the problems of monitoring spacecraft subsystems; applied object-oriented systems of artificial intelligence, neural networks and high-performance computing for control, diagnostics and decision support. Currently, the following issues of monitoring the fault tolerance and reliability of space systems are relevant: the development of concepts for flight control centers that control multi-satellite systems and imply high reliability, processing of extra-large amounts of information, autonomy of operation, transfer offunctions on board, reduced human participation in control; analysis of possible failures; substantiation of the use of the on-board system for diagnostics, monitoring and localization of failures and malfunctions, restoration of working capacity; ensuring the maximum achievable (close to absolute) safety, eliminating the catastrophic consequences of possible failures of individual elements and subsystems; development and improvement of the methodology for setting requirements, assessing and controlling the quality and reliability of spacecraft and their components; improvement of international standardization of commercial rocket and space technology; optimal integrated risk management of space activities. The work is devoted to the control and diagnostics of subsystems of communication and orientation of the spacecraft. An artificial neural network, pair and multiple correlation functions are used as a mathematical apparatus to support the operator's decision making. It is shown that the methods make it possible to detect malfunctions and failures of sensors in real time. Calculations based on data obtained from spacecraft sensors are presented.

Outage Management of Active Distribution Systems With Data Fusion From Multiple Sensors Given Sensor Failures

Complex scenarios of multiple faults, sensor failures, and the fault current from distributed generations (DGs) pose a challenge for efficient outage management in distribution systems. To solve the issue, this study proposes a new algorithm for sequential determination of outage scenarios with data fusion from multiple sensors, i.e., smart meters and remote fault indicators (RFIs). To this end, an active distribution system is modeled as an undirected graph with vertices representing load zones, protective devices and RFIs, and edges denoting their adjacency. Set operation is proposed to derive the expected RFI alarms for each load zone if faulted and smart meter outage reports for each protection device. This serves as an input to the proposed evidence-driven methodology to sequentially quantify the credibility of the potential outage scenarios supported by RFIs and smart meters. By ranking the credibility, the most credible scenario of the activated protection and fault location, either on the primary feeder or laterals, is determined for expedited power recovery. Test cases of the IEEE System and a utility circuit show that the proposed approach outperforms the state-of-the-art technologies in terms of its accuracy and computational efficiency, leading to more efficient outage management of distribution systems.