scholarly journals A locally adaptive ensemble approach for data-driven prognostics of heterogeneous fleets

2017 ◽  
Vol 231 (4) ◽  
pp. 350-363 ◽  
Author(s):  
Sameer Al-Dahidi ◽  
Francesco Di Maio ◽  
Piero Baraldi ◽  
Enrico Zio
Keyword(s):  
Markov Model ◽  
Remaining Useful Life ◽  
Operating Conditions ◽  
Data Driven ◽  
Prognostic Models ◽  
Ensemble Approach ◽  
Fuzzy Similarity ◽  
Finite State ◽  
Useful Life ◽  
Life Predictions

In this work, we consider the problem of predicting the remaining useful life of a piece of equipment, based on data collected from a heterogeneous fleet working under different operating conditions. When the equipment experiences variable operating conditions, individual data-driven prognostic models are not able to accurately predict the remaining useful life during the entire equipment life. The objective of this work is to develop an ensemble approach of different prognostic models for aggregating their remaining useful life predictions in an adaptive way, for good performance throughout the degradation progression. Two data-driven prognostic models are considered, a homogeneous discrete-time finite-state semi-Markov model and a fuzzy similarity–based model. The ensemble approach is based on a locally weighted strategy that aggregates the outcomes of the two prognostic models of the ensemble by assigning to each model a weight and a bias related to its local performance, that is, the accuracy in predicting the remaining useful life of patterns of a validation set similar to the one under study. The proposed approach is applied to a case study regarding a heterogeneous fleet of aluminum electrolytic capacitors used in electric vehicle powertrains. The results have shown that the proposed ensemble approach is able to provide more accurate remaining useful life predictions throughout the entire life of the equipment compared to an alternative ensemble approach and to each individual homogeneous discrete-time finite-state semi-Markov model and fuzzy similarity–based models.

scholarly journals Validation of Data-Driven Labeling Approaches Using a Novel Deep Network Structure for Remaining Useful Life Predictions

IEEE Access ◽  
2019 ◽  
Vol 7 ◽  
pp. 71563-71575 ◽  
Author(s):  
Andre Listou Ellefsen ◽  
Sergey Ushakov ◽  
Vilmar Aesoy ◽  
Houxiang Zhang
Keyword(s):  
Network Structure ◽  
Remaining Useful Life ◽  
Data Driven ◽  
Deep Network ◽  
Useful Life ◽  
Life Predictions

Data Driven Approach to Analyzing The Impact of Power Plant Cycling on Air Preheater Degradation and Remaining Useful Life

2021 ◽  
Author(s):  
Himanshu Sharma ◽  
Veronica Adetola ◽  
Laurentiu Marinovici ◽  
Herbert T. Schaef
Keyword(s):  
Power Plant ◽  
Operation Time ◽  
Remaining Useful Life ◽  
Operating Conditions ◽  
Dew Point ◽  
Data Driven ◽  
Air Preheater ◽  
Operational Life ◽  
Useful Life ◽  
The Impact

Abstract Due to the increased penetration of renewable energy generation sources, and fluctuations of the oil and gas prices, modern coal burning power plants deal with increased variability in the demand for power generation. These varying demands result in their intermittent under-capacity operation (cycling). Periodical ramping down and back up to follow the daily power demands causes damages to the plant components reducing its operational life. In this paper we analyze the impact of cycling on a rotary Ljungstrom air preheater (APH) unit installed at a coal fire power plant in the US. An inefficient air preheater can significantly impact boiler performance. Due to the repeated boiler’s hot-cold start, the APH experiences fluctuating operating conditions that result in accelerated degradation mechanisms, such as dew-point corrosion, fouling/deposition plugging, and air heater leakage. The analysis in this paper utilizes field data related to APH basket replacement, and the number of cycles experienced by the boiler to model the life expectancy of the baskets. The data-driven model enables preventive maintenance strategies for the APH by predicting how long the APH baskets will last in a probabilistic sense. The analysis showed that an increase in cycling for a fixed operation time can reduce the APH basket remaining useful life by about 30%.

scholarly journals Data-Driven Deep Learning-Based Attention Mechanism for Remaining Useful Life Prediction: Case Study Application to Turbofan Engine Analysis

Electronics ◽  
2021 ◽  
Vol 10 (20) ◽  
pp. 2453
Author(s):  
Amgad Muneer ◽  
Shakirah Mohd Taib ◽  
Sheraz Naseer ◽  
Rao Faizan Ali ◽  
Izzatdin Abdul Aziz
Keyword(s):  
Time Window ◽  
Attention Mechanism ◽  
Remaining Useful Life ◽  
Data Driven ◽  
Sensor Data ◽  
Prognostic Models ◽  
Prediction Ability ◽  
Turbofan Engine ◽  
Useful Life ◽  
Engine System

Accurately predicting the remaining useful life (RUL) of the turbofan engine is of great significance for improving the reliability and safety of the engine system. Due to the high dimension and complex features of sensor data in RUL prediction, this paper proposes four data-driven prognostic models based on deep neural networks (DNNs) with an attention mechanism. To improve DNN feature extraction, data are prepared using a sliding time window technique. The raw data collected after normalizing is simply fed into the suggested network, requiring no prior knowledge of prognostics or signal processing and simplifying the proposed method’s applicability. In order to verify the RUL prediction ability of the proposed DNN techniques, the C-MAPSS benchmark dataset of the turbofan engine system is validated. The experimental results showed that the developed long short-term memory (LSTM) model with attention mechanism achieved accurate RUL prediction in both scenarios with a high degree of robustness and generalization ability. Furthermore, the proposed model performance outperforms several state-of-the-art prognosis methods, where the LSTM-based model with attention mechanism achieved an RMSE of 12.87 and 11.23 for FD002 and FD003 subset of data, respectively.

Failure Prognosis Based on Adaptive State Space Models

2016 ◽  
Author(s):  
Guangxing Bai ◽  
Amirmahyar Abdolsamadi ◽  
Pingfeng Wang
Keyword(s):  
State Space ◽  
Environmental Changes ◽  
Remaining Useful Life ◽  
State Space Models ◽  
Data Driven ◽  
Operational Conditions ◽  
Useful Life ◽  
Failure Prognosis ◽  
Model Recognition ◽  
Life Predictions

This paper presents a generic data-driven failure prognosis method based on adaptive state space models for engineering systems, which integrates adaptive model recognition with a dynamic system model for remaining useful life prediction. The developed approach employs a statistical learning framework for adaptively learning of time-series degradation performance, and then a Bayesian technique for self-updating of data-driven models to adapt the operational or environmental changes. With the developed approach, the prognosis technique can eliminate the dependence to system specific models and be adaptive to system performance changes due to degradation or variation of system operational conditions, thereby yielding accurate remaining useful life predictions. The developed methodology is demonstrated by an engineering case study.

scholarly journals Data-driven remaining useful life prediction based on domain adaptation

2021 ◽  
Vol 7 ◽  
pp. e690
Author(s):  
Bin cheng Wen ◽  
Ming qing Xiao ◽  
Xue qi Wang ◽  
Xin Zhao ◽  
Jian feng Li ◽  
...  
Keyword(s):  
Test Data ◽  
Health Management ◽  
Domain Adaptation ◽  
Current Data ◽  
Remaining Useful Life ◽  
Operating Conditions ◽  
Training Data ◽  
Data Driven ◽  
Target Domain ◽  
Useful Life

As an important part of prognostics and health management, remaining useful life (RUL) prediction can provide users and managers with system life information and improve the reliability of maintenance systems. Data-driven methods are powerful tools for RUL prediction because of their great modeling abilities. However, most current data-driven studies require large amounts of labeled training data and assume that the training data and test data follow similar distributions. In fact, the collected data are often variable due to different equipment operating conditions, fault modes, and noise distributions. As a result, the assumption that the training data and the test data obey the same distribution may not be valid. In response to the above problems, this paper proposes a data-driven framework with domain adaptability using a bidirectional gated recurrent unit (BGRU). The framework uses a domain-adversarial neural network (DANN) to implement transfer learning (TL) from the source domain to the target domain, which contains only sensor information. To verify the effectiveness of the proposed method, we analyze the IEEE PHM 2012 Challenge datasets and use them for verification. The experimental results show that the generalization ability of the model is effectively improved through the domain adaptation approach.

scholarly journals A New Hybrid Prognostic Methodology

2019 ◽  
Vol 10 (2) ◽  
Author(s):  
Omer F. Eker ◽  
Fatih Camci ◽  
Ian K. Jennions
Keyword(s):  
Hybrid Methods ◽  
Industrial Applications ◽  
Remaining Useful Life ◽  
Data Driven ◽  
Prognostic Models ◽  
Advantages And Disadvantages ◽  
Minimal Data ◽  
Useful Life ◽  
Filter Clogging ◽  
Short Term Prediction

Methodologies for prognostics usually centre on physics-based or data-driven approaches. Both have advantages and disadvantages, but accurate prediction relies on extensive data being available. For industrial applications, this is very rarely the case, and hence the chosen method’s performance can deteriorate quite markedly from optimal. For this reason, a hybrid methodology, merging physics-based and data-driven approaches, has been developed and is reported here. Most, if not all, hybrid methods apply physics-based and data-driven approaches in different steps of the prognostics process (i.e. state estimation and state forecasting). The presented technique combines both methods in forecasting, and integrates the short-term prediction of a physics-based model with the longer-term projection of a similarity-based data-driven model, to obtain remaining useful life estimation. The proposed hybrid prognostic methodology has been tested on two engineering datasets, one for crack growth and the other for filter clogging. The performance of the presented methodology has been evaluated by comparing remaining useful life estimations obtained from both hybrid and individual prognostic models. The results show that the presented methodology improves accuracy, robustness and applicability, especially in the case of minimal data being available.

scholarly journals Remaining Useful Life Prediction of MOSFETs via the Takagi–Sugeno Framework

Energies ◽  
2021 ◽  
Vol 14 (8) ◽  
pp. 2135
Author(s):  
Marcin Witczak ◽  
Marcin Mrugalski ◽  
Bogdan Lipiec
Keyword(s):  
Historical Data ◽  
Field Effect Transistors ◽  
Failure Process ◽  
Practical Method ◽  
Remaining Useful Life ◽  
Data Driven ◽  
Data Repository ◽  
Uncertainty Interval ◽  
Useful Life ◽  
Takagi Sugeno

The paper presents a new method of predicting the remaining useful life of technical devices. The proposed soft computing approach bridges the gap between analytical and data-driven health prognostic approaches. Whilst the former ones are based on the classical exponential shape of degradation, the latter ones learn the degradation behavior from the observed historical data. As a result of the proposed fusion, a practical method for calculating components’ remaining useful life is proposed. Contrarily to the approaches presented in the literature, the proposed ensemble of analytical and data-driven approaches forms the uncertainty interval containing an expected remaining useful life. In particular, a Takagi–Sugeno multiple models-based framework is used as a data-driven approach while an exponential curve fitting on-line approach serves as an analytical one. Unlike conventional data-driven methods, the proposed approach is designed on the basis of the historical data that apart from learning is also applied to support the diagnostic decisions. Finally, the entire scheme is used to predict power Metal Oxide Field Effect Transistors’ (MOSFETs) health status. The status of the currently operating MOSFET is determined taking into consideration the knowledge obtained from the preceding MOSFETs, which went through the run-to-failure process. Finally, the proposed approach is validated with the application of real data obtained from the NASA Ames Prognostics Data Repository.

A Clustering-Based Framework for Performance Degradation Prediction of Slewing Bearing Using Multiple Physical Signals

2019 ◽  
Vol 5 (2) ◽  
Author(s):  
Peng Ding ◽  
Hua Wang ◽  
Yongfen Dai
Keyword(s):  
Degradation Process ◽  
Remaining Useful Life ◽  
Data Driven ◽  
Self Organizing Map ◽  
Rolling Bearings ◽  
Heavy Load ◽  
Joint Application ◽  
Learning Modes ◽  
Useful Life ◽  
Slewing Bearings

Diagnosing the failure or predicting the performance state of low-speed and heavy-load slewing bearings is a practical and effective method to reduce unexpected stoppage or optimize the maintenances. Many literatures focus on the performance prediction of small rolling bearings, while studies on slewing bearings' health evaluation are very rare. Among these rare studies, supervised or unsupervised data-driven models are often used alone, few researchers devote to remaining useful life (RUL) prediction using the joint application of two learning modes which could fully take diversity and complexity of slewing bearings' degradation and damage into consideration. Therefore, this paper proposes a clustering-based framework with aids of supervised models and multiple physical signals. Correlation analysis and principle component analysis (PCA)-based multiple sensitive features in time-domain are used to establish the performance recession indicators (PRIs) of torque, temperature, and vibration. Subsequently, these three indicators are divided into several parts representing different degradation periods via optimized self-organizing map (OSOM). Finally, corresponding data-driven life models of these degradation periods are generated. Experimental results indicate that multiple physical signals can effectively describe the degradation process. The proposed clustering-based framework is provided with a more accurate prediction of slewing bearings' RUL and well reflects the performance recession periods.

Sign in / Sign up

Export Citation Format

Share Document