基于MoG-HMM的齿轮箱状态识别与剩余使用寿命预测研究

提出了基于混合高斯隐马尔可夫模型的齿轮箱状态识别与剩余使用寿命预测新方法。建立了基于聚类评价指标的状态数优化方法,通过计算待识别特征向量的概率值来识别齿轮箱当前状态。在状态识别的基础上,提出了剩余使用寿命计算方法。最后,利用齿轮箱全寿命实验数据进行验证,结果表明,该方法可以有效的识别齿轮箱状态并实现了剩余使用寿命预测,平均预测正确率为90.94%,为齿轮箱的健康管理提供了科学依据。     

基于混合高斯输出贝叶斯信念网络模型的设备退化状态识别与剩余使用寿命预测方法研究

提出了基于混合高斯输出贝叶斯信念网络模型的设备退化状态识别与剩余使用寿命预测新方法,将变量消元和期望最大化算法相结合对模型进行推理,应用聚类评价指标对状态数进行优化,通过计算待识别特征向量的概率值来确定设备当前的退化状态,在退化状态识别的基础上,提出了剩余使用寿命预测方法。最后,分别应用50组轴承全寿命仿真数据和3组轴承全寿命实验数据对模型进行验证。结果表明,该模型可有效地识别设备的退化状态并对剩余使用寿命进行预测。     

Machine and component residual life estimation through the application of neural networks

This paper concerns the use of neural networks for predicting the residual life of machines and components. In addition, the advantage of using condition-monitoring data to enhance the predictive capability of these neural networks was also investigated. A number of neural network variations were trained and tested with the data of two different reliability-related datasets. The first dataset represents the renewal case where the failed unit is repaired and restored to a good-as-new condition. Data were collected in the laboratory by subjecting a series of similar test pieces to fatigue loading with a hydraulic actuator. The average prediction error of the various neural networks being compared varied from 431 to 841 s on this dataset, where test pieces had a characteristic life of 8971 s. The second dataset were collected from a group of pumps used to circulate a water and magnetite solution within a plant. The data therefore originated from a repaired system affected by reliability degradation. When optimized, the multi-layer perceptron neural networks trained with the Levenberg-Marquardt algorithm and the general regression neural network produced a sum-of-squares error within 11.1% of each other for the renewal dataset. The small number of inputs and poorly mapped input space on the second dataset meant that much larger errors were recorded on some of the test data. The potential for using neural networks for residual life prediction and the advantage of incorporating condition-based data into the model was nevertheless proven for both examples.     

Remaining useful life prediction for hard failures using joint model with extended hazard

In this paper, we investigate a joint modeling method for hard failures where both degradation signals and time‐to‐event data are available. The mixed‐effects model is used to model degradation signals, and extended hazard model is used for the time‐to‐event data. The extended hazard is a general model which includes two well‐known hazard rate models, the Cox proportional hazards model and accelerated failure time model, as special cases. A two‐stage estimation approach is used to obtain model parameters, based on which remaining useful life for the in‐service unit can be predicted. The performance of the method is demonstrated through both simulation studies and a real case study.     

Stress‐life prediction of 25°C polypropylene materials based on calibration of Zhurkov fatigue life model

In this study, to evaluate the chemical and mechanical properties of polypropylene (PP), activation‐energy and tensile tests were performed at room temperature (25°C) on pure PP and PP reinforced with glass fibre (GF). To improve the prediction accuracy of the fatigue life, three models based on the calibration of the Zhurkov model were proposed: a regression model, modified strain‐rate model and lethargy coefficient‐based model. Based on the experimental data analysis and statistical assessment results, we proposed a modified strain‐rate model that satisfies the dependency of the physical parameters and is congruent with the predicted fatigue life data. The experimental data and modified strain‐rate model were compared with the direct cyclic analysis results. The tendency of the frequency factor as a correction parameter in the modified strain‐rate model corresponded to the experimental activation energy and the increasing GF content.     

融合多源数据的非线性退化建模与剩余寿命预测

针对目前基于单个传感器剩余寿命预测方法存在预测精度不高的问题,该文提出一种融合多源传感器数据的非线性退化建模与剩余寿命预测方法。该方法包括复合健康指标的构建、模型参数的估计和传感器融合系数的确定,在确定融合系数后,结合设备历史寿命数据与实时监测数据,利用Bayesian参数更新公式推导出设备的剩余寿命概率分布,实现设备的剩余寿命在线预测。最后通过由商用模块化航空推进系统仿真生成的发动机退化数据集进行仿真实验,结果表明该文所提方法能够有效提高设备剩余寿命预测的准确性。     

Fatigue life prediction of woven composite laminates with initial delamination

An engineering approach for fatigue life prediction of fibre‐reinforced polymer composite materials is highly desirable for industries due to the complexity in damage mechanisms and their interactions. This paper presents a fatigue‐driven residual strength model considering the effect of initial delamination size and stress ratio. Static and constant amplitude fatigue tests of woven composite specimens with delamination diameters of 0, 4 and 6 mm were carried out to determine the model parameters. Good agreement with experimental results has been achieved when the modified residual strength model has been applied for fatigue life prediction of the woven composite laminate with an initial delamination diameter of 8 mm under constant amplitude load and block fatigue load. It has been demonstrated that the residual strength degradation‐based model can effectively reflect the load sequence effect on fatigue damage and hence provide more accurate fatigue life prediction than the traditional linear damage accumulation models.     

Probabilistic life analysis of vital components considering multistage degradation under variable working conditions

The lifespan of a mechanical product is related to its working conditions; the product's performance typically shows a multistage degradation pattern throughout its life profile. The performance degradation is generally researched under constant test conditions, while the effects of different working conditions on life are seldom considered. This paper proposes a staged recursive derivation method for the multistage degradation under variable working conditions. The proposed method works by merging measured degradation data with an empirical degradation model. The measured degradation data of a new prototype are utilized to update the staged degradation model based on a Bayesian posterior probability analysis. The staged degradation model is derived stage by stage, and then the probabilistic life of the new prototype is predicted. The degradation data of a machine‐gun barrel are used as a case study to demonstrate and validate the proposed method. The results show that the probabilistic life of the test prototype can be predicted effectively in the case of relatively little measured degradation data at the product development stage. Furthermore, the proposed method appears to be especially suited to mechanical components requiring short test periods or low test costs.     

Fatigue life prediction and failure analysis of a gas turbine disc using the finite-element method

A numerical prediction of the life of a gas turbine model disc by means of the finite‐element technique is presented and the solution is compared with an experimental rim‐spinning test. The finite‐element method was used to obtain the K solution for a disc with two types of cracks, both at the notch root of the blade insert and located in the corner and in the centre. A crack aspect ratio of (a/c) = 1 was assumed. The fracture mechanics parameters J‐integral and K were used in the assessment, which were computed with linear elastic and elastic–plastic material behaviour. Using a crack propagation program with appropriate fatigue‐creep crack growth‐rate data, previously obtained in specimens for the nickel‐based superalloy IN718 at 600 °C, fatigue life predictions were made. The predicted life results were checked against experimental data obtained in real model discs. The numerical method, based on experimental fatigue data obtained in small laboratory specimens, shows great potential for development, and may be able to reduce the enormous costs involved in the testing of model and full‐size components.     

Hybrid approach for remaining useful life prediction of ball bearings

Remaining useful life (RUL) prediction plays an important role in predictive maintenance systems to support decision‐makers for arranging maintenance tasks and related resources. We propose a hybrid approach that is combined an exponential weighted moving average (EWMA) control chart for anomaly detection and machine learning models such as support vector regression (SVR) and random forest regression (RFR) with differential evolution (DE) algorithm to predict the RULs of ball bearings. Here, DE algorithm is used to find the optimal hyperparameters of SVR model. The datasets of ball bearings from the Prognostics Data Repository of NASA are used to compare the prediction performance of different methods. The degradation behavior of training data from the anomaly time to the end of life is used to transfer learning for the testing data in the SVR and RFR models. The results indicate that the proposed methods outperform the other four existing methods in terms of score. Therefore, the proposed hybrid approach is a reliable tool for the RUL prediction of ball bearings.     

An ensemble model considering health index based classification for remaining useful life prediction

Accurate prediction of remaining useful life (RUL) plays an important role in the formulation of maintenance strategies. However, due to the diversity of the failure mode of equipment, there are significant differences between the degradation data, which greatly affects the accuracy of RUL prediction. In this case, an ensemble prediction model considering health index-based (HI-based) classification is proposed in this paper. Firstly, the stacked autoencoder (SAE) is employed to construct the HI. Then, the time window is used to sequentially process the HI sequence, so that many data segments with the same length can be achieved. To differentiate the data with the similar degradation process, K-means and Xgboost are selected to construct offline and online data classification models respectively. Finally, according to the results of the data classification, the ensemble model that integrates multiple machine learning methods are separately trained and then adaptively used for RUL prediction. In addition, integrating multiple methods helps to improve the generalization ability of the model. The NASA C-MAPSS dataset is applied to verify the effectiveness of the proposed method, and the results show that the proposed method achieves a higher prediction accuracy and shorter computational time than other existing prediction models.     

Prediction of residual fatigue life under interspersed mixed-mode (I and II) overloads by Artificial Neural Network

Mixed-mode (I and II) overloads are often encountered in an engineering structure due to either alteration of the loading direction or the presence of randomly oriented defects. Prediction of fatigue life in these cases is more complex than that of mode-I overloads. The objective of this study is to explore the use of an artificial neural network (ANN) model for the prediction of fatigue crack growth rate under interspersed mixed-mode (I and II) overload. The crack growth rates as predicted by the ANN method on two aluminium alloys, 7020 T7 and 2024 T3 have been compared with the experimental data and an Exponential Model. It is observed that the predicted results are in good agreement and facilitate determination of residual fatigue life.     

Low cycle fatigue life prediction in shot‐peened components of different geometries – part II: life prediction

This study investigates the effects of shot peening on the low‐cycle fatigue performance of a low‐pressure steam turbine blade material. The finite element model incorporating shot‐peening effects, which has been introduced in part I, has been used to predict the stabilised stress/strain state in shot‐peened samples during fatigue loading. The application of this model has been extended to different notched geometries in this study. Based on the modelling results, both the Smith–Watson–Topper and Fatemi–Socie critical plane fatigue criteria have been used to predict the fatigue life of shot‐peened samples (treated with two different peening intensities) with varying notched geometries. A good agreement between experiments and predictions was obtained. The application of a critical distance method considering the stress and strain hardening gradients near the shot‐peened surface has been found to improve the life prediction results. The effects of surface defects on the accuracy of life predictions using the proposed method were also discussed.     

Prognosis of gas turbine remaining useful life using particle filter approach

Gas turbines are commonly used in distributed power generation. Because of high speed nature, they require good maintenance for increased reliability and availability. Remaining useful life prediction is therefore an essential part of condition‐based maintenance to better foresee future state hence guaranteeing design efficiency, reduced maintenance cost, and improved safety. Gas turbines also contain a lot of sensors data that need to be processed for better prediction. In this paper, a probabilistic approach called particle filter is used for prediction. The proposed approach is tested using Turbofan degradation data provided by NASA as a benchmark problem. Meanwhile, through time the gas turbines experiences a change from normal state to degraded state attributed to aging, corrosion and erosion etc. Hence, in the context of abundant data, it is helpful to know the transition between states. For the same reason, the present paper suggests a statistical approach called Z‐test. The test results show that the proposed technique provides score and MAPE values of 559.9 and 21.6 respectively, comparable to past reported performance.     

Property Modeling across Transition Temperatures in PMC's: Part III. Bending Fatigue

Parts I [1] and II [2] of the present paper introduced systematic models for the computation of thermal effects on strength and stiffness of unidirectional polymer matrix composites (PMC's) as well as the life prediction of these materials in end-loaded bending at elevated temperatures. The last step of this study was the possibility of introducing such models in durability codes such as MRLife [3]. A recent method was developed for the experimental characterization of end-loaded bending fatigue behavior of composites at elevated temperatures. The literature dealing with the durability of composite materials in bending focuses mainly on 3 and 4 point bending [4–6]. A limited set of data as well as the basis for theoretical modeling for fatigue end-loaded bending is available in the literature [7]. However, the life prediction scheme required elevated temperature experiments. New experiments in fatigue bending were performed in order to complete the available data. Microscopic observations revealed new information for the understanding of the damage process of unidirectional AS4/PPS composites in end-loaded fatigue bending. Finally, the models developed in Parts I and II were integrated into the MRLife integral enabling the life prediction of unidirectional PMC's under combined mechanical and thermal loads from room temperature experimental data.     

An anisotropic damage model based on microstructure of boom–panel for the fatigue life prediction of structural components

Because of their simplicity, many isotropic damage models have been used to approximately predict the fatigue life of metallic engineering components. However, experimental observations confirm that the anisotropic damage evolves at probable failure sites even for isotropic materials. In this study, a model of microstructure of boom–panel is constructed to simulate a representative volume element (RVE), and the anisotropic damage of the RVE is described by the independent isotropic damage of boom and panel. Firstly, the constitutive equation of the RVE in terms of stiffness of boom–panel is deduced by the principle of deformation and static consistency. Then the expressions of damage‐driving force for boom and panel based on the principle of thermodynamics are introduced, and the damage evolution equations are constructed. The parameters of boom and panel are identified from fatigue test data of uniaxial tension and pure torsion, respectively. Finally, the aforementioned method is applied to predict the fatigue life of two structures: one is Pitch‐Change‐Link, which is a kind of structure in helicopter, and the other is a specimen under tension–torsion. The prediction results all fit well with the experimental data.     

基于Gamma退化过程的直升机主减速器行星架剩余寿命预测模型

鉴于Gamma过程具有平稳、独立增量等退化建模所需的属性,将其用于描述设备退化过程,并针对缺乏故障数据时难以进行剩余寿命预测的问题,利用设备运行中采集的表征其退化状态的大量间接状态参数和少量直接状态参数,建立了基于Gamma退化过程的剩余寿命预测模型;针对经验最大化算法中似然函数难以解析求解的问题,引入粒子滤波算法实现了模型参数估计;最后将模型应用于直升机主减速器行星架的剩余寿命预测,得到了不同时刻的预测结果及95%置信区间,验证了预测模型的有效性和准确性。     

基于性能衰退评估的轴承寿命状态识别方法研究

针对滚动轴承退化性能难以评估、寿命状态难以识别的难题,提出一种基于性能衰退评估的轴承寿命状态识别新方法,该方法基于卷积自编码器(convolutional autoencoder,CAE)与多维尺度分析(multidimensional scaling,MDS)算法构建轴承性能衰退指标,再根据构建指标和改进卷积神经网络(convolutional neural network,CNN)建立轴承寿命状态识别模型,实现轴承寿命状态识别。将轴承信号样本输入CAE,实现轴承寿命状态特征的自动提取与表达,再将所提取的特征通过MDS算法进行约简获得低维特征,在低维特征空间构造欧氏距离作为轴承性能衰退指标,依据指标实现轴承数据标签化。使用标签化的轴承数据训练CNN,建立轴承寿命状态识别模型。在训练过程中,为抑制过拟合,对原始训练样本进行加噪处理,为提高模型抗干扰能力,将Leaky ReLU(LReLU)函数和dropout作为激活函数。运用轴承全寿命试验数据对识别模型进行检验,通过对比验证,结果表明所提出的轴承寿命状态识别方法能更准确的实现轴承寿命状态识别。     

Planning constant-stress accelerated life tests with multiple stresses based on D-optimal design

With the rapid technological advances, products are becoming more reliable. Then, multistress accelerated life testing (MALT) has been adopted in engineering to obtain failure information in a limited time. In order to make the testing procedure more efficient, it is necessary to better design the test plan. However, to date, relevant research on planning of MALT is limited. Multiple stresses will lead to plenty of stress-level combinations that require too much cost and time to implement. Besides, there may be interactions among multiple stresses, which need more experiments for parameter estimation. To solve these problems, we propose a novel planning method for constant-stress MALT under lognormal distribution using D-optimal design, which can reduce required test points efficiently and deal with second-order effects in models. In ALT, the log-linear model is often used to describe the life-stress relationship. Hence, D-optimal design is adopted in this paper to select test points from the whole test region. Then, optimal unit allocation plans are formulated under V/D-optimality criterion, respectively, where type I and type II censoring are both discussed. A real case of light-emitting device (LED) is presented to compare the proposed approach with other two existing methods. The results show that the proposed method performs better than other two existing methods both in prediction accuracy and estimation precision. Moreover, a sensitivity analysis reveals the robustness of the optimal plans determined by the proposed method.     

An assessment of three creep–fatigue life prediction methods for nickel‐based superalloy GH4049

High‐temperature low‐cycle fatigue tests with and without a 10‐s strain hold period in a cycle were performed on a nickel base superalloy GH4049 under a fully reversed axial total strain control mode. Three creep–fatigue life prediction methods are chosen to analyse the experimental data. These methods are the linear damage summation method (LDS), the strain range partitioning method (SRP) and the strain energy partitioning method (SEP). Their ability to predict creep‐fatigue lives of GH4049 at 700, 800 and 850 °C has been evaluated. It is found that the SEP method shows an advantage over the SRP method for all the tests under consideration. At 850 °C, the LDS and SEP methods give a more satisfactory prediction for creep–fatigue lives. At the temperatures of 700 and 800 °C, the SRP and SEP methods can correlate the life data better than the LDS method. In addition, the differences in predictive ability of these methods have also been analysed. The scanning electron microscopy (SEM) examination of fracture surfaces reveals that under creep–fatigue test conditions crack initiation mode is transgranular, while crack propagation mode is either intergranular plus transgranular or entirely intergranular, dependent on test temperature.