Big Machinery Data Preprocessing Methodology for Data-Driven Models in Prognostics and Health Management

Cofre-Martel, Sergio; Droguett, Enrique López; Modarres, Mohammad

doi:10.3390/s21206841

Cited by 27 publications

(9 citation statements)

References 63 publications

(76 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Griffiths et al (2022) also make recommendations for integrating contextual data sources, such as linking contextual data sources to time series data for labeling the data while also mitigating the large storage needs of streaming data. Cofre-Martel, Lopez Droguett and Modarres developed a step-by-step guideline for processing sensor data for PHM modeling with emphasis on handling the high volume of data and incorporation of expert knowledge from field engineers (Cofre-Martel, Lopez Droguett, & Modarres, 2021). They stress the importance of reproducible and consistent processes for data pre-processing and show the impact of pre-processing decisions on final PHM models.…”

Section: Related Literaturementioning

confidence: 99%

“…Data reduction and feature extraction and selection are important data preparation steps for prognostics model building which are well-covered in the literature. Data reduction typically includes identifying highly correlated variables in order to discard redundant variables as well as variables with low variability (Eg: close to constant value) which may not act strongly as explanatory variables (Cofre-Martel et al, 2021;Nguyen et al, 2019;Griffiths et al, 2022). Feature extraction and selection may involve preparing data ranging from calculating features such as lag times to employing analytics for identifying significant explanatory variables.…”

Section: Related Literaturementioning

confidence: 99%

See 1 more Smart Citation

Data Quality Scorecard for Assessing the Suitability of Asset Condition Data for Prognostics Modeling

Lukens¹,

Rousis

Baer

et al. 2022

PHM_CONF

View full text Add to dashboard Cite

High efficacy algorithm development for prognostics requires quality data from sensors and other contextual sources, such as maintenance, usage and inspection data. Data quality challenges, such as lack of sensor-based history (depth) across the entire fleet of components (breadth), can prohibit the ability to develop algorithms which are both cost-effective and useful. Therefore, the first step in prognostics modeling is determining the sufficiency of the data required to support the development of predictive algorithms. We present an assessment process for determining data suitability in the development of prognostic models based on available data that determines which modeling approaches are feasible, allowing for a first determination of decision-making for data adequacy. The assessment process follows a full data quality framework which also identifies where data eligibility and quality may be further enhanced using advanced technologies for data quality improvement approaches such as imputation, increasing the probability of obtaining the required data needed for the successful development of predictive algorithms. Use of this framework maximizes the quantity of quality data harvested from industrial data sources, increasing the probability of obtaining the required data needed for the successful development of predictive algorithms. Additionally, repeating this assessment as further data becomes available enables further expansion of the set of usable prognostic models as data availability grows.

show abstract

Section: Related Literaturementioning

confidence: 99%

Section: Related Literaturementioning

confidence: 99%

Data Quality Scorecard for Assessing the Suitability of Asset Condition Data for Prognostics Modeling

Lukens¹,

Rousis

Baer

et al. 2022

PHM_CONF

View full text Add to dashboard Cite

show abstract

“…In recent years, various institutions have increasingly generated huge volumes of structured, semi-structured, and unstructured data, referred to as big data. A wide range of devices, applications, and research activities generate heterogeneous data every day that needs to be stored, managed, or processed [7]. As the global population grows and health patterns change rapidly, healthcare providers and clinicians are expected to develop and implement treatment models that evolve based on these changes [8].…”

Section: Theoretical Backgroundmentioning

confidence: 99%

Impact of Big Data Analysis on Health

Rosário¹,

Dias²

2022

Preprint

View full text Add to dashboard Cite

Big data analytics tools are the use of advanced analytic techniques targeting large and diverse volumes of data that include structured, semi-structured, and unstructured data from different sources and in different sizes from terabytes to zetabytes. The health sector is faced with the need to generate and manage large data sets from various health systems, such as electronic health records and clinical decision support systems. This data can be used by providers, clinicians, and policymakers to plan and implement interventions, detect disease more quickly, predict outcomes, and personalize care delivery. However, little attention is paid to the connection between big data analytics tools and the health sector. Thus, a systematic review of the bibliometric literature (LRSB) was developed to study how the adoption of big data analytics tools and infrastructures will revolutionize the healthcare industry. The review integrated 77 scientific and/or academic documents indexed in SCOPUS presenting up‐to‐date knowledge on current insights on how big data analytics technologies influence the healthcare sector and the different big data analytical tools used. The LRSB provides findings related to the impact of Big Data analytics on the health sector by introducing opportunities and technologies that provide practical solutions to various challenges.

show abstract

“…The monitoring data of the system actually reflect many characteristics of the system, and by deeply mining the system monitoring data, a large amount of relevant information about the system’s health status can be obtained. For example, in [ 9 ], several data-driven models have been proposed and applied for diagnostics and prognostics purposes in complex systems based on monitoring data. In [ 10 ], a PHM model for the prediction of component failures and the system lifetime is proposed by combining monitoring data, time-to-failure data, and background engineering knowledge of the systems.…”

Section: Introductionmentioning

confidence: 99%

An Unsupervised Machine Learning Approach for Monitoring Data Fusion and Health Indicator Construction

Huang,

Pan,

Liu

et al. 2023

Sensors

View full text Add to dashboard Cite

The prediction of system degradation is very important as it serves as an important basis for the formulation of condition-based maintenance strategies. An effective health indicator (HI) plays a key role in the prediction of system degradation as it enables vital information for critical tasks ranging from fault diagnosis to remaining useful life prediction. To address this issue, a method for monitoring data fusion and health indicator construction based on an autoencoder (AE) and a long short-term memory (LSTM) network is proposed in this study to improve the predictability and effectiveness of health indicators. Firstly, an unsupervised method and overall framework for HI construction is built based on a deep autoencoder and an LSTM neural network. The neural network is trained fully based on the normal operating monitoring data and then the construction error of the AE model is adopted as the health indicator of the system. Secondly, we propose related machine learning techniques for monitoring data processing to overcome the issue of data fusion, such as mutual information for sensor selection and t-distributed stochastic neighbor embedding (T-SNE) for operating condition identification. Thirdly, in order to verify the performance of the proposed method, experiments are conducted based on the CMAPSS dataset and results are compared with algorithms of principal component analysis (PCA) and a vanilla autoencoder model. Result shows that the LSTM-AE model outperforms the PCA and Vanilla-AE model in the metrics of monotonicity, trendability, prognosability, and fitness. Fourthly, in order to analyze the impact of the time step of the LSMT-AE model on HI construction, we construct and analyze the system HI curve under different time steps of 5, 10, 15, 20, and 25 cycles. Finally, the results demonstrate that the proposed method for HI construction can effectively characterize the health state of a system, which is helpful for the development of further failure prognostics and converting the scheduled maintenance into condition-based maintenance.

show abstract

Big Machinery Data Preprocessing Methodology for Data-Driven Models in Prognostics and Health Management

Cited by 27 publications

References 63 publications

Data Quality Scorecard for Assessing the Suitability of Asset Condition Data for Prognostics Modeling

Data Quality Scorecard for Assessing the Suitability of Asset Condition Data for Prognostics Modeling

Impact of Big Data Analysis on Health

An Unsupervised Machine Learning Approach for Monitoring Data Fusion and Health Indicator Construction

Contact Info

Product

Resources

About