Joint Learning of Failure Mode Recognition and Prognostics for Degradation Processes.

Real-time monitoring of tool health is of great importance for the production of mechanical equipment. The sensor signals that monitor the health of the tool are indicative of the tool wear. The sensor signals collected are typically acoustic signals, force signals, vibration signals and spindle power signals. The acquired signals are first pre-processed and then feature decomposed to extract time domain frequency domain and time-frequency domain features. Multiple sensor signals can effectively increase the accuracy of monitoring, but can produce a large amount of redundant information. It is particularly important to match the number of sensor signals and feature extraction methods to complement each other and combine them organically.KeywordsReal-time monitoringSensor signalsTool wear

Markov models are commonly used for unavailability analysis of redundant systems. However, due to the exploding states of Markov models for redundant systems, the states need to be merged to simplify the computation, which is called micro-Markov models. However, how to derive the failure rates and repair rates of the newly developed micro-Markov models has not been studied thoroughly. Therefore, this paper proposes detailed explanations and rules to derive the static unavailability by the micro-Markov models for thek-out-of-n:G systems with multiple failure modes. Firstly, two properties about applying the Markov models to the repairable system with independent multiple failure modes are presented. Based on these two properties, two rules are proposed for implementing the micro-Markov models. The micro-Markov models provide the exact same results for the repairablek-out-of-n:G system with multiple independent failure modes and repair mechanisms and approximate results for systems with multiple hybrid failure modes. A case study of safety integrity verification for safety instrumented systems is provided to illustrate the application of the proposed method. The conceptual comparison and numerical examples demonstrate the reasonability and usefulness of the proposed micro-Markov models.

In this paper, we propose the "trace data analytics" for classifying fault conditions from multivariate time series sensor signals using well-known deep CNN models. In our approach, multiple sensor signals are converted into two dimensional representations using the proposed conversion methods to optimize the classification performance. Many studies on the prediction of manufacturing results using sensor signals have been conducted in the field of fault detection and classification for display and semiconductor manufacturing processes. It is challenging to apply machine learning to real-life manufacturing problems due to practical limitations, class imbalance and data insufficiency, which also make it difficult to produce a generalized model. To overcome these challenges, we propose using omni-supervised learning but with a new approach to knowledge distillation that ensembles predictions from multiple instantiations of a CNN model of synthetically generated data samples from a deep generative model. Our experiment results show that the fault classification accuracy improves substantially by applying trace data analytics to manufacturing data from display fabrication lines. The results also show that the quality of trained CNN models using the proposed knowledge distillation is maintained steadily and stably.

Portable electronic products subjected to transient shock may exhibit multiple failure modes in vicinity of the interconnects. Failure is often diagnosed by loss of functionality using techniques including the built-in self test, which provides limited insight into reliability and remaining useful life. In this paper prognostic framework for electronic systems has been developed with neural network based self organizing maps. The presented approach resides in the pre-failure space with a focus on electronic systems with multiple failure modes. Unsupervised learning of the neural net has been used to train the neural net for identification of individual failure modes. Feature vectors have been developed based on damage pre-cursors from time-spectral measurements. The clustered damage pre-cursors have been correlated with failure modes of the underlying damage. Several chip-scale packages have been studied, with leadfree second-level interconnects including SAC305, SAC405 alloys. Transient strain has been measured during the drop-event using digital image correlation and high-speed cameras operating at 100,000 fps. Continuity has been monitored simultaneously for failure identification. In addition, explicit finite element models have been developed and various kinds of failure modes have been simulated such as solder ball cracking, trace fracture, package falloff and solder ball failure. Fault modes predicted by simulation based pre-cursors have been correlated with those from experimental data. Activation of different neurons in the lattice for various failure modes and combinations of failure modes has been demonstrated. Previously the authors have developed techniques based on statistical pattern recognition for leading indication of impending failure and detection of damage initiation and progression [Lall 2006 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a</sup> , 2007 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a</sup> , 2008]. Early classification of multiple failure modes in the pre-failure space is new.

Despite the substantive literature on remaining useful life (RUL) prediction, less attention is paid to the influence of epistemic uncertainty and aleatory uncertainty in multiple failure behaviors in the accuracy of RUL. The research question in this study was: can uncertainties be quantified in predicting the RUL of systems with multiple failure modes? The first objective was to quantify the uncertainties in the prediction of RUL, considering known multiple failure modes. This objective used vibration data from accelerated degradation experiments of rolling element bearings. The second objective was to calculate the uncertainties in the prediction of RUL, considering the multiple failure modes as unknown. The experimental data used in this objective were from run-to-failure tests of Li-ion batteries. An analysis was performed on how the uncertainties affect the RUL prediction in systems with known multiple failure modes and systems where the multiple failure modes were unknown. A Bayesian neural network (BNN) was used to quantify epistemic and aleatory uncertainty while predicting RUL. The results of the qualitative uncertainties on RUL in systems with multiple failure modes were presented and discussed. Also, the study yielded an RUL uncertainty quantification model for multiple failure modes. The proposed framework's performance in the RUL prediction was demonstrated. Finally, the epistemic and aleatory uncertainties were quantified in the system's RUL. It was shown that systems that fail due to the same failure mode tend to have similar uncertainty values over time. The results in this paper may lead to the design of more reliable systems that exhibit multiple failure modes.

Many modern structural systems usually consist of multiple failure modes. One failure mode may affect other failure modes due to the same working environment and random input variables, meaning that these failure modes are not independent. The assumption of independence among failure modes simplifies the calculation of reliability of structural systems, but it adds approximation error. Different from the dependence between two failure modes, the dependence of multiple failure modes is more complicated. In this paper, on the basis of vine copula function, which is a flexible tool to describe the multivariate dependence, the dependence of multiple failure modes of the structural systems is mainly studied and analyzed. Then Rosenblatt transformation and Monte Carlo simulation method are utilized to evaluate the reliability of structural systems. Furthermore, in order to research the importance of failure modes, two indices combined with empirical copula functions are extended, which can quantitatively measure the importance of failure modes. The multiple failure modes can be ranked based on proposed importance analysis, and the ones that have greater influence on the system can be found out, thus simplifying the system analysis. Finally, in order to confirm the applicability and rationality of the proposed method, an engineering case about a mechanism system with five failure modes is presented.

An anomaly detection and failure mode classification method has been developed for electronic assemblies with multiple failure modes. The presented prognostic health management method targets the pre-failure space of the electronic assembly life to trigger repair or replacement of impending failures. Presently, health monitoring systems focus on reactive diagnostic detection of failure modes. Examples of diagnostic detection include the built in self test and on-board diagnostics. In this paper, damage pre-cursors from time-spectral measurements of the electronic assemblies has been measured under applied vibration and shock stimulus. The time-evolution of spectral content of the damage pre-cursors has been studied using joint time frequency analysis in a full-field manner on the printed circuit assembly. Frequency moments have been used to build a feature vector. Evolution of the feature vector with damage initiation and progression has been studied under shock and vibration. The feature vector from multiple locations in the board assemblies has been mapped into a de-correlated feature space using Sammon’s mapping. Several chip-scale packages have been studied, with SAC305 and SAC405 leadfree second-level interconnects. Transient strain has been measured during the drop-event using digital image correlation and high-speed cameras operating at 100,000 fps. Continuity has been monitored simultaneously for failure identification. In addition, explicit finite element models have been developed and various kinds of failure modes have been simulated such as solder ball cracking, trace fracture, package falloff and solder ball failure. The neural net has been trained using simulated data-sets created from error-seeded models with specific failure modes. The neural net has then been used to identify and classify the failure modes in board assemblies experimentally. Supervised learning of multilayer neural net in conjunction with parity has been used to identify the hard-separation boundaries between failure mode clusters in the de-correlated feature space. The assemblies have been cross-sectioned to verify the identified failure modes. Cross-sections indicate that the experimentally measured failures modes correlate well with the position of the cluster in the de-correlated feature space.

Offshore system safety and reliability considering microbial influenced multiple failure modes and their interdependencies

In recent years, time‐variant reliability issues have garnered widespread attention, particularly due to their consideration of the cumulative effects of material degradation and the time dependency of stochastic process loads. Methods based on the outcrossing rate are typically employed for time‐variant reliability analysis. However, they fail to effectively capture the correlations among multiple failure modes. To address this limitation, this paper proposes a time‐variant reliability analysis method that accounts for the correlation of multiple failure modes by integrating copula and PHI2. The method first utilizes Copula to analyze the correlations among various failure modes, followed by applying the PHI2 method for the time‐dependent reliability analysis of each individual failure mode. Ultimately, case studies confirm that the proposed method proficiently evaluates the time‐variant reliability of systems with multiple failure modes, yielding a maximum relative error below 5% compared to MCS and significant performance enhancements.

Taking the main shaft system of hydraulic turbine-generator unit as an object,the nonlinear dynamic equation is established by the finite element method.Based on the equation,the nonlinear vibration characteristics of the system are analyzed and the excitation frequencies of the system are obtained.Considering the randomness of the structure parameters,material parameters,hydraulic parameters and electromagnetic parameters of system,the gradients of natural frequencies to the random parameters are obtained.According to the criterion that the absolute values of difference between the natural frequencies and excitation frequencies of system should be less than specific values,the failure probability with one single failure mode and the combined failure probability with two failure modes are calculated by dividing the variable space into several simple areas.The nonlinear vibration reliability model of the main shaft system of hydraulic turbine-generator units with correlated failure modes is constructed and the method by which the nonlinear vibration reliability of system with multiple failure modes can be analyzed is proposed.An example is presented and the nonlinear vibration reliability model is proved by the Monte Carlo method.

SUMMARY & CONCLUSIONSIn the realm of reliability testing, one of the fundamental challenges lies in the identification and selection of the most critical failure modes and corresponding performance parameters. With a vast array of potential failure mechanisms and performance metrics, engineers often struggle to pinpoint those that truly reflect real-world degradation and field performance. Traditionally, this selection process is guided by expert judgment and established references such as the NSWC Mechanical Handbook, and other well-defined handbooks. However, practical constraints including limited time, budget, and test resources make exhaustive testing for all potential failure modes unfeasible.This paper introduces the Critical Failure Mode Prioritization (CFMP) methodology for reliability test design. Unlike traditional approaches that rely heavily on expert opinion and static reference standards, CFMP integrates real-world operating conditions and focuses on prioritizing the most significant and probable failure modes. By leveraging tools such as Parameter Diagram (P-Diagram), Design Failure Mode and Effects Analysis (DFMEA), actual field usage data, and historical field failure data, this methodology moves beyond subjective biases. It identifies and models failure modes based on their occurrences, Mean Time to Failure (MTTF), and percentage of contributions, thereby enhancing the relevance and accuracy of the testing process.The key distinction of the CFMP approach lies in its ability to handle products with multiple failure modes and mechanisms. It emphasizes identifying the critical failure mode or mechanism that could lead to system failure as early as possible. This method is more comprehensive, focusing on end-to-end processes and prioritizing sub-assembly and system-level testing from the end user's perspective, rather than primarily from an engineer’s point of view.This paper addresses the highly relevant topic of prioritizing different failure modes in reliability testing. It mentions relevant exemplary data, especially on the side of requirements.Moreover, the CFMP methodology is grounded in core reliability engineering principles. By incorporating physics-of-failure (POF) analysis to explore optimal failure mechanisms, degradation modeling to forecast failure progression, and preventive maintenance strategies for proactive decision-making, CFMP ensures that test conditions are aligned with actual usage profiles. This results in more accurate, relevant, and efficient testing, directly supporting the prediction of field performance and facilitating better-informed decisions.

During the product development processes, components and subsystems go through test-fix-test cycles until the final configurations surpass the design life to exceed the reliability goals. When failure modes are uncovered before reaching the design life, design configurations are improved to eliminate failures. Early failures can be mixed with several failure modes commonly lead to lower shape parameters. Systems that are mature that survive close to design life may have higher and steeper slope parameters. During the component-subsystem development process and design maturity, shape parameter changes associated with reliability growth are illustrated with examples. As designs reach or surpass design life with design maturity, failure modes can be single out to unique shape parameters. Separating different shapes and patterns of Weibull curves help identifying different failure modes thereby leading to failure mode-wise reliability analysis. Such analysis help prioritizing the fixing process of each failure mode based on returns on investment.On the other hand, if a product was launched before it reaches the design life and meets reliability goals, field failures become inevitable. Such field failures can also be modeled with Weibull Analysis. Parts that fail short of the design life are commonly called under-developed designs. Such designs can be mixed with multiple failure modes. Different shapes and shift of curves can be strategically isolated for separating design weaknesses and investigations before catastrophic failures erase evidence of root causes. For example, shape parameters and patterns of underdeveloped parts can be helpful in isolating manufacturing issues from design issues. This paper addresses how to perform data analysis in order to strategically enhance the further use of Weibull plots, its shapes and pattern recognition for speedier product development and customer satisfaction.

The failure-in-time (FIT) rate is widely used to quantify the reliability of a electronic component. It fails to indicate the portion of the failures due to either environmental or electrical stresses or issues that are related to process/handing, manufacturing and applications. To meet this end, FIT-based corrective action driven metrics are proposed to link the failure mode (FM) with components and non-component faults. First the conventional failure mode pareto is reviewed and its deficiency is discussed. Then a new index called the failure mode rate (FMR) is introduced to monitor the FM trend and evaluate the effectiveness of corrective actions (C/As). Based on the FMR, the FIT rate is extended to non-component failure mode and further to individual failure mode in predicting the reliability of electronic products. The extended FIT rate enables product designers to narrow down the root-cause of the failure, identify the C/A ownership, and estimate the MTBF improvement. The new metrics provide a guideline for prioritizing resources to attack the critical failures with the minimum cost.

Failure-mode importance measures in structural system with multiple failure modes and its estimation using copula

Artificial intelligence (AI) and big data profoundly impact people’s way of life and way of thinking, and college ideological and political education (IPE) has gradually entered the era of online education. On account of this, this study designs an online education algorithm based on AI technology to help teachers better understand the situation of students’ online IPE teaching and improve the management of IPE in universities. Firstly, the learning features of students are extracted through the Back Propagation Neural Network (BP) model. This model summarizes the shortcomings of feature extraction in machine learning, and can simultaneously obtain depth information from the signals of multiple sensors, thus increasing the overall algorithm classification accuracy. Secondly, combined with the human behavior recognition model, the status and behavior of students’ IPE teaching can be obtained in real-time from students’ listening devices. Finally, the algorithm’s classification performance is evaluated by experiments and compared with the designed model. The results reveal that the recognition accuracy of the designed classification algorithms for the sample students is 98.59%, 98.99%, 99.21%, 100%, 97.10%, 95.61%, and 100%, respectively. In addition, comparing the algorithm with similar recognition algorithms, its index values of accuracy and precision are 97.83% and 97.82%, respectively, which are better than similar classification algorithms. Finally, through the experimental samples, the accuracy of the human recognition model is tested and compared with other recognition models. The results reveal that the designed model has high recognition accuracy. This study is of great significance for improving teachers’ innovative IPE methods and optimizing the management level of online IPE teaching.