Real-time Failure Research Articles

A Spatiotemporal Locomotive Axle Temperature Prediction Approach Based on Ensemble Graph Convolutional Recurrent Unit Networks

Spatiotemporal axle temperature forecasting is crucial for real-time failure detection in locomotive control systems, significantly enhancing reliability and facilitating early maintenance. Motivated by the need for more accurate and reliable prediction models, this paper proposes a novel ensemble graph convolutional recurrent unit network. This innovative approach aims to develop a highly reliable and accurate spatiotemporal axle temperature forecasting model, thereby increasing locomotive safety and operational efficiency. The modeling structure involves three key steps: (1) the GCN module extracts and aggregates spatiotemporal temperature data and deep feature information from the raw data of different axles; (2) these features are fed into GRU and BiLSTM networks for modeling and forecasting; (3) the ICA algorithm optimizes the fusion weight coefficients to combine the forecasting results from GRU and BiLSTM, achieving superior outcomes. Comparative experiments demonstrate that the proposed model achieves RMSE values of 0.2517 °C, 0.2011 °C, and 0.2079 °C across three temperature series, respectively, indicating superior prediction accuracy and reduced errors compared to benchmark models in all experimental scenarios. The Wilcoxon signed-rank test further confirms the statistical significance of the result improvements with high confidence.

Multi-dimensional early warning of the entire supply chain of power materials based on RFID technology

Early warning system needs to process a large amount of real-time data, and carry out in-depth analysis and mining of these data to identify potential risks and hidden dangers. However, existing data processing and analysis capabilities may not be able to meet this demand. To this end, a multi-dimensional early warning of the entire supply chain of power materials based on RFID technology is designed. According to the real-time failure probability of power materials, the probability of early warning accidents is calculated and classified, and the risk is graded based on these probabilities. Through the methods of bonus points, deduction points and grade evaluation, the risk early warning indicators of different stages in the whole supply chain of power materials were quantified. The objective and rationality of risk assessment can be ensured by means of comprehensive weight. A multi-dimensional early warning system based on RFID technology is established, combining multiple linear regression model and particle swarm optimization algorithm to determine the time window of multi-dimensional early warning, and carry out dynamic monitoring and early warning of supply chain. The experimental results show that the early warning effect of the design method can reach 95% and the highest early warning effect can reach 98% at 10 seconds. The average warning error is only 2.91%, and the average warning time is only 1.34 seconds, which is more accurate in identifying the number of first-level risks, second-level risks and third-level risks.

Dynamic strengthening of UHMWPE yarns by incorporating ZrO2/PU coatings

Wrapping is an effective technique for reducing defects in fabric weaving and promoting even wetting in composite manufacturing, which is potential to enhance the impact resistance of armor materials. In this study, we impregnated UHMWPE yarns with a polyurethane (PU) compound and cured them to create composite yarns, and ZrO2 was introduced to improve the toughness. The microstructures were analyzed using three-dimensional X-ray computed tomography, optical microscopy, and scanning electron microscopy (SEM). Quasi-static and dynamic tensile experiments were conducted using a universal testing machine and Kolsky tension bar, respectively. High-speed imaging was used to capture the real-time failure process, and post-fracture analysis was performed to examine the fracture morphology. The effects of strain rate, gauge length, and the presence of PU and ZrO2 on the mechanical behavior of the yarns were investigated. The results showed that the peak force of the composite yarns with PU coatings exhibited strain-rate dependency at low strain rates (below 4.4 × 10−2 s−1) but became insensitive to strain rate in the higher range of 4.4 × 10−2 s−1 to 1500 s−1. Furthermore, the introduction of ZrO2 effectively increased the peak force, potentially by strengthening the bonding between individual fibers within the yarn.

Dynamic rock tensile failure under hydrostatic pressure analyzed with wavelet transform of acoustic emission signals

We develop a novel experimental apparatus, integrated into a dynamic testing system, for the real-time monitoring of the failure process of deep rocks. Four groups of Brazilian disc (BD) granite specimens subjected to various hydrostatic confinements (0, 10, 20, 30 MPa) are tested under different loading rates. A strain gauge and an acoustic emission (AE) transducer are attached to each specimen to obtain the real-time failure signals during the impact. The continuous wavelet transform is utilized to analyze the recorded AE signals. It is observed that tensile crack initiation and propagation effectively absorb AE signals produced by stress waves above 50 kHz, which indicates that post-failure AE signals are predominantly generated by the cracks themselves. It is also found that the proportion of AE signals within the 100–150 kHz range increases with the loading rate, suggesting that higher loading rates encourage the emergence of smaller cracks. Moreover, higher hydrostatic stress enhances the dynamic tensile strength of the rock, thereby restraining crack initiation. When the confining stress is applied, the AE signals exceeding 100 kHz are delayed in their occurrence as compare with the failure, contrasting with those within the 50–100 kHz range that appear simultaneously with the failure. This pattern indicates that cracks generate AE signals within the 50–100 kHz range during the tensile failure, while those with frequencies above 100 kHz are produced by tensile crack propagation following the failure.

A sensor-driven operations and maintenance planning approach for large-scale leased manufacturing systems

The rise of mass customisation drives manufacturers to adopt leasing agreements for their machinery rather than outright their ownership. This industrial trend results in large-scale leased manufacturing systems, where the asset conditions and maintenance requirements for a fleet of machines and/or production lines from several manufacturers are continuously monitored and managed by a lessor. In this paper, we propose an operations and maintenance planning model that explicitly models (i) dynamic real-time failure rate predictions for machines based on sensor-driven degradation data, (ii) optimal routes for maintenance teams across a number of geographically-distributed manufacturing sites, and (iii) operational outcomes. The model can handle complex maintenance and operational outcome interdependencies between multiple machines, and incorporate demand for different products, heterogeneous machine capacities and capabilities, factory topology, and the resulting production flow capacities of products. To demonstrate the model’s practicality, it is applied to three experimental case studies with various numbers of machines, operating schedules, product types, and flow capacities. The results show that the proposed model significantly outperforms the traditional reliability-based maintenance model in key metrics such as the number of unexpected failures, the cost of preventive maintenance actions, machine downtime, the percentage of unsatisfied demand, and total cost.

Load-bearing response of deep content mixing soft soil composite foundation based on the geotechnical centrifugal model test

Deep cement mixing (DCM) is an effective method of treating soft underwater soil foundations. Currently, the real-time failure development of a composite foundation cannot be obtained by most model tests of DCM pile foundation reinforcement. In this study, a circuit device for identifying pile failure and that can reflect the real-time working state of piles arranged at different positions in a soft soil composite foundation was developed. By applying this device to a geotechnical centrifugal model test, the load-bearing response process of a bank slope reinforced with a soft-soil composite foundation and DCM piles was studied. The failure development law of the DCM piles, ultimate surcharge of the bank slope, failure mode of the foundation, and the pile-soil stress ratio were determined from the test results. The law obtained in this study provides a reference for the application of DCM piles in underwater soft soil foundation engineering.

AI-driven real-time failure detection in additive manufacturing

The optimisation of 3D printing parameters for manufacturing biomedical devices is an emerging interdisciplinary field that incorporates artificial intelligence techniques such as machine learning and deep learning. In this particular study, the focus is on the fabrication of biocompatible finger splints using digital light processing 3D printing technology, followed by UV curing, to evaluate their quality. By leveraging vibration data from printers, which cannot be captured through visual inspection of layer defects, this study aims to develop a predictive model for assessing the failures of printed parts. Here, a closed-loop detection system is proposed to identify failure phenomena in 3D resin printing, combining both cloud and edge computing technologies to effectively detect and address potential failures in the printing process.

Bayesian change point prediction for downhole drilling pressures with hidden Markov models

AbstractIn the drilling of oil wells, the need to accurately detect downhole formation pressure transitions has long been established as critical for safety and economics. In this article, we examine the application of Hidden Markov Models (HMMs) to oilwell drilling processes with a focus on the real time evolution of downhole formation pressures in its partially observed state. The downhole drilling pressure system can be viewed as a nonlinear, non‐degrading stochastic process whose optimum performance is in a region in its warning state prior to random failure in time. The differential pressure system is modeled as a hidden 3 state continuous time Markov process. States 0 and 1 are not observable and represent the normally pressured (initiating ) and abnormally pressured or warning (reducing ) states respectively. State 2 is the observable failure state (from negative and loss of well control). The signal process of the evolution of differential pressure is identified in the changes in the observable rate of penetration (ROP) encoded in drilling performance data. The state and observation parameters of the HMM are estimated using the Expectation Maximization (EM) algorithm and we show, for a univariate system with a depth dependent time relationship, that the model parameter updates of the EM algorithm equation have explicit solutions. A Bayesian inference model, to determine the safety threshold of the system and early failure prediction at each sampling epoch, is thereafter proposed. The application of our stochastic model of the dynamic evolution of downhole pressures in operational time is illustrated with a hindcast case example. The analysis showed strong early indication of probable failure in real time and was validated in the field post drilling system failure that resulted in significant recovery costs. The potential to predict the differential pressure state transitions ahead of the bit represents a capability not currently available in the industry. This opens up significant opportunity for value creation from optimizing drilling operations to deliver substantial savings in well construction costs.

基于自编码神经网络的玉米籽粒收获机故障预测模型

The corn grain harvester serves as an example of complex farming machinery with a condition monitoring system that collects a lot of working condition data, making it challenging to identify the true change pattern due to the data coming from the equipment in various states. Firstly, the overall structure of the corn grain harvester is analyzed, and the common causes and mechanisms of corn grain harvester failures are analyzed, leading to the cutting table as the main research object; Secondly, by collecting historical failure data of corn grain harvester as well as real-time failure information for collation and pre-processing, eliminating interference such as noise and missing data, establishing a failure matrix, extracting internal characteristics between failure causes and establishing a mapping between failure causes and failure phenomena; Finally, the future failure phenomena of the corn grain harvester are predicted according to different failure causes. The simulation analysis results show that the self-coding neural network fault prediction model can better predict the occurrence probability and types of faults and provide data support for fault maintenance and decision making of agricultural machinery.

Assessing the Fracturing Process of Rocks Based on Burst–Brittleness Ratio (BBR) Governed by Point Load Testing

AbstractBrittleness is an intrinsic mechanical property of rock materials that has attracted significant attention to be properly quantified as it plays an important role in characterization of brittle fracturing. Endeavors have led to the establishment of many Brittleness Indices (BIs) for various rock types and widespread engineering applications. Among them, assessing burst proneness as a serious challenge in underground mining has received considerable attention. Parallel to BIs' development, various Bursting Liability Indices (BLIs) have been proposed to specifically assess coal bursting phenomenon. Despite having different names, both BI and BLI in principle have aimed at evaluating the burst–brittleness level of different rocks for different applications. In this study, the principles of burst and brittleness were discussed followed by the development of a novel so-called burst–brittleness ratio (BBR) to assess the relative burst–brittleness of rock types irrespective of their applications. To do so, the proposed BBR was governed by point load testing (PLT) which has significant advantages over the other rock testing methods used in BI estimation such as direct or indirect tensile testing. To examine the suitability of the proposed ratio, three different rock types from various geological origins including coal, granite and sandstone were selected and tested under uniaxial compressive, indirect tensile Brazilian and point loadings. The high-speed imaging technique and Acoustic Emission (AE) were utilized to characterize the cracking process (e.g., failure under shear or tension) and to monitor the real-time failure behavior of samples under different loading conditions. The resulting data revealed that the severity of strength loss in coal samples was significantly higher than that observed in other rock types particularly under uniaxial compression endorsing the validity of the proposed BBR.

On-Board Deep-Learning-Based Unmanned Aerial Vehicle Fault Cause Detection and Classification via FPGAs

With the increase in the use of unmanned aerial vehicles (UAVs)/drones, it is important to detect and identify causes of failure in real time for proper recovery from a potential crash-like scenario or postincident forensics analysis. The cause of crash could be either a fault in the sensor/actuator system, a physical damage/attack, or a cyber attack on the drone's software. In this work, we propose novel architectures based on deep convolutional and long short-term memory neural networks to detect (via autoencoder) and classify drone misoperations based on real-time sensor data. The proposed architectures are able to learn high-level features automatically from the raw sensor data. Empirical results show that our solution is able to detect (with over 90% accuracy) and classify various types of drone misoperations [with about 99% accuracy (simulation data) and up to 85% accuracy (experimental data)]. Furthermore, with the help of field programmable gate array-based hardware acceleration, we achieved a speedup of 40x ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\sim\!\text{2.6 ms}$</tex-math></inline-formula> ) for detection, while consuming half the amount of power compared with onboard GPU devices, such as NVIDIA Jetson TX2.

Testing coverage-based software modeling incorporating random effect with change point and its release

The use of computers has been increasingly prevalent in our social lives in recent years. As a result, software engineers must create trustworthy software systems. Companies often release improved versions of the core program due to the constant demand and growing competition in the software field. A variety of growth models have been created to track and measure reliability by software managers and engineers. In the testing phase, the fault content and size of the software system increase, and consequently, the fault content found and eliminated during each debugging process decreases in comparison to the fault content present at the initial stage. In this scenario, we can consider the software fault detection process to be stochastic. The fault detection process is expressed in terms of testing coverage with random effects. In this study, we construct a testing coverage-based software model incorporating random effect with change point. We have used different testing coverage functions such as Exponential and Delayed S-shaped to study the effect of randomness. Further, multi-release planning for the proposed model has been studied and validated on the real-time failure dataset from Tandem Computers with four releases. Different performance measures and goodness-of-fit have been presented using a graphical representation.

Online Failure Detection using Deep Learning in FPGA PCB Interface

This research paper is aimed to present a real-time failure detection technique while working with Field Programmable Gate Arrays (FPGA) and interfaced Printed Circuit Boards (PCBs). In this research, we explored the feasibility of currently available innovative Deep Learning (DL) algorithms to detect the defects in variety of PCBs. In our proposed technique, we trained the YOLOv5 (You Only Look Once) algorithm with a few hundreds of defective PCBs’ images, which were obtained from Kaggle, an online community of data scientists and machine learning practitioners. The advantage of using YOLOv5 is that the detection is carried out in real-time. In the next phase, after training, the algorithm undergoes validation and testing, where we tested with different images. The obtained results are promising, as the Deep Learning process successfully detects the defects on the PCBs.

Single-Wheel Failure Stability Control for Vehicle Equipped with Brake-by-Wire System

In order to solve the problem of vehicle stability control after a single-wheel brake failure in the brake-by-wire system, a control strategy of braking force redistribution with a yaw moment is proposed to ensure the braking efficiency and stability of vehicles. In this strategy, a two-layer architecture is adopted. In the upper layer control, a fault factor is introduced to represent the real-time failure degree of the wheel, and the driver’s braking intention is perceived through the pedal travel and pedal speed of the driver. The braking force redistribution algorithm of the remaining three wheels is designed based on the wheel failure degree and braking intensity. In the lower control, according to the state parameters of the vehicle, the additional yaw moment, which controls the yaw rate and the sideslip angle of the vehicle, is calculated by using the sliding mode control theory, and the yaw moment is reasonably allocated to the normal wheel. By using MATLAB/Simulink and Carsim co-simulation, different braking strength and failure types are selected for simulation analysis. The simulation results show that the proposed control strategy can improve the braking efficiency and stability of the vehicle under different braking conditions.

Application of Machine Learning Algorithms for Tool Condition Monitoring in Milling Chipboard Process.

In this article, we present a novel approach to tool condition monitoring in the chipboard milling process using machine learning algorithms. The presented study aims to address the challenges of detecting tool wear and predicting tool failure in real time, which can significantly improve the efficiency and productivity of the manufacturing process. A combination of feature engineering and machine learning techniques was applied in order to analyze 11 signals generated during the milling process. The presented approach achieved high accuracy in detecting tool wear and predicting tool failure, outperforming traditional methods. The final findings demonstrate the potential of machine learning algorithms in improving tool condition monitoring in the manufacturing industry. This study contributes to the growing body of research on the application of artificial intelligence in industrial processes. In conclusion, the presented research highlights the importance of adopting innovative approaches to address the challenges of tool condition monitoring in the manufacturing industry. The final results provide valuable insights for practitioners and researchers in the field of industrial automation and machine learning.

The Development of Exhaust Fan Housing With Ceiling Mounting For High Rise Buildings by Using DFMA

Design for manufacturing and assembly (DFMA) is widely applied in many industries to optimize the manufacturing and assembly process at the early stage of design, with the aides of the CAD model. Many researchers apply the DFMA to increase assembly efficiency, by decreasing the number of parts from a product, decreasing the manufacturing cost, and reducing assembly time. Therefore, this research applies DFMA to develop exhaust fan housing with ceiling mounting for high rise building type with the same purpose, and at the same time to justify that the method can overcome the problem of assembly time in a production line. Both designs from before and after the application of DFMA, are being compared by using finite element simulation and experimental. The simulation employs stress analysis, to predict the strength of those designs. While the experimental uses a manufacturing cost survey, real assembly time survey and failure test to show the advantages of DFMA design results. The research result shows that the DFMA method can decrease the manufacturing cost by 0.44%, and the assembly time by up to 2%, and able to withstand the entire mass of the ceiling mounting fan.

Detection of Water Vapor by Chemiluminescence

We examined the possibility of detecting water vapor by chemiluminescence using the reaction of popular “chemical light” (bis(2,4,5-trichlorophenyl-6-carbopentoxyphenyl)oxalate with H2O2). H2O2 is released from sodium percarbonate exposed to water molecules as in the oxygen bleach. The release of H2O2 by water vapor was confirmed by mass spectrometry in a vacuum. The chemiluminescence from the mixed reagents was observed when exposed to water vapor. This method opens the way to locally detect the faulty points of water barrier films and observe the real-time failure of the barrier films during bending tests of flexible packing materials. A molecular dynamics simulation was performed to study the diffusion of H2O2 molecules in polymers.

Failure diagnosis of rotating Machines for steam turbine in Cap-Djinet thermal power plant

The Cap-Djinet thermal power plant is a 1872-megawatt (MW) gas power plant located in Djinet in Algeria. The steam turbine is an important strategic machine in this plant. A malfunction of the turbine can generate a high maintenance and production cost due to the classical diagnosis method based on conditional maintenance. The important amount of information given by the different sensors keeps the diagnosis operation more difficult; it takes more time to be effective, resulting in an additional cost for failures. The present paper emphasises a new professional approach capable of diagnosing the different failures in real time. This approach is based on the combination of the Fault Tree (FT) method and the Artificial Neural Network (ANN) method. Only six failure modes have been considered in this study. Each theme has been represented by an integrated turbine sensor. The FT is used to model all the probable causes of the different failure modes. On the other hand, ANN is used to accelerate the diagnosis procedure by learning all the possible combinations of failures given by FT analysis. The ANN was tested using different failure scenarios. The learning and testing results were successfully completed and the proposed ANN was able to detect the failures. This proposed method can help the maintenance agent to find fastly, in real-time, the probable failure of the steam turbine. The precision and accuracy of this method have been validated using a one-hundred simulation test.

Cloud-Edge Architecture With Virtualized Hardware Functionality for Real-Time Diagnosis of Transients in Smart Grids

Edge Cloud is providing unprecedented opportunities for IoT and WAMS (Wide Area Monitoring Systems) in electrical grid operation. It is an orchestrated environment able to address low latency events through appropriate edge-cloud computing configurations.Transient State Estimation (TSE) is a key monitoring tool for capturing a reliable knowledge of the Smart Grid status in real-time, given the impediments introduced by the increasing penetration of Distributed Energy Resources in the energy mix. Frequency response anomalies, large scale transients, and voltage swings can be captured by TSE for real time or post failure data analytics. This work presents a cloud edge framework for the efficient calculation of TSE which, albeit its benefits, demands high computational resources at the edge (close to the measurement units) along with ultra low latency communications. The framework enables TSE as a service through the coordination of Virtual Machines (VMs) running on virtualized infrastructure and other non-virtualized physical nodes. In order to support the stringent time requirements, part of the TSE is offloaded to dedicated hardware acceleration units (FPGA). The proposed TSE framework is validated using an IEEE 30-bus, and the results show a significant superiority in terms of total latency compared to conventional cloud and edge deployments.

Identification of micro‐failure processes of HDPE‐henequen fiber composite material by using acoustic emission monitoring: Effect of fiber surface modification

AbstractIt is well known that fiber‐matrix interface region is crucial to ensure an efficient stress transfer between matrix and reinforcement and that it controls the composite behavior when subjected to external loads. In this experimental study, micro‐damage mechanisms of HDPE/surface modified henequen fiber reinforced composites are investigated in tensile quasi‐static mode. Two essential parameters of the materials composition were taken as main indicators: fiber length and quality of the fiber matrix adhesion. Essential Work of Fracture theory (EWF) was applied on double edge notch tensile (DENT) specimen geometry. Acoustic Emission (AE) technique was coupled to composite samples to obtain the characteristics of stress waves signals to monitor in real time damage evolution and failure mechanisms detection. Results demonstrated the substantial effect of chemical bonds interaction in the interface area on the composite mechanical properties. The total fracture work (Wf) exhibited an increasing value in both components, We and Wp, in chemically treated fiber composites. Likewise, mechanical properties are reported to be higher by 37% in maximum load in such materials. Elastic waves detected by AE were identified and correlated to micromechanical events during composite damage sequence. Most of the signals corresponded to microcracks (signals of low amplitude, short duration), propagation of microcracks (signals of medium amplitude), fracture of fibers and matrix (signals of long duration and high amplitude). AE signals detected exhibiting higher energy were associated to an enhanced fiber/matrix interface.