Data-driven Reliability Research Articles

Data-driven fast reliability assessment of offshore structures based on real-time sensor data

With the advancement of monitoring technologies for offshore structures, sensor data are becoming increasingly available in real-time. The objective of this paper is to develop fast methods for reliability assessment of offshore structures for the extreme and fatigue failure modes, based on the latest real-time sensor data, assumed to be four months. The continuously updated assessment informs operators on the failure probability over the next few years. Reliability analysis using real-time data is challenging owing to the need to extrapolate the observed data to more severe sea states and across many orders of magnitudes of probabilities, and there is a lack of related studies. Here, the environment comprises the short-term and long-term wave characteristics. Three data-driven reliability approaches are developed for three types of wave sensor data. The first approach applies statistical models to the response when wave data are unavailable. The second relies on Gaussian Process Regression (GPR) when only the wave spectral parameters are known. The third applies the Gaussian-experienced NARX network to predict the response time series when the wave time series are recorded. The proposed approaches are applied to a jack-up platform, and implementation of benchmark reliability analysis is enabled by establishing an emulator trained from dynamic simulations for fast generation of synthetic sensor data. The GPR approach is found to show good extrapolation capabilities and able to provide relatively accurate and robust reliability predictions.

Data-Driven Reliability Prediction for District Heating Networks

As district heating networks age, current asset management practices, such as those relying on static life expectancies and age- and rule-based approaches, need to be replaced by data-driven asset management. As an alternative to physics-of-failure models that are typically preferred in the literature, this paper explores the application of more accessible traditional and novel machine learning-enabled reliability models for analyzing the reliability of district heating pipes and demonstrates how common data deficiencies can be accommodated by modifying the models’ likelihood expressions. The tested models comprised the Herz, Weibull, and the Neural Weibull Proportional Hazard models. An assessment of these models on data from an actual district heating network in Funen, Denmark showed that the relative youth of the network complicated the validation of the models’ distributional assumptions. However, a comparative evaluation of the models showed that there is a significant benefit in employing data-driven reliability modeling as they enable pipes to be differentiated based on the their working conditions and intrinsic features. Therefore, it is concluded that data-driven reliability models outperform current asset management practices such as age-based vulnerability ranking.

Multi-fidelity wavelet neural operator surrogate model for time-independent and time-dependent reliability analysis

Operator learning frameworks have recently emerged as an effective scientific machine learning tool for learning complex nonlinear operators of differential equations. Since neural operators learn an infinite-dimensional functional mapping, it is useful in applications requiring rapid prediction of solutions for a wide range of input functions. A task of a similar nature arises in many applications of uncertainty quantification, including reliability estimation and design under uncertainty, each of which demands thousands of samples subjected to a wide range of possible input conditions, an aspect to which neural operators are specialized. Although the neural operators are capable of learning complex nonlinear solution operators, they require an extensive amount of data for successful training. Unlike the applications in computer vision, the computational complexity of the numerical simulations and the cost of physical experiments contributing to the synthetic and real training data compromise the performance of the trained neural operator model, thereby directly impacting the accuracy of uncertainty quantification results. We aim to alleviate the data bottleneck by using multi-fidelity learning in neural operators, where a neural operator is trained by using a large amount of inexpensive low-fidelity data along with a small amount of expensive high-fidelity data. We propose the multi-fidelity wavelet neural operator, capable of learning solution operators from a multi-fidelity dataset, for efficient and effective data-driven reliability analysis of dynamical systems. We illustrate the performance of the proposed framework on bi-fidelity data simulated on coarse and refined grids for spatial and spatiotemporal systems.

Data-driven time-variant reliability assessment of bridge girders based on deep learning

This article presents a new data-driven reliability analysis framework through combining the sample convolution and interaction network (SCINet) with Bayesian probability recursion to predict the time-variant reliability of bridge girders. The structural response predicted by SCINet and the state parameter (i.e., the variance of normal distribution) estimated by Bayesian dynamic linear model (BDLM) are used to form the limit state function to predict time-variant reliability. The results with a practical bridge show that the proposed method can predict future structural responses and time-variant reliability more accurately than BDLM, long short-term memory (LSTM), and long- and short-term time-series network (LSTNet) methods.

Data-driven reliability framework for qualitative damage states of reinforced concrete beams under blast loading

The incessant demand for protective infrastructure research has been motivated by recent accidental and intentional explosion events. This has led to a significant increase in blast design approaches and modern methods of structural analysis. When considering the analysis and response of blast-loaded members, like the identification of crack patterns and failure modes (CP & FM) of reinforced concrete (RC) beams, the current literature heavily depends on complex numerical modeling. Recently, the adoption of machine learning (ML) modeling to readily predict damage state labels of blast-loaded RC beams has emerged as a competitor to numerical methods. In either modeling method, outputs of damage states are often oversimplified as deterministic labels without elaborating on the possibility of alternative damage state. To extend on the practical usage of deterministic modeling, the present study introduces a probabilistic framework which integrates ML models towards ascertaining the probability of different CP & FM classes via the Monte Carlo simulation. The framework accounts for the uncertainties of different beam and blast parameters and incorporates four damage states of flexural cracking, bending failure, flexural-shear cracking, and crushing failure. The developed framework was evaluated against forty-two blast-loaded RC beams available from several previous experimental studies and exhibited excellent predictive performance while providing significant insight to the occurrence of all four damage states. In addition, extensive Blast Reliability Curves (BRC) were developed for eighteen RC beam cases over a wide range of blast intensities and the effects of beam and blast parameters on qualitative damage states were evaluated. The most notable observation was that the sensitivity of both concrete and steel strengths on damage state probabilities increased with reinforcement ratio. The BRCs depicted an elaborate transition of damage states with varying blast magnitudes. Also, the limitations, sources of error, and further improvements associated with the proposed methodology were clearly defined. Overall, the resulting probabilistic structural blast analysis introduced by the proposed framework serves as an intricate approach to providing greater comprehension to the CP & FM response of RC beams under blast loading which aims to promote confidence in decision making within protective design processes.

A fuzzy data-driven reliability analysis for risk assessment and decision making using Temporal Fault Trees

Fuzzy data-driven reliability analysis has been used in different safety-critical domains for risk assessment and decision-making where precise failure data is non-existent. Expert judgements and fuzzy set theory have been combined with different variants of fault trees as part of fuzzy data-driven reliability analysis studies. In such fuzzy fault tree analyses, different people represented failure data using different membership functions for the fuzzy set, and different parameters were set differently in the expert opinion elicitation process. Due to the availability of a wide variety of options, it is possible to obtain different outcomes when choosing one option over another. This article performed an analysis in the context of fuzzy data-based temporal fault tree analysis to investigate the effect of choosing different membership functions on the estimated system reliability and criticality ranking of different failure events. Moreover, the effect of using different values for the relaxation factor, a parameter set during the expert elicitation process, was studied on the system reliability and criticality evaluation. The experiments on the fuel distribution system case study show system reliability did not vary when triangular and trapezoidal fuzzy numbers were used with the same upper and lower bounds. However, it was seen that the criticality rankings of a couple of events were changed due to choosing different membership functions and different values of relaxation factor

Data-Driven Reliability Models of Quantum Circuit: From Traditional ML to Graph Neural Network

The current advancement in quantum computers has been focusing on increasing the number of qubits and enhancing their fidelity. However, the available quantum devices, known as Intermediate Scale Quantum (NISQ) computers, still suffer from different sources of noise that impact their reliability. Thus, practical noise modeling is of great importance in the development of quantum error mitigation approaches. In this paper, we propose a Machine Learning (ML)-based scheme to predict the output fidelity of the quantum circuit executed on NISQ devices. We show the benefit of using Graph Neural Network (GNN)-based models compared to traditional ML-based models in capturing the quantum circuit structure in addition to its gates’ features, which enable characterizing unpredicted quantum circuit errors. We use different metrics to measure the fidelity of the quantum circuit output. Our experimental results using different quantum algorithms executed on IBM Q Guadalupe quantum computer show the high prediction accuracy of our ML reliability models. Our results also show that our models can guide the single-qubit gate rescheduling to improve the output fidelity of the quantum circuit without the need for prior execution of dedicated calibration circuits.

Data-driven reliability and resilience measure of transportation systems considering disaster levels

Data-driven reliability and resilience measure of transportation systems considering disaster levels

MAntRA: A framework for model agnostic reliability analysis

We propose a novel model-agnostic data-driven reliability analysis framework for time-dependent reliability analysis. The proposed approach – referred to as MAntRA – combines Bayesian inference and stochastic differential equations to evaluate the reliability of stochastically-driven dynamical systems for which the governing physics is a priori unknown. A two-stage approach is adopted: in the first stage, an efficient variational Bayesian equation discovery algorithm is developed to determine the governing physics of an underlying stochastic differential equation (SDE) from measured output-only data. The developed algorithm is efficient and accounts for epistemic uncertainty due to limited and noisy data and aleatoric uncertainty because of environmental effects and external excitation. In the second stage, the discovered SDE is solved using a stochastic integration scheme, and the probability of failure is computed. The efficacy of the proposed approach is illustrated in four numerical examples. The results obtained indicate the possible application of the proposed approach for reliability analysis of in-situ and heritage structures from on-site measurements.

Data-driven reliability assessment with scarce samples considering multidimensional dependence

This study proposes a data-driven method for assessing reliability, based on the scarce input dataset with multidimensional correlation. Since considering the distribution parameters estimated from the scarce dataset as those of the population may lead to epistemic uncertainty, the bootstrap resampling algorithm is adopted to infer the distribution parameters as interval parameters. To account for the variable dependence, vine copula theory is utilized to construct the joint probability density function (PDF) of input variables, and maximum likelihood estimation (MLE) and Akaike information criterion (AIC) analysis are employed to select optimal copulas based on the samples for the vine structure. Subsequently, the failure probability bounds of a response function are calculated based on the constructed joint PDF with interval distribution parameters by the active learning Kriging (AK) method combining the sparse grid integration (SGI) method. Finally, several examples are provided to demonstrate the feasibility and efficiency of the proposed method.

The Distance of Random Permutation Set

The Random Permutation Set (RPS) is a newly proposed type of set. An RPS takes the permutation of a certain set into consideration and can be regarded as a generalization of evidence theory. Since distance is an important metric, an immediate question which arises is how to calculate the distance between RPSs. To address this problem, the distance between RPSs is first presented in this paper. Based on a geometrical interpretation of RPS, the distance between RPSs is measured by considering the similarity between their permutation sets, both in the number and order of elements. Numerical examples demonstrate the behavior of the proposed RPS distance. Thereafter, based on the proposed RPS distance, a Data-Driven Reliability Determination (DDRD) algorithm is put forth and applied in threat assessment. The experimental results show that the proposed DDRD algorithm can effectively identify the fusion order of RPSs. The corresponding fusion result is in agreement with theoretical predictions and has the highest accuracy of identification.

Enhancing Reliability Analysis with Multisource Data: Mitigating Adverse Selection Problems in Bridge Monitoring and Management

Data collected using sensors plays an essential role in active bridge health monitoring. When analyzing a large number of bridges in the U.S., the National Bridge Inventory data as been widely used. Yet, the database does not provide information about live loads, one of the most indeterminate variables for monitoring bridges. Such asymmetric information can lead to an adverse selection problem in making maintenance, rehabilitation, and repair decisions. This study proposes a data-driven reliability analysis to assess probabilities of bridge failure by synthesizing NBI data and Weigh-In-Motion (WIM) data for a large number of bridges in Georgia. On the resistance side, tree ensemble methods are employed to support the hypothesis that the NBI operating load rating represents the distribution of bridge resistance capacities which change over time. On the loading side, the live load distribution is derived from field data collected using WIM sensors. Our results show that the proposed WIM data-enabled reliability analysis substantially enhances information symmetry and provides a reliability index that supports monitoring of bridge conditions, depending on live loads and load-carrying capacities.

Data-driven reliability assessment of dynamic structures based on power spectrum classification

The power spectral density function is a widely used tool to determine the frequency components and amplitudes of environmental processes, such as earthquakes or wind loads. It is an important technique especially in the engineering field of vibration analysis and in determining the response of structures. When using a large amount of data, a load model can be generated, which describes the characteristics of the underlying stochastic process. This load model enables artificially generated excitations to be created within the framework of Monte Carlo simulations. If multiple data records are utilised, a problem that can occur is that the individual records have a high variance in the frequency domain and are therefore too dissimilar from each other, even though they appear to be similar in the time domain. A load model derived from this data does not represent the entire data set, because not the whole spectral range is covered. Therefore, every attempt must be made to group the records according to their characteristics and thus combine similar data to derive two or more load models accordingly. In this work, an approach is proposed to classify real earthquake ground motion records using the k-means algorithm based on similarities within the data ensemble as determined by the Bhattacharyya distance. The silhouette method enables the identification of the optimal number of groups for the classification. The classified data thus form a subset of the entire data set from which load models can be generated and can be applied separately to the structure under investigation, leading to more accurate simulation results. The advantages of this classification approach are illustrated by means of an academic example and a seismic-isolated bridge pier model as a non-linear dynamic system.

Data-Driven Reliability Evaluation of the Integrated Energy System Considering Optimal Service Restoration

The demand for environmental protection and energy utilization transformation has promoted the rapid development of integrated energy systems (IES). Reliability evaluation is a fundamental element in designing IES as it could instruct the planning and operation of IES. This study proposes a novel data-driven reliability improvement and evaluation method considering the three-state reliability model and an optimal service restoration model (OSR). First, a multi-energy flow model is introduced and linearized in order to reduce the computing complexity. Next, a three-state reliability model is developed, considering the transitional process and partial failure mode. Furthermore, an optimal service restoration model is established to determine the best repairment moment for minimizing the load curtailment, and a data-driven reliability evaluation method is developed that integrates OSR and models the stochastic state transition process using the historical measurement data of the smart meters. Finally, the proposed reliability evaluation method is tested on a test IES, and the numerical results validate its effectiveness in evaluating the reliability of IES and improving the overall reliability.

Data-driven optimization of reliability using buffered failure probability

Design and operation of complex engineering systems rely on reliability optimization. Such optimization requires us to account for uncertainties expressed in terms of complicated, high-dimensional probability distributions, for which only samples or data might be available. However, using data or samples often degrades the computational efficiency, particularly as the conventional failure probability is estimated using the indicator function whose gradient is not defined at zero. To address this issue, by leveraging the buffered failure probability, the paper develops the buffered optimization and reliability method (BORM) for efficient, data-driven optimization of reliability. The proposed formulations, algorithms, and strategies greatly improve the computational efficiency of the optimization and thereby address the needs of high-dimensional and nonlinear problems. In addition, an analytical formula is developed to estimate the reliability sensitivity, a subject fraught with difficulty when using the conventional failure probability. The buffered failure probability is thoroughly investigated in the context of many different distributions, leading to a novel measure of tail-heaviness called the buffered tail index. The efficiency and accuracy of the proposed optimization methodology are demonstrated by three numerical examples. Although they might show slight deviations from a target failure probability because of sampling errors and other inaccuracies, the results underline unique advantages and potentials of the buffered failure probability for data-driven reliability analysis.

Collaborative data-driven reliability analysis of multi-state fault trees

Fault tree modeling and failure analysis of systems that are equipped with sensors and meters are becoming more automated and less human-dependent. For a single system to benefit from its own collected data, it will need to wait for a long time to collect sufficient data to build representative models to increase its reliability. Therefore, if multiple systems with similar functionalities cooperate, the resolution of the collected data will increase. This leads to extracting fault trees with higher accuracy in failure detection and prediction. In this paper, we present an extended approach for collaborative Data-Driven Fault Tree Analysis (DDFTA) of a system which extracts repairable fault trees from time series data streaming from multiple systems/machines sharing similar functionalities. Results are analyzed to estimate the system’s reliability measures and investigate the effect of number of machines cooperating in data collection. Our method is not limited to binary (two states) components, nor to exponential distributions. Results show that applying collaborative data analytics significantly increases the accuracy of data-driven fault tree analysis, specifically for systems following nonexponential distributions.

Robust Inertial Motion Tracking through Deep Sensor Fusion across Smart Earbuds and Smartphone

IMU based inertial tracking plays an indispensable role in many mobility centric tasks, such as robotic control, indoor navigation and virtual reality gaming. Despite its mature application in rigid machine mobility (e.g., robot and aircraft), tracking human users via mobile devices remains a fundamental challenge due to the intractable gait/posture patterns. Recent data-driven models have tackled sensor drifting, one key issue that plagues inertial tracking. However, these systems still assume the devices are held or attached to the user body with a relatively fixed posture. In practice, natural body activities may rotate/translate the device which may be mistaken as whole body movement. Such motion artifacts remain as the dominating factor that fails existing inertial tracing systems in practical uncontrolled settings. Inspired by the observation that human heads induces far less intensive movement relative to the body during walking, compared to other parts, we propose a novel multi-stage sensor fusion pipeline called DeepIT, which realizes inertial tracking by synthesizing the IMU measurements from a smartphone and an associated earbud. DeepIT introduces a data-driven reliability aware attention model, which assesses the reliability of each IMU and opportunistically synthesizes their data to mitigate the impacts of motion noise. Furthermore, DeepIT uses a reliability aware magnetometer compensation scheme to combat the angular drifting problem caused by unrestricted motion artifacts. We validate DeepIT on the first large-scale inertial navigation dataset involving both smartphone and earbud IMUs. The evaluation results show that DeepIT achieves multiple folds of accuracy improvement on the challenging uncontrolled natural walking scenarios, compared with state-of-the-art closed-form and data-driven models.

Study on potential energy curves and ro-vibrational energies of DT, HT and T2 molecules

Accurately monitoring and effectively controlling the tritium compounds based on their ro-vibrational energy structure are important issues in various nuclear systems. Because of their radioactivity, it is difficult to obtain the corresponding energies directly through experiments. In this paper, the potential energy curves and the corresponding ro-vibrational full spectrum of DT, HT and T2 systems are derived by ab initio methods. However, it is difficult to verify the reliability of the calculated results due to the lack of direct experimental support. Therefore, a data-driven reliability analysis method is proposed, which can confirm the reliability by extracting information from the relevant calculations and multiple experimental data (the vibrational level, rotational level, and molar heat capacity) of similar systems (HD, H2, D2). The results show that: 1) The potential energy curves obtained by the ab initio method can provide the full ro-vibrational energy spectrum with an accuracy of approximately 10 cm−1; 2) Macroscopic heat capacity information can be used to distinguish and calibrate the overall reliability of microscopic ro-vibrational energies; 3) For the isotopic energy level structure of hydrogen, the influence of isotopes is mainly mass effect.

Data-driven reliability assessment method of Integrated Energy Systems based on probabilistic deep learning and Gaussian mixture Model-Hidden Markov Model

Reliability analysis of IESs (Integrated Energy System) is complicated because of the complexity of system topology and dynamics and different kinds of uncertainties. Reliability is often calculated based on statistic methods, which always focus on historical performances and neglect the importance of their dynamics and structure. To overcome this problem, in this paper, a systematic framework for dynamically analysing the real-time reliability of IESs is proposed by integrating different machine learning methods and statistics. Firstly, the bootstrap-based Extreme Learning Machine is developed to forecast the conditional probability distributions of the productions of renewable energies and the energy consumptions. Then, the dynamic behaviour of IESs is simulated based on a stacked auto-encoder model, instead of using traditional mechanism-based simulation models, for improving computational efficiency. Besides, the variables representing the transient properties of natural gas pipeline networks, such as delivery pressures and flow rates, are taken as the indicators for quantifying the energy supply security in natural gas pipeline networks. The time-dependent relationships among these indicators and their statistic correlations are modelled for improving the effectiveness of the analysis results. Finally, the reliability assessment is performed by estimating the probability distribution of each functional state of the target IES. A case study of a realistic bi-directional IES is carried out to demonstrate the effectiveness of the proposed method. The results show that the method is able to effectively evaluate the reliability of IESs, which can provide useful information for system operation and management.

Towards Data-Driven Reliability Modeling for Cyber-Physical Production Systems

Abstract Reliability is one of the most important performance indicators in contemporary production facilities. Increasing reliability of manufacturing systems results in their prolonged lifetimes, and reduced maintenance and repair costs. Reliability modeling is a common technique for deriving reliability measurements and illustrating relevant fault-dependencies. There is a significant body of research focusing on hardware- and software reliability models, such as Fault Trees, Petri Nets and Markov Chains. Up until now, development of reliability models has been a labor-intensive and expert-knowledge-driven process. To remedy that, through the prevalence of data stemming from the new and technologically advanced manufacturing systems, we propose that data generated in modern manufacturing lines could be used to either automate or at least to support development of reliability models. In this paper, we elaborate on the details of our proposed framework for data-driven reliability assessment of cyber-physical production systems. We, furthermore, introduce a case study that will aid the development and testing of the proposed novel data-driven approach.