Two stage deep learning for prognostics using multi-loss encoder and convolutional composite features

Pillai, S. Muthu Perumal; Vadakkepat, Prahlad

doi:10.1016/j.eswa.2021.114569

Cited by 26 publications

(6 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The HI curve can be generated using autoencoders of various [19,20] principles for similarity-based curve matching techniques. In [21] as a feature extractor, a convolution-based autoencoder is used to construct a latent representation of the engine state that is maximally correlated with the function-based curve constructed from training sample sensor values. In the article, [22] the authors proposed a comprehensive approach using a feature extractor built by applying multilayer CNN layers to a time-sequence sliding window.…”

Section: Auxiliary Health Characteristicmentioning

confidence: 99%

“…In addition to this assumption, the accuracy of the validation of the constructed model is also used to determine the value. Due to the various methods of estimation, most of the empirical estimates of RUL lie in the range of '115' to '160' switching cycles [27,28,21,29,9,30,31,32]. The most popular approach to solving the RUL curve prediction problem is to use structures consisting of feature extractors and a regression head.…”

Section: Direct Rul Predictionmentioning

confidence: 99%

See 1 more Smart Citation

Interaction models for remaining useful life estimation

Zhevnenko¹,

Kazantsev²,

Makarov³

2023

Preprint

View full text Add to dashboard Cite

The paper deals with the problem of controlling the state of industrial devices according to the readings of their sensors. The current methods rely on one approach to feature extraction in which the prediction occurs. We proposed a technique to build a scalable model that combines multiple different feature extractor blocks. A new model based on sequential sensor space analysis achieves state-of-the-art results on the C-MAPSS benchmark for equipment remaining useful life estimation. The resulting model performance was validated including the prediction changes with scaling.

show abstract

Section: Auxiliary Health Characteristicmentioning

confidence: 99%

Section: Direct Rul Predictionmentioning

confidence: 99%

Interaction models for remaining useful life estimation

Zhevnenko¹,

Kazantsev²,

Makarov³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Examples of AE in the context of feature extraction are found in Mishra and Huhtala (2019) and Chen, Zhu, et al (2021), extracting in all cases operating profiles from time‐series. Pillai and Vadakkepat (2021) utilize a multilayer convolutional AE based on a multiloss objective function. The relevant information maximizing encoder generates high‐dimensional representation that is processed in a second stage through a depth‐wise separable convolution that learns temporal features.…”

Section: Data Mining In Predictive Maintenancementioning

confidence: 99%

“…It integrates the model in a complete platform for PdM based on a modular software solution for edge computing gateways. In the case of Pillai and Vadakkepat (2021), the CNN model starts from high-dimensional data extracted with an AE. Then, the temporal features and encoded representation are concatenated, forming convolutional composite features that are used to train the model for RUL prediction.…”

Section: Artificial Neural Networkmentioning

confidence: 99%

“…However, the most common DL method for feature extraction is the Autoencoder (AE), in which the data are compressed or reduced to lower dimensions and then decoded to the desired dimension. Examples of AE in the context of feature extraction are found inMishra and Huhtala (2019) andChen, Zhu, et al (2021), extracting in all cases operating profiles from time-series Pillai and Vadakkepat (2021). utilize a multilayer convolutional AE based on a multiloss objective function.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Data mining in predictive maintenance systems: A taxonomy and systematic review

Esteban

Zafra

Ventura

2022

WIREs Data Min & Knowl

View full text Add to dashboard Cite

Predictive maintenance is a field of study whose main objective is to optimize the timing and type of maintenance to perform on various industrial systems. This aim involves maximizing the availability time of the monitored system and minimizing the number of resources used in maintenance. Predictive maintenance is currently undergoing a revolution thanks to advances in industrial systems monitoring within the Industry 4.0 paradigm. Likewise, advances in artificial intelligence and data mining allow the processing of a great amount of data to provide more accurate and advanced predictive models. In this context, many actors have become interested in predictive maintenance research, becoming one of the most active areas of research in computing, where academia and industry converge. The objective of this paper is to conduct a systematic literature review that provides an overview of the current state of research concerning predictive maintenance from a data mining perspective. The review presents a first taxonomy that implies different phases considered in any data mining process to solve a predictive maintenance problem, relating the predictive maintenance tasks with the main data mining tasks to solve them. Finally, the paper presents significant challenges and future research directions in terms of the potential of data mining applied to predictive maintenance. This article is categorized under: Application Areas > Industry Specific Applications Technologies > Internet of Things

show abstract