Life Prediction Method Research Articles

A new approach to multiaxial fatigue life prediction: A multi-dimensional multi-scale composite neural network with multi-depth

Many multiaxial fatigue life prediction methods tend to increase additional parameters and model depth as the complexity of the loading path increases. This leads to issues such as poor model robustness, limited flexibility, and single-dimensional approaches. In this study, a multi-dimensional multi-scale composite neural network with multi-depth is proposed to address these challenges and enhance prediction accuracy. Initially, physical and sensitive features serve as input data for the proposed model, to enhance the richness of input features. Subsequently, a multi-dimensional feature extraction module is deployed to extract feature information from the composite data. To process these features, an improved multi-domain query cascaded transformer network (IMQCT) is employed as the feature processing module of the proposed model. The proposed model is verified to have better prediction accuracy and extrapolation capability by using experimental data from nine materials and comparing its performance with six machine learning models.

Research on the Remaining Useful Life Prediction Method of Energy Storage Battery Based on Multimodel Integration.

The remaining useful life (RUL) of lithium-ion batteries (LIBs) needs to be accurately predicted to enhance equipment safety and battery management system design. Currently, a single machine learning approach (including an improved machine learning approach) has poor generalization performance due to stochasticity, and the combined prediction approach lacks sufficient theoretical support at the same time. In this paper, we first analyze the prediction principles and applicability of models such as long and short-term memory networks and random forests, and then propose a method for predicting the RUL of batteries based on the integration of multiple-model, and finally validate the proposed model by using experimental data. The experimental results show that (1) for the proposed model, in the best case, the root-mean-square error (RMSE) does not exceed 0.14%, which has a stronger generalization; (2) for the comparison with the single model used, the average RMSE is reduced by 46.2%, 43.7%, and 80.6%, which has a better fitting performance. These results show that the model has good prediction accuracy and application prospects for predicting the RUL of energy storage batteries.

SO2 removal performance and breakthrough prediction of activated carbon-based chemical filters for electronics cleanroom air purification

Electronic cleanrooms have strict control requirements for SO2 concentration due its significant influence on product yield, and application of chemical filters is the main measure of SO2 control. In this study, 8 activated carbon (AC)-based chemical filters were collected and tested for their SO2 purification performance and resistance. These filters had very similar initial efficiencies (all >95 %) but large differences in 20 % breakthrough adsorption capacity (49–294 g at 9 ppm and 0.35 m/s). The adsorption capacity is found to be positively correlated with pleat density, which is a main structural parameter of pleated chemical filter. Meanwhile, the filter resistance (9.3–34 Pa at 0.35 m/s) demonstrates a decreasing and then increasing trend with increasing pleat density. This indicated the existence of an optimal range of pleat numbers that can be used to achieve a balance between resistance and adsorption capacity. A comprehensive performance evaluation index (Z′-index) is proposed for assessing the integrated resistant and adsorption capacity performance of chemical filters (19–210 g/W for 8 samples). A lumped parameter model was proposed for predicting the breakthrough curves at low concentration based on the traditional semi-empirical breakthrough models. The prediction deviation of the breakthrough curve at real cleanroom scenario (2 ppb) is less than 15 %. In addition, a rapid filter life prediction method based on filter design parameters is developed from the lumped parameter model. This work helps the understanding of the impact of filter structure parameters on performance and facilitate the development of high-performance chemical filters for SO2 contaminant control in electronic cleanrooms.

Determination of equivalent crack length for voids in metallic alloys based on continuum damage mechanics modeling

Metals contain different kinds of spatial defects significantly reducing the fatigue performance. Accurate defect-based life prediction methods are required to improve structural integrity analysis and life design reliability. The present work developed a method to determine the equivalent crack length for defects based on continuum damage mechanics. The spatial effects of defects in the calculation of the equivalent crack were quantified and verified by fatigue tests. It is confirmed that the equivalent crack length of both slit and elliptic defects is affected not only by the defect geometry but also by the applied load. Based on the equivalent crack and combined with Paris’ law, the predicted fatigue lives of various slit holes under different loading orientations are within the double scatter band and are much more accurate than the conventional projection method.

An aircraft engine remaining useful life prediction method based on predictive vector angle minimization and feature fusion gate improved transformer model

Remaining useful life (RUL) prediction is crucial for achieving intelligent and predictive maintenance of aircraft engines. In practical applications, advance prediction values smaller than the true values can prevent serious deferred maintenance accidents. Using asymmetric loss functions directly leads to noticeable accuracy degradation, and existing methods often fail to satisfy accuracy and advance prediction requirements. To address this problem, this paper proposes a novel RUL prediction method based on the Prediction Vector Angle (PVA) minimization and Feature Fusion Gate (FFG) improved Transformer network. Specifically, the FFG is proposed to enhance Transformer prediction accuracy by dynamically fusing global and local features. The concept of PVA is first introduced based on the tilting properties of the RUL descent process. The target of the prediction model is cleverly transformed from error minimization to PVA minimization through the cosine similarity loss function. Various experiments on the CMAPSS dataset demonstrate the effectiveness of the proposed method in achieving high accuracy and advanced prediction. Compared to the state-of-the-art method, RMSE is reduced by at least 2.94 % and Score by 7.00 %. Finally, the PVA minimization mechanism significantly improves long short-term memory and convolutional neural network performance. The proposed method is noteworthy for its superiority and applicability.

Life prediction method of battery energy storage system in frequency modulation application

Abstract To tackle the challenge of lifespan reduction in lithium batteries during frequency modulation, this study introduces a novel Remaining Useful Life (RUL) prediction methodology. The proposed approach integrates Variational Mode Decomposition (VMD) with Gated Recurrent Unit (GRU) networks, thereby efficiently synthesizing data from both operational parameters and capacity metrics. Firstly, VMD is utilized to decompose the initial capacity data of lithium batteries into separate components characterized by high and low frequencies. Subsequently, for the high-frequency elements, GRU is employed for rolling iterative prediction, while for the low-frequency elements, features are extracted from the operational data and inputted into GRU for prediction. Finally, the component prediction results are restored to obtain capacity prediction values. Validation conducted using the NASA dataset shows that the root mean square error of capacity prediction is minimized to 0.0122, with a corresponding minimum average absolute error of 0.0096, and RUL prediction errors are generally within 2 cycles.

Experimental study on low-cycle fatigue performance of high-strength bolted shear connections

In strong earthquakes, low-cycle fatigue fractures may occur in the bolted connections of steel braced frames. Ensuring a high low-cycle fatigue life for high-strength bolted connections is of paramount significance in enhancing the overall safety and collapse resistance of steel structures. Despite substantial research on plate fatigue failure within bolted connections, there remains a research gap regarding low-cycle fatigue failure of high-strength bolts in these connections. This study aims to address the aforementioned shortcomings by conducting 34 sets of low-cycle fatigue tests on high-strength bolted connections. The test specimens all experienced low-cycle shear fatigue fracture at the threaded or unthreaded portion of the bolt shank as the dominant failure mode. Residual deformations were observed in the holes of connection plates. Based on the test data, ultimate shear capacity of a single shear connection is around 0.61Fy (tensile yield bearing capacity of bolt) when the shear force passes through the unthreaded portion and drops to 0.54Fy for the shear force at the thread. The untreated Q355 (minimum yield strength of 355 MPa) steel surface has an average anti-slip coefficient of around 0.37, while milling reduces it to approximately 0.30. The influence patterns of loading amplitude, bolt material, minimum plate thickness, bolt diameter, shear force position and connection type on low-cycle fatigue life are given by range analysis. Among them, variance analysis shows that loading amplitude and bolt material are the primary influencing factors. The correlation coefficient between loading amplitude and low-cycle fatigue life is −0.92, indicating a strong negative correlation. The degradation amount and rate of anti-slip capacity and shear capacity of bolted connections are also investigated. To propose reliable life prediction method, three models were used to establish the relationship between dimensionless shear deformation and low-cycle fatigue life of 10.9-grade high-strength bolted connections. It is not appropriate to use the fatigue categories defined in current standards to predict the low-cycle shear fatigue life of high-strength bolted connections.

Low-Cycle Fatigue Damage Mechanism and Life Prediction of High-Strength Compacted Graphite Cast Iron at Different Temperatures.

Tensile and low-cycle fatigue tests of high-strength compacted graphite cast iron (CGI, RuT450) were carried out at 25 °C, 400 °C, and 500 °C, respectively. The results show that with the increase in temperature, the tensile strength decreases slowly and then decreases rapidly. The fatigue life decreases, and the life reduction increases at high temperature and high strain amplitude. The oxide layer appears around the graphite and cracks at high temperature, and the dependence of crack propagation on ferrite gradually decreases. With the increase in strain amplitude, the initial cyclic stress of compacted graphite cast iron increases at three temperatures, and the cyclic hardening phenomenon is obvious. The fatigue life prediction method based on the energy method and damage mechanism for compacted graphite cast iron is summarized and proposed after comparing and analyzing a large amount of fatigue data.

A multi-source data fusion driven power field effect transistor health state assessment and remaining useful life prediction method

Abstract The metal oxide semiconductor field effect transistor (MOSFET) is subjected to various stresses due to the external and internal operating environments, leading to ageing and failure of the device. With multiple degradation mechanisms, a single piece of information can no longer fully reflect the health state of MOSFETs, so how to use multi-source data to characterise the state of the device and predict the remaining useful life (RUL) is an issue worth exploring. To address this problem, a method for constructing health indicators (HI) as well as predicting RUL using multi-source data is proposed. In this method, firstly, the features are computed by selecting the appropriate ageing signal from the ageing mechanism. Secondly, the extracted features are filtered using Pearson’s algorithm to find the features that are strongly correlated with longevity. Then, the filtered features are merged by dimensionality reduction using the kernel principal component analysis algorithm and the HI is constructed from the reduced result. Finally, an unsupervised clustering algorithm is used to classify the states based on the data distribution in HI, and the filtered features are used as input to the grey wolf optimisation bidirectional long short-term memory neural network to predict the RUL of the device. In this paper, the proposed method is validated using data from the MOSFET Accelerated Aging Experiment at the NASA Ames Centre of Excellence for Prediction. The results show that the method is able to achieve good results in health state assessment and RUL prediction of MOSFETs. The proposed method is an effective and comprehensive strategy that is more helpful for the operation and maintenance of electronics.

Research on remaining useful life prediction methods for rolling bearings based on a novel gated unit

Research on remaining useful life prediction methods for rolling bearings based on a novel gated unit

A real-time prediction method for PEMFC life under actual operating conditions

The life prediction of proton exchange membrane fuel cell (PEMFC) is one of the methods to reduce the cost of using fuel cells. This helps to solve the problem of high using costs for fuel cell vehicles, which makes it difficult to commercialize them on a large scale. However, there is still a lack of real-time life prediction methods that can be applied to fuel cell vehicles. In this article, the first step is conducting actual road vehicle life experiments, which include urban roads, suburban roads, and highways. Collect experimental data and preprocess it to obtain a dataset of actual vehicle life. The second step is to divide the output power of the fuel cell into multiple operating conditions, calculate the voltage decay rate of each operating condition, and establish a real-time prediction (RTP) model for the fuel cell life. The third step is conducting fuel cell life prediction. The results show that compared with the Pei method, the RMSE and MAPE of the proposed RTP method for fuel cell life have decreased by 21% and 14% in the long-term prediction of 100 h, and the average reduction is 41% and 35% in the short-term prediction of 10 h.

A Small Database with an Adaptive Data Selection Method for Solder Joint Fatigue Life Prediction in Advanced Packaging.

There has always been high interest in predicting the solder joint fatigue life in advanced packaging with high accuracy and efficiency. Artificial Intelligence Plus (AI+) is becoming increasingly popular as computational facilities continue to develop. This study will introduce machine learning (a core component of AI). With machine learning, metamodels that approximate the attributes of systems or functions are created to predict the fatigue life of advanced packaging. However, the prediction ability is highly dependent on the size and distribution of the training data. Increasing the amount of training data is the most intuitive approach to improve prediction performance, but this implies a higher computational cost. In this research, the adaptive sampling methods are applied to build the machine learning model with a small dataset sampled from an existing database. The performance of the model will be visualized using predefined criteria. Moreover, ensemble learning can be used to improve the performance of AI models after they have been fully trained.

An enhanced fatigue damage model based on strength degradation of composite materials

AbstractAn enhanced nonlinear fatigue damage cumulative model proposed is based on the strength degradation characteristics of composites, aiming to investigate damage progression under fatigue loading. Building upon this foundation, given the assumption of a linear correlation between fatigue cumulative damage and stress level, a methodology is presented for extrapolating the damage curve of untested stress levels from that of tested stress levels. The model substantiates its reliability by validating against experimental data from several distinct material types. The reliability of the model has been validated using experimental data from multiple groups of materials. The experimental results indicate that the model can effectively reflect the fatigue damage development characteristics of composite materials. Simultaneously, the predicted stress levels derived from the proposed methodology show lesser deviation from the fitted data. Finally, a life prediction method founded on the proposed model is introduced, validated for its high prediction accuracy through experimentation.

Study on the wear performance of V-combined tandem seals

In this thesis, a numerical simulation model of tandem reciprocating sealing system is established for a V-type combination seal for large-size piston rods according to the material and structural characteristics of the seals to achieve the quantitative calculation of leakage rate, friction and other sealing performance indexes; Based on Rhee's wear theory, combined with the hybrid lubrication model, a wear life prediction method for V-type combination seals is proposed to predict and analyse the evolution of lip profile with working time, and a V-type combination seal bench is built for experimental verification.The results show that: with the increase of lip wear, the contact pressure decreases, the contact width increases, the wear is maximal at the lip, the friction of seals decreases gradually, the leakage rate increases gradually, and the total leakage meets the working requirements. Comparison of experimental and simulation results with high degree of overlap, the life prediction method has high reliability; this paper proves that, in 300 h service cycle, the V-type combination seals have good abrasion resistance and wear compensation ability, the seals have good sealing performance, meet the requirements of work use. This study provides a reference for predicting the wear of V-combination tandem seals.

Prediction of remaining useful life for rolling bearing based on ISOMAP and multi-head self-attention with gated recurrent unit

The remaining useful life of rolling element bearing affects the reliability and stability of the machine. Accurately predicting the remaining useful life of rolling element bearings is always necessary to make maintenance decisions in practical engineering. However, the remaining useful life prediction methods using traditional machine learning are incompetent for the task of training without labels, which consumes computational cost, financial resources, and labor. Therefore, a gated recurrent unit model based on feature dimensionality reduction and the multi-head self-attention mechanism (MHGRU) is established, and a homology algorithm is proposed to train the model and make the prediction. The isometric feature mapping algorithm and a homology algorithm are used in the model training, which incorporates the advantages of appreciable computational efficiency and preservation of automatic labeling for training in engineering. First, 24 basic features are extracted from the life-cycle vibration signals of rolling element bearings to reconstruct fusion features by the isometric feature mapping algorithm, which aims to reduce the feature dimension and improve computational efficiency. Since the multi-head self-attention mechanism in the MHGRU has the ability to comprehensively highlight the attention coefficient for long-term fusion features at different moments, it gives the gated recurrent unit considerable performance in reducing the computational complexity of extremely long time series and improving the accuracy of prediction results. In addition, two open-source IEEE PHM Challenge and XJTU-SY bearing datasets sampled by the sensor are adopted to assess the prediction performance of the MHGRU with homology algorithm. Finally, some existing remaining useful life prediction approaches of rolling element bearings are used for comparison. The experimental results show that MHGRU is suitable for the remaining useful life prediction of rolling element bearings and is superior to other models.

A data augmentation method for multiaxial fatigue life prediction based on physics-informed tabular generative adversarial network

Abstract Currently, in the multiaxial fatigue life prediction problem, machine learning (ML) algorithms are still limited by small samples. For ML tasks requiring large amounts of data, this paper proposes a Conditional Tabular Generative Adversarial Network (CTGAN) model using a physical equation as a generation constraint. The proposed method can make synthetic samples satisfy specific physical knowledge. The method is evaluated using two datasets of different materials. ML-based algorithms verify the feasibility of using synthetic datasets for life prediction. Verification analysis shows that the proposed method successfully synthesizes high-quality multiaxial fatigue datasets and effectively improves the prediction accuracy of ML algorithms.

Adaptive remaining useful life prediction method for aircraft engines under variable operational conditions based on multi-channel transformer

Adaptive remaining useful life prediction method for aircraft engines under variable operational conditions based on multi-channel transformer

Life prediction based on fatigue-environment damage under multiaxial thermo-mechanical loading

In this paper, the competition mechanism between protectiveness and harmfulness of oxidation, as well as the effects of cycle period, mean stress and temperature on the damage behavior of TA15 titanium alloy are revealed under multiaxial thermo-mechanical fatigue loading. Based on the damage mechanism, a high-temperature environmental damage model is proposed under multiaxial thermo-mechanical fatigue loading first, then a life prediction method is developed by combining a multiaxial fatigue damage model. The experimental verification results under uniaxial and multiaxial thermo-mechanical fatigue loadings showed that almost all of the prediction errors are within a factor of 2.

Creep–fatigue life prediction of a titanium alloy deep-sea submersible using a continuum damage mechanics-informed BP neural network model

Deep-sea submersibles are subjected to trapezoidal wave loading during operation, which can cause creep–fatigue failure of pressure hull structures. To ensure structural safety, it is necessary to develop accurate and efficient creep–fatigue life prediction methods. This paper proposes a continuum damage mechanics-informed BPNN approach for predicting the creep–fatigue life of titanium alloy submersibles. To obtain a reliable sample set, the creep–fatigue life is calculated based on the titanium alloy room-temperature creep–fatigue damage model and numerical analysis methods proposed by the authors. The operating duration, operating pressure, axial stress, and circumferential stress were selected as the input variables, whereas the number of cycles and actual operating duration were chosen as the output variables. The proposed model fills the current gap of a lack of research on the creep‒fatigue life prediction of titanium alloy deep-sea submersibles via machine learning methods. The research results show that the neural network model established in this paper can quickly and accurately predict the creep–fatigue life of titanium alloy deep-sea submersibles. A sensitivity analysis of the input parameters revealed that the creep–fatigue life is more sensitive to the operational pressure than to the dwell time.

A weakest link theory‐based probabilistic fatigue life prediction method for the turbine disc considering the influence of the number of critical sections

A weakest link theory‐based probabilistic fatigue life prediction method for the turbine disc considering the influence of the number of critical sections