phuc do - Academia.edu (original) (raw)

Papers by phuc do

Pesquisa Operacional

Mathematical models in maintenance are an old topic: more than fifty years ago, researcher Efrain... more Mathematical models in maintenance are an old topic: more than fifty years ago, researcher Efrain Turban published a paper in the prestigious journal "Management Science" (Turban, 1967) that analysed a national survey for identifying the gaps between the theory and practice. Specifically, their main question was why very few maintenance decision makers use the mathematical models to solve the main problems facing them. Surprisingly, however, the gap mentioned by the author is still present today. Fair to say that the gap is becoming smaller, which may be a result of some factors. Maintenance has earned a high level of reputation thanks to the recognition of its relationship with the good performance of systems. Failures in production processes and service delivery systems have very negative consequences, in financial, reliability and safety dimensions. As a result, the arsenal of models and approaches to handle maintenance problems has been increased substantially. A large spectrum of new techniques have been emerging from different areas and are being associated with classical operations research approaches applied in maintenance and reliability. This special issue brings a blend of the important problems on maintenance and reliability being treated under the field of operational research.

Pesquisa Operacional

We present a novel framework for diagnostics and prognostics for multi-component systems with wea... more We present a novel framework for diagnostics and prognostics for multi-component systems with wear interaction between components. The principal elements of this framework are: health-state indicator extraction using signal-processing; clustering of wear phases using a Gaussian mixture model; a stochastic multivariate wear model; and prediction of the remaining-useful-life of components using particle-filtering. These elements of the framework are illustrated and verified using an experimental platform that generates real data. Our diagnostics study shows that different clusters not only indicate the wear-state, but also the wear-rate of the components. Furthermore, our prognostics study shows that the wear-interaction between components has an significant impact in predicting the remaining-useful-life for components. Thus, we demonstrate, for prognostics and health management, the importance of modeling wear interactions in the prognostic process of multi-component systems.

Neural Computing and Applications

2018 10th International Conference on Knowledge and Systems Engineering (KSE)

Similarity measurement on objects in a HIN is considered as a challenging problem of networked da... more Similarity measurement on objects in a HIN is considered as a challenging problem of networked data mining. There are several proposed models that are used to compute the similarity score between objects in the context of multi-typed objects and links network, such as most well-known PathSim, HeteSim, etc. which are considered as the meta-path-based approaches. However, these meta-path-based models involve the information loss due to the shortage of evaluating the similarity in the text data of content-based objects such as papers. Moreover, these models also encounter the shortage of evaluating the common neighbor relevancy between two pairwise objects. Furthermore, the capability of handling big networked data is also common challenge for prior similarity search models which are designed to work on standalone environment. Therefore, in this paper, we present the W-PathSim++ model which are the extended version of our previous work for solving the problems related to content-based similarity between objects in HIN mining. Moreover, our works in this paper also focus on improving the accuracy of similarity measurement between two objects via thoroughly evaluating their common neighborhood objects. The proposed model is also designed for taking the advantages of DataLog in network representation and meta-path querying as well as implementing within the Spark-based distributed environment in order to enable for handling large-scaled HINs. We test the W-PathSim++ model in compare with previous similarity measure approaches within DBLP dataset in order to demonstrate the effectiveness of our proposed model.

PHM Society European Conference

Maintenance planning for complex systems has still been a challenging problem. Firstly, integrati... more Maintenance planning for complex systems has still been a challenging problem. Firstly, integrating multiple dependency types into maintenance models makes them more realistic, however, more complicated to solve and analyze. Secondly, the number of maintenance decision variables needed to be optimized increases rapidly in the number of components, causing computational expensive for optimization algorithms. To face these issues, this thesis aims to incorporate multiple kinds of dependencies into maintenance models as well as to take advantage of recent advances in artificial intelligence field to effectively optimize maintenance polices for large-scale multi-component systems.

PHM Society European Conference

It is well-known that maintenance decision optimization for multi-component systems faces the cur... more It is well-known that maintenance decision optimization for multi-component systems faces the curse of dimensionality. Specifically, the number of decision variables needed to be optimized grows exponentially in the number of components causing computational expensive for optimization algorithms. To address this issue, we customize a multi-agent deep reinforcement learning algorithm, namely Weighted QMIX, in the case where system states can be fully observed to obtain cost-effective policies. A case study is conducted on a 13- component system to examine the effectiveness of the customized algorithm. The obtained results confirmed its performance.

2021 IEEE 71st Electronic Components and Technology Conference (ECTC)

Light-emitting diodes (LEDs) are a solid-state light source being used in numerous applications, ... more Light-emitting diodes (LEDs) are a solid-state light source being used in numerous applications, including display, communications, medical services, etc. However, the reliability assessment of LED components is still challenging due to the growing up of the LED complexity and/or the miniaturization of assembly technologies. To face this challenge, this paper proposes a novel accelerated degradation testing (ADT) model considering the self-heating impact in the degradation process of a LED component. So, self-heating impact is first analyzed and modeled. In fact, the junction temperature of a LED component depends not only on the heat generation (e.g., drive current, dispersing heat) but also on the current state (degradation level) of the component. Then, a modified stochastic difference equation is developed for modelling the degradation process by considering the self-heating impact. The LED reliability formulation is finally derived. In addition, an estimation method based on the maximum likelihood is developed to estimate the proposed model's parameters from experimental data. To validate our models, a case study for LED light sources is implemented. The obtained results show that, compared to the TM-21 standard and the conventional ADT methods, our proposed approach achieves better prediction performance in the LED reliability assessment.

CIRP Annals

Anticipating system failures using predictive strategies based on efficient prognostics has becom... more Anticipating system failures using predictive strategies based on efficient prognostics has become an important topic in manufacturing where maintenance plays a crucial role. As such, promising prognostics approaches use data-driven machine learning techniques, though the initial data set for learning is often small as failure occurrences are rare. Therefore, this study investigates data augmentation methods for improving prognostics by increasing data set size using samples generated by altering existing ones. First, a method is proposed for quantifying the gain from additional data. Thereafter, augmentation methods are assessed through a benchmark. Finally, contributions are illustrated in a steel industry case-study.

Le Centre pour la Communication Scientifique Directe - HAL - Diderot, Jul 8, 2014

ABSTRACT The paper presents firstly an overview of various definitions/concepts of energy efficie... more ABSTRACT The paper presents firstly an overview of various definitions/concepts of energy efficiency and their related applications in different contexts, especially in industrial sectors. Each definition/concept is analyzed and recommended for different decision-making levels. Then a multi-level approach is described in detail for evaluating energy efficiency index of an industrial process. In addition, the paper discusses potential prognostic approaches in order to forecast energy efficiency index by underlining difficulties and opportunities to implement such approaches. Finally, a specific example based on an air-fan system is introduced to illustrate energy efficiency concepts and the added value of the prognostics to predict energy efficiency evolution.

Annual Conference of the Prognostics and Health Management Society 2012, PHM Conference 2012, Sep 23, 2012

This paper deals with a condition-based maintenance (CBM) model considering both perfect and impe... more This paper deals with a condition-based maintenance (CBM) model considering both perfect and imperfect maintenance actions for a deteriorating system whose condition is aperiodically monitored according to a remaining useful life (RUL) based-inspection policy. Perfect maintenance actions restore completely the system to the 'as good as new' state. Their related cost are however often high. Imperfect preventive maintenance restores partially the system with reduced maintenance cost. Nevertheless, it may however make the system more susceptible to future deterioration. The aim of the paper is to propose a CBM model which can help to construct optimal maintenance policies when both perfect and imperfect maintenance actions are possible. To illustrate the use of the proposed CBM model, a numerical example finally is introduced.

Proc. of the 11th IMA International Conference on Modelling in Industrial Maintenance and Reliability, 2021

Maintenance decisions in multi-component systems are of great interest to maintenance managers. T... more Maintenance decisions in multi-component systems are of great interest to maintenance managers. The equipment during its operation, produces and stores a large amount of data, especially discrete event data such as alarm, failed situation, change of operation modes, stop of the systems, and so forth, and produced via processings supported by the programmable logic controller (PLCs), supervision system, SCADA. Considering this data to assist in maintenance management and decisions is an area with a growing interest in maintenance management. In this paper, we study the stochastic dependency in a multi-component system through data from PLCs database. We use appriory algorithm and affinity function to identify failure dependence in multi-component systems. The results of failure dependence can be used as input for planning group maintenance, purchase spare parts, or planning opportunistic maintenance.

Le Centre pour la Communication Scientifique Directe - HAL - memSIC, Jul 8, 2014

ABSTRACT The paper presents a joint predictive maintenance and spare parts provisioning policy fo... more ABSTRACT The paper presents a joint predictive maintenance and spare parts provisioning policy for gradually deteriorating multi-component systems with complex structure. The decision-making process related to maintenance, spare parts ordering, as well as inspections scheduling is based on both RUL prediction and structural importance measure. Moreover, economic dependency between components is studied and integrated in decision rules. In addition, the impacts of the system structure on components deterioration process are also investigated. This dependency may have a significant influence on the RUL estimation of components. In order to evaluate the performance of the proposed joint predictive policy, a cost model is used. Finally, a numerical example of a 6-component system is introduced to illustrate the use and the advantages of the proposed joint maintenance and spare parts provisioning policy.

Le Centre pour la Communication Scientifique Directe - HAL - Inria, Jul 1, 2018

This study addresses prognostics and health management (PHM) for manufacturing machines. Differen... more This study addresses prognostics and health management (PHM) for manufacturing machines. Different from previous researches where continuous monitoring is assumed for PHM, we investigate the issue with discrete event data. Various event data were recorded during system operation, which can provide useful information for fault diagnosis and failure prediction. We focus on discovery of association rules based on the industrial discrete event data. Events that occur together frequently are classified into event groups. Apriori algorithm is employed to discover the frequent event groups and identify strong association rules (occurrence of the events is highly dependent). To accommodate the algorithm, the initial event data is transformed into the form of transactional data. The obtained association rule estimates the occurrence probability of certain significant events within specified time interval. It is concluded through a case study that the number of frequent event groups and strong association rules increases with the time interval that the events are grouped as one transaction. Bin Liu et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 3.0 United States License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

The Study on "Maximizing the Spread of Influence through a Social Network" has been strongly attr... more The Study on "Maximizing the Spread of Influence through a Social Network" has been strongly attracted attentions recently. One of the most important problems is figuring those who are able to make the strongest influence on the others in spreading information based on a specific topic. We would like to build a system to support viral marketing on social networks and to promote the topic modeling, the propagation model and propagation algorithm to find out the most influential group of users of each topic. The system consists of several steps, such as word extracting, data processing and finding the most influential users in exchanged topics. We especially focus on calculating the users' influence probabilities via an action log file, using the propagation model-TLT (Topic-aware Linear Threshold) and the propagation algorithm-CELF (Cost Effective Lazy Forward). Furthermore, we also experiment our model with Enron email data, which includes 11,177 emails exchanged among 147 users and estimated in 50 topics. We have received many useful topics and the most influential group of users.

Proceedings of the Eighth International Symposium on Information and Communication Technology, 2017

In this paper, a solution of storing and inferring on citation network using graph database and t... more In this paper, a solution of storing and inferring on citation network using graph database and topic model is proposed. Citation network is a very large directed graph containing nodes and edges. Each node is a paper and each directed edge is a link between paper and its citing papers. In citation network analysis, each node usually contains basic properties of paper such as paper ID, publication year, paper title, authors. In this research, we propose one more property called "topic vector". This property contains topic distribution of a specific paper which is gained by LDA algorithm. After that, we propose a new approach to store and manage the citation network using graph database. Finally, we use the graph query language to develop some functions of citation network analysis and visualize topic propagation through the network. We also compare our approach with the traditional method in which the relational database is used to store and manage citation network. Experimental results show that the performance of our approach is higher than the traditional one and a different view of citation network analysis is also discussed.

Intelligent Computing & Optimization, 2018

Recently, heterogeneous network mining has gained tremendous attention from researcher due to its... more Recently, heterogeneous network mining has gained tremendous attention from researcher due to its wide applications. Link prediction is one of the most important task in information network mining. From the past, most of the networked data mining approaches are mainly applied for homogenous network which is considered as single-typed objects and links. Moreover, there are remained challenges related to thoroughly evaluating the content of linked objects which are considered as important in predicting the potential relationships between objects. Like a common problem of predicting co-authorship in bibliographic network such as: DBLP, DBIS, etc. There is no doubt that an author who is interesting in "data mining" field tend to cooperate with the other authors who contribute on this field only. Hence, predicting co-authorships between authors work on "data mining" with others who work on "hardware" is dull as well. Moreover, in the context of large-scaled network, traditional standalone computing mechanism also is not affordable due to low-performance in time-consuming. To overcome these challenges, n this paper, we propose an approach of topic-driven meta-path-based prediction in heterogeneous network, called T-MPP which is implemented on distributed computing environment of Spark. The T-MPP not only enables to discover potential relationships in given bibliographic network but also supports to capture the topic similarity between authors. We present experiments on a real-world DBLP network. The outputs show that our proposed T-MPP model can generate more accurate prediction results as compared to previous approaches.

2019 4th International Conference on System Reliability and Safety (ICSRS), 2019

In this paper, a study on the use of artificial neural networks for predicting the system reliabi... more In this paper, a study on the use of artificial neural networks for predicting the system reliability and maintenance cost of a system with multistate components is presented. TensorFlow and Keras APIs are used to build and train deep learning models under Python environment. Different numerical experimentations are carried out to illustrate the use of the robustness of the prediction approach. The obtained results show that artificial neural networks with TensorFlow and Keras APIs are a relevant tool for reliability and maintenance cost prediction.

Advances in Safety, Reliability and Risk Management, 2011

This paper presents a dynamic grouping maintenance strategy for multi-component systems with posi... more This paper presents a dynamic grouping maintenance strategy for multi-component systems with positive economic dependence, which implies that combining maintenance activities is cheaper than performing maintenance on components separately. Preventive maintenance durations and occurrences of maintenance activities within the scheduling horizon are considered. Moreover, in presence of opportunities with limited durations in which some maintenance activities could be executed with reduced maintenance costs, the present paper proposes a new algorithm to optimally update online the grouping maintenance planning. A numerical example of a five components system is finally introduced to illustrate the proposed dynamic grouping maintenance strategy.

Proceedings of the Ninth International Symposium on Information and Communication Technology - SoICT 2018, 2018

Along with rapid development of electronic scientific publication repositories, automatic topics ... more Along with rapid development of electronic scientific publication repositories, automatic topics identification from papers has helped a lot for the researchers in their research. Latent Dirichlet Allocation (LDA) model is the most popular method which is used to discover hidden topics in texts basing on the co-occurrence of words in a corpus. LDA algorithm has achieved good results for large documents. However, article repositories usually only store title and abstract that are too short for LDA algorithm to work effectively. In this paper, we propose CitationLDA++ model that can improve the performance of the LDA algorithm in inferring topics of the papers basing on the title or/and abstract and citation information. The proposed model is based on the assumption that the topics of the cited papers also reflects the topics of the original paper. In this study, we divide the dataset into two sets. The first one is used to build prior knowledge source using LDA algorithm. The second is training dataset used in CitationLDA++. In the inference process with Gibbs sampling, CitationLDA++ algorithm use topics distribution of prior knowledge source and citation information to guide the process of assigning the topic to words in the text. The use of topics of cited papers helps to tackle the limit of word co-occurrence in case of linked short text. Experiments with the AMiner dataset including title or/and abstract of papers and citation information, CitationLDA++ algorithm gains better perplexity measurement than no additional knowledge. Experimental results suggest that the citation information can improve the performance of LDA algorithm to discover topics of papers in the case of full content of them are not available.

Pesquisa Operacional

Mathematical models in maintenance are an old topic: more than fifty years ago, researcher Efrain... more Mathematical models in maintenance are an old topic: more than fifty years ago, researcher Efrain Turban published a paper in the prestigious journal "Management Science" (Turban, 1967) that analysed a national survey for identifying the gaps between the theory and practice. Specifically, their main question was why very few maintenance decision makers use the mathematical models to solve the main problems facing them. Surprisingly, however, the gap mentioned by the author is still present today. Fair to say that the gap is becoming smaller, which may be a result of some factors. Maintenance has earned a high level of reputation thanks to the recognition of its relationship with the good performance of systems. Failures in production processes and service delivery systems have very negative consequences, in financial, reliability and safety dimensions. As a result, the arsenal of models and approaches to handle maintenance problems has been increased substantially. A large spectrum of new techniques have been emerging from different areas and are being associated with classical operations research approaches applied in maintenance and reliability. This special issue brings a blend of the important problems on maintenance and reliability being treated under the field of operational research.

Pesquisa Operacional

We present a novel framework for diagnostics and prognostics for multi-component systems with wea... more We present a novel framework for diagnostics and prognostics for multi-component systems with wear interaction between components. The principal elements of this framework are: health-state indicator extraction using signal-processing; clustering of wear phases using a Gaussian mixture model; a stochastic multivariate wear model; and prediction of the remaining-useful-life of components using particle-filtering. These elements of the framework are illustrated and verified using an experimental platform that generates real data. Our diagnostics study shows that different clusters not only indicate the wear-state, but also the wear-rate of the components. Furthermore, our prognostics study shows that the wear-interaction between components has an significant impact in predicting the remaining-useful-life for components. Thus, we demonstrate, for prognostics and health management, the importance of modeling wear interactions in the prognostic process of multi-component systems.

Neural Computing and Applications

2018 10th International Conference on Knowledge and Systems Engineering (KSE)

Similarity measurement on objects in a HIN is considered as a challenging problem of networked da... more Similarity measurement on objects in a HIN is considered as a challenging problem of networked data mining. There are several proposed models that are used to compute the similarity score between objects in the context of multi-typed objects and links network, such as most well-known PathSim, HeteSim, etc. which are considered as the meta-path-based approaches. However, these meta-path-based models involve the information loss due to the shortage of evaluating the similarity in the text data of content-based objects such as papers. Moreover, these models also encounter the shortage of evaluating the common neighbor relevancy between two pairwise objects. Furthermore, the capability of handling big networked data is also common challenge for prior similarity search models which are designed to work on standalone environment. Therefore, in this paper, we present the W-PathSim++ model which are the extended version of our previous work for solving the problems related to content-based similarity between objects in HIN mining. Moreover, our works in this paper also focus on improving the accuracy of similarity measurement between two objects via thoroughly evaluating their common neighborhood objects. The proposed model is also designed for taking the advantages of DataLog in network representation and meta-path querying as well as implementing within the Spark-based distributed environment in order to enable for handling large-scaled HINs. We test the W-PathSim++ model in compare with previous similarity measure approaches within DBLP dataset in order to demonstrate the effectiveness of our proposed model.

PHM Society European Conference

Maintenance planning for complex systems has still been a challenging problem. Firstly, integrati... more Maintenance planning for complex systems has still been a challenging problem. Firstly, integrating multiple dependency types into maintenance models makes them more realistic, however, more complicated to solve and analyze. Secondly, the number of maintenance decision variables needed to be optimized increases rapidly in the number of components, causing computational expensive for optimization algorithms. To face these issues, this thesis aims to incorporate multiple kinds of dependencies into maintenance models as well as to take advantage of recent advances in artificial intelligence field to effectively optimize maintenance polices for large-scale multi-component systems.

PHM Society European Conference

It is well-known that maintenance decision optimization for multi-component systems faces the cur... more It is well-known that maintenance decision optimization for multi-component systems faces the curse of dimensionality. Specifically, the number of decision variables needed to be optimized grows exponentially in the number of components causing computational expensive for optimization algorithms. To address this issue, we customize a multi-agent deep reinforcement learning algorithm, namely Weighted QMIX, in the case where system states can be fully observed to obtain cost-effective policies. A case study is conducted on a 13- component system to examine the effectiveness of the customized algorithm. The obtained results confirmed its performance.

2021 IEEE 71st Electronic Components and Technology Conference (ECTC)

Light-emitting diodes (LEDs) are a solid-state light source being used in numerous applications, ... more Light-emitting diodes (LEDs) are a solid-state light source being used in numerous applications, including display, communications, medical services, etc. However, the reliability assessment of LED components is still challenging due to the growing up of the LED complexity and/or the miniaturization of assembly technologies. To face this challenge, this paper proposes a novel accelerated degradation testing (ADT) model considering the self-heating impact in the degradation process of a LED component. So, self-heating impact is first analyzed and modeled. In fact, the junction temperature of a LED component depends not only on the heat generation (e.g., drive current, dispersing heat) but also on the current state (degradation level) of the component. Then, a modified stochastic difference equation is developed for modelling the degradation process by considering the self-heating impact. The LED reliability formulation is finally derived. In addition, an estimation method based on the maximum likelihood is developed to estimate the proposed model's parameters from experimental data. To validate our models, a case study for LED light sources is implemented. The obtained results show that, compared to the TM-21 standard and the conventional ADT methods, our proposed approach achieves better prediction performance in the LED reliability assessment.

CIRP Annals

Anticipating system failures using predictive strategies based on efficient prognostics has becom... more Anticipating system failures using predictive strategies based on efficient prognostics has become an important topic in manufacturing where maintenance plays a crucial role. As such, promising prognostics approaches use data-driven machine learning techniques, though the initial data set for learning is often small as failure occurrences are rare. Therefore, this study investigates data augmentation methods for improving prognostics by increasing data set size using samples generated by altering existing ones. First, a method is proposed for quantifying the gain from additional data. Thereafter, augmentation methods are assessed through a benchmark. Finally, contributions are illustrated in a steel industry case-study.

Le Centre pour la Communication Scientifique Directe - HAL - Diderot, Jul 8, 2014

ABSTRACT The paper presents firstly an overview of various definitions/concepts of energy efficie... more ABSTRACT The paper presents firstly an overview of various definitions/concepts of energy efficiency and their related applications in different contexts, especially in industrial sectors. Each definition/concept is analyzed and recommended for different decision-making levels. Then a multi-level approach is described in detail for evaluating energy efficiency index of an industrial process. In addition, the paper discusses potential prognostic approaches in order to forecast energy efficiency index by underlining difficulties and opportunities to implement such approaches. Finally, a specific example based on an air-fan system is introduced to illustrate energy efficiency concepts and the added value of the prognostics to predict energy efficiency evolution.

Annual Conference of the Prognostics and Health Management Society 2012, PHM Conference 2012, Sep 23, 2012

This paper deals with a condition-based maintenance (CBM) model considering both perfect and impe... more This paper deals with a condition-based maintenance (CBM) model considering both perfect and imperfect maintenance actions for a deteriorating system whose condition is aperiodically monitored according to a remaining useful life (RUL) based-inspection policy. Perfect maintenance actions restore completely the system to the 'as good as new' state. Their related cost are however often high. Imperfect preventive maintenance restores partially the system with reduced maintenance cost. Nevertheless, it may however make the system more susceptible to future deterioration. The aim of the paper is to propose a CBM model which can help to construct optimal maintenance policies when both perfect and imperfect maintenance actions are possible. To illustrate the use of the proposed CBM model, a numerical example finally is introduced.

Proc. of the 11th IMA International Conference on Modelling in Industrial Maintenance and Reliability, 2021

Maintenance decisions in multi-component systems are of great interest to maintenance managers. T... more Maintenance decisions in multi-component systems are of great interest to maintenance managers. The equipment during its operation, produces and stores a large amount of data, especially discrete event data such as alarm, failed situation, change of operation modes, stop of the systems, and so forth, and produced via processings supported by the programmable logic controller (PLCs), supervision system, SCADA. Considering this data to assist in maintenance management and decisions is an area with a growing interest in maintenance management. In this paper, we study the stochastic dependency in a multi-component system through data from PLCs database. We use appriory algorithm and affinity function to identify failure dependence in multi-component systems. The results of failure dependence can be used as input for planning group maintenance, purchase spare parts, or planning opportunistic maintenance.

Le Centre pour la Communication Scientifique Directe - HAL - memSIC, Jul 8, 2014

ABSTRACT The paper presents a joint predictive maintenance and spare parts provisioning policy fo... more ABSTRACT The paper presents a joint predictive maintenance and spare parts provisioning policy for gradually deteriorating multi-component systems with complex structure. The decision-making process related to maintenance, spare parts ordering, as well as inspections scheduling is based on both RUL prediction and structural importance measure. Moreover, economic dependency between components is studied and integrated in decision rules. In addition, the impacts of the system structure on components deterioration process are also investigated. This dependency may have a significant influence on the RUL estimation of components. In order to evaluate the performance of the proposed joint predictive policy, a cost model is used. Finally, a numerical example of a 6-component system is introduced to illustrate the use and the advantages of the proposed joint maintenance and spare parts provisioning policy.

Le Centre pour la Communication Scientifique Directe - HAL - Inria, Jul 1, 2018

This study addresses prognostics and health management (PHM) for manufacturing machines. Differen... more This study addresses prognostics and health management (PHM) for manufacturing machines. Different from previous researches where continuous monitoring is assumed for PHM, we investigate the issue with discrete event data. Various event data were recorded during system operation, which can provide useful information for fault diagnosis and failure prediction. We focus on discovery of association rules based on the industrial discrete event data. Events that occur together frequently are classified into event groups. Apriori algorithm is employed to discover the frequent event groups and identify strong association rules (occurrence of the events is highly dependent). To accommodate the algorithm, the initial event data is transformed into the form of transactional data. The obtained association rule estimates the occurrence probability of certain significant events within specified time interval. It is concluded through a case study that the number of frequent event groups and strong association rules increases with the time interval that the events are grouped as one transaction. Bin Liu et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 3.0 United States License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

The Study on "Maximizing the Spread of Influence through a Social Network" has been strongly attr... more The Study on "Maximizing the Spread of Influence through a Social Network" has been strongly attracted attentions recently. One of the most important problems is figuring those who are able to make the strongest influence on the others in spreading information based on a specific topic. We would like to build a system to support viral marketing on social networks and to promote the topic modeling, the propagation model and propagation algorithm to find out the most influential group of users of each topic. The system consists of several steps, such as word extracting, data processing and finding the most influential users in exchanged topics. We especially focus on calculating the users' influence probabilities via an action log file, using the propagation model-TLT (Topic-aware Linear Threshold) and the propagation algorithm-CELF (Cost Effective Lazy Forward). Furthermore, we also experiment our model with Enron email data, which includes 11,177 emails exchanged among 147 users and estimated in 50 topics. We have received many useful topics and the most influential group of users.

Proceedings of the Eighth International Symposium on Information and Communication Technology, 2017

In this paper, a solution of storing and inferring on citation network using graph database and t... more In this paper, a solution of storing and inferring on citation network using graph database and topic model is proposed. Citation network is a very large directed graph containing nodes and edges. Each node is a paper and each directed edge is a link between paper and its citing papers. In citation network analysis, each node usually contains basic properties of paper such as paper ID, publication year, paper title, authors. In this research, we propose one more property called "topic vector". This property contains topic distribution of a specific paper which is gained by LDA algorithm. After that, we propose a new approach to store and manage the citation network using graph database. Finally, we use the graph query language to develop some functions of citation network analysis and visualize topic propagation through the network. We also compare our approach with the traditional method in which the relational database is used to store and manage citation network. Experimental results show that the performance of our approach is higher than the traditional one and a different view of citation network analysis is also discussed.

Intelligent Computing & Optimization, 2018

Recently, heterogeneous network mining has gained tremendous attention from researcher due to its... more Recently, heterogeneous network mining has gained tremendous attention from researcher due to its wide applications. Link prediction is one of the most important task in information network mining. From the past, most of the networked data mining approaches are mainly applied for homogenous network which is considered as single-typed objects and links. Moreover, there are remained challenges related to thoroughly evaluating the content of linked objects which are considered as important in predicting the potential relationships between objects. Like a common problem of predicting co-authorship in bibliographic network such as: DBLP, DBIS, etc. There is no doubt that an author who is interesting in "data mining" field tend to cooperate with the other authors who contribute on this field only. Hence, predicting co-authorships between authors work on "data mining" with others who work on "hardware" is dull as well. Moreover, in the context of large-scaled network, traditional standalone computing mechanism also is not affordable due to low-performance in time-consuming. To overcome these challenges, n this paper, we propose an approach of topic-driven meta-path-based prediction in heterogeneous network, called T-MPP which is implemented on distributed computing environment of Spark. The T-MPP not only enables to discover potential relationships in given bibliographic network but also supports to capture the topic similarity between authors. We present experiments on a real-world DBLP network. The outputs show that our proposed T-MPP model can generate more accurate prediction results as compared to previous approaches.

2019 4th International Conference on System Reliability and Safety (ICSRS), 2019

In this paper, a study on the use of artificial neural networks for predicting the system reliabi... more In this paper, a study on the use of artificial neural networks for predicting the system reliability and maintenance cost of a system with multistate components is presented. TensorFlow and Keras APIs are used to build and train deep learning models under Python environment. Different numerical experimentations are carried out to illustrate the use of the robustness of the prediction approach. The obtained results show that artificial neural networks with TensorFlow and Keras APIs are a relevant tool for reliability and maintenance cost prediction.

Advances in Safety, Reliability and Risk Management, 2011

This paper presents a dynamic grouping maintenance strategy for multi-component systems with posi... more This paper presents a dynamic grouping maintenance strategy for multi-component systems with positive economic dependence, which implies that combining maintenance activities is cheaper than performing maintenance on components separately. Preventive maintenance durations and occurrences of maintenance activities within the scheduling horizon are considered. Moreover, in presence of opportunities with limited durations in which some maintenance activities could be executed with reduced maintenance costs, the present paper proposes a new algorithm to optimally update online the grouping maintenance planning. A numerical example of a five components system is finally introduced to illustrate the proposed dynamic grouping maintenance strategy.

Proceedings of the Ninth International Symposium on Information and Communication Technology - SoICT 2018, 2018

Along with rapid development of electronic scientific publication repositories, automatic topics ... more Along with rapid development of electronic scientific publication repositories, automatic topics identification from papers has helped a lot for the researchers in their research. Latent Dirichlet Allocation (LDA) model is the most popular method which is used to discover hidden topics in texts basing on the co-occurrence of words in a corpus. LDA algorithm has achieved good results for large documents. However, article repositories usually only store title and abstract that are too short for LDA algorithm to work effectively. In this paper, we propose CitationLDA++ model that can improve the performance of the LDA algorithm in inferring topics of the papers basing on the title or/and abstract and citation information. The proposed model is based on the assumption that the topics of the cited papers also reflects the topics of the original paper. In this study, we divide the dataset into two sets. The first one is used to build prior knowledge source using LDA algorithm. The second is training dataset used in CitationLDA++. In the inference process with Gibbs sampling, CitationLDA++ algorithm use topics distribution of prior knowledge source and citation information to guide the process of assigning the topic to words in the text. The use of topics of cited papers helps to tackle the limit of word co-occurrence in case of linked short text. Experiments with the AMiner dataset including title or/and abstract of papers and citation information, CitationLDA++ algorithm gains better perplexity measurement than no additional knowledge. Experimental results suggest that the citation information can improve the performance of LDA algorithm to discover topics of papers in the case of full content of them are not available.