A predictive maintenance system for integral type faults based on support vector machines:An application to ion implantation

In semiconductor fabrication processes, effective management of maintenance operations is fundamental to decrease costs associated with failures and downtime. Predictive Maintenance (PdM) approaches, based on statistical methods and historical data, are becoming popular for their predictive capabilities and low (potentially zero) added costs. We present here a PdM module based on Support Vector Machines for prediction of integral type faults, that is, the kind of failures that happen due to machine usage and stress of equipment parts. The proposed module may also be employed as a health factor indicator. The module has been applied to a frequent maintenance problem in semiconductor manufacturing industry, namely the breaking of the filament in the ion-source of ion-implantation tools. The PdM has been tested on a real production dataset. © 2013 IEEE.

Machine Learning for Predictive Maintenance: A Multiple Classifiers Approach

In this paper a multiple classifier machine learning methodology for Predictive Maintenance (PdM) is presented. PdM is a prominent strategy for dealing with maintenance issues given the increasing need to minimize downtime and associated costs. One of the challenges with PdM is generating so called ’health factors’ or quantitative indicators of the status of a system associated with a given maintenance issue, and determining their relationship to operating costs and failure risk. The proposed PdM methodology allows dynamical decision rules to be adopted for maintenance management and can be used with high-dimensional and censored data problems. This is achieved by training multiple classification modules with different prediction horizons to provide different performance trade-offs in terms of frequency of unexpected breaks and unexploited lifetime and then employing this information in an operating cost based maintenance decision system to minimise expected costs. The effectiveness of the methodology is demonstrated using a simulated example and a benchmark semiconductor manufacturing maintenance problem.

Predictive Maintenance for Improved Sustainability - An Ion Beam Etch Endpoint Detection System Use Case

In modern semiconductor manufacturing facilities maintenance strategies are increasingly shifting from traditional preventive maintenance (PM) based approaches to more efficient and sustainable predictive maintenance (PdM) approaches. This paper describes the development of such an online PdM module for the endpoint detection system of an ion beam etch tool in semiconductor manufacturing. The developed system uses optical emission spectroscopy (OES) data from the endpoint detection system to estimate the RUL of lenses, a key detector component that degrades over time. Simulation studies for historical data for the use case demonstrate the effectiveness of the proposed PdM solution and the potential for improved sustainability that it affords.

An Adaptive Machine Learning Decision System for Flexible Predictive Maintenance

Process monitoring and Predictive Maintenance (PdM) are gaining increasing attention in most manufacturing environments as a means of reducing maintenance related costs and downtime. This is especially true in industries that are data intensive such as semiconductor manufacturing. In this paper an adaptive PdM based flexible maintenance scheduling decision support system, which pays particular attention to associated opportunity and risk costs, is presented. The proposed system, which employs Machine Learning and regularized regression methods, exploits new information as it becomes available from newly processed components to refine remaining useful life estimates and associated costs and risks. The system has been validated on a real industrial dataset related to an Ion Beam Etching process for semiconductor manufacturing.

Design of a labVIEW system applied to predictive maintenance

The implementation of vibration analysis techniques based on virtual instrumentation has spread increasingly in the academic and industrial branch, since the use of any software for this type of analysis brings good results at low cost. Among the existing software for programming and creation of virtual instruments, the LabVIEW was chosen for this project. This software has good interface with the method of graphical programming. In this project, it was developed a system of rotating machine condition monitoring. This monitoring system is applied in a test stand, simulating large scale applications, such as in hydroelectric, nuclear and oil exploration companies. It was initially used a test stand, where an instrumentation for data acquisition was inserted, composed of accelerometers and inductive proximity sensors. The data collection system was structured on the basis of an NI 6008 A/D converter of National Instruments. An electronic circuit command was developed through the A/D converter for a remote firing of the test stand. The equipment monitoring is performed through the data collected from the sensors. The vibration signals collected by accelerometers are processed in the time domain and frequency. Also, proximity probes were used for the axis orbit evaluation and an inductive sensor for the rotation and trigger measurement. © (2013) Trans Tech Publications, Switzerland.

Predictive maintenance of a worm reducer with an unbalanced load

Among all predictive maintenance techniques the oil analysis and vibration analysis are the most important for monitoring some mechanical systems. The integration of these techniques has potential to improve industrial maintenance practices and provide a better economic gain for industries. To study the integration of these two techniques, a test rig was set up to obtain an extreme working condition for the worm reducer used in this paper. The test rig was composed by a motor connected to a reducer through a flexible coupling and with an unbalanced load. The analysis of the results carried out by using a sample of the oil recommended by the manufacturer in extreme conditions, and using liquid contaminant is presented. From the results it was observed that if there is an abnormal instantaneous load in a system, the subsequent vibration analysis may not perceive what occurred if there was no permanent damage, which is not the case with the lubricant analysis.

An advanced real-time predictive maintenance framework for large scale machine systems

This thesis introduces and develops a novel real-time predictive maintenance system to estimate the machine system parameters using the motion current signature. Recently, motion current signature analysis has been addressed as an alternative to the use of sensors for monitoring internal faults of a motor. A maintenance system based upon the analysis of motion current signature avoids the need for the implementation and maintenance of expensive motion sensing technology. By developing nonlinear dynamical analysis for motion current signature, the research described in this thesis implements a novel real-time predictive maintenance system for current and future manufacturing machine systems. A crucial concept underpinning this project is that the motion current signature contains infor­mation relating to the machine system parameters and that this information can be extracted using nonlinear mapping techniques, such as neural networks. Towards this end, a proof of con­cept procedure is performed, which substantiates this concept. A simulation model, TuneLearn, is developed to simulate the large amount of training data required by the neural network ap­proach. Statistical validation and verification of the model is performed to ascertain confidence in the simulated motion current signature. Validation experiment concludes that, although, the simulation model generates a good macro-dynamical mapping of the motion current signature, it fails to accurately map the micro-dynamical structure due to the lack of knowledge regarding performance of higher order and nonlinear factors, such as backlash and compliance. Failure of the simulation model to determine the micro-dynamical structure suggests the pres­ence of nonlinearity in the motion current signature. This motivated us to perform surrogate data testing for nonlinearity in the motion current signature. Results confirm the presence of nonlinearity in the motion current signature, thereby, motivating the use of nonlinear tech­niques for further analysis. Outcomes of the experiment show that nonlinear noise reduction combined with the linear reverse algorithm offers precise machine system parameter estimation using the motion current signature for the implementation of the real-time predictive maintenance system. Finally, a linear reverse algorithm, BJEST, is developed and applied to the motion current signature to estimate the machine system parameters.

A New Model For Reliability Centered Maintenance In Petroleum Refineries

Refiners today operate their equipment for prolonged periods without shutdown. This is primarily due to the increased pressures of the market resulting in extended shutdown-to-shutdown intervals. This places extreme demands on the reliability of the plant equipment. The traditional methods of reliability assurance, like Preventive Maintenance, Predictive Maintenance and Condition Based Maintenance become inadequate in the face of such demands. The alternate approaches to reliability improvement, being adopted the world over are implementation of RCFA programs and Reliability Centered Maintenance. However refiners and process plants find it difficult to adopt this standardized methodology of RCM mainly due to the complexity and the large amount of analysis that needs to be done, resulting in a long drawn out implementation, requiring the services of a number of skilled people. These results in either an implementation restricted to only few equipment or alternately, one that is non-standard. The paper presents the current models in use, the core requirements of a standard RCM model, the alternatives to classical RCM, limitations in the existing model, classical RCM and available alternatives to RCM and will then go on to present an ‗Accelerated‘ approach to RCM implementation, that, while ensuring close conformance to the standard, does not place a large burden on the implementers

Probabilistic models to assist maintenance of multiple instruments

The paper discusses maintenance challenges of organisations with a huge number of devices and proposes the use of probabilistic models to assist monitoring and maintenance planning. The proposal assumes connectivity of instruments to report relevant features for monitoring. Also, the existence of enough historical registers with diagnosed breakdowns is required to make probabilistic models reliable and useful for predictive maintenance strategies based on them. Regular Markov models based on estimated failure and repair rates are proposed to calculate the availability of the instruments and Dynamic Bayesian Networks are proposed to model cause-effect relationships to trigger predictive maintenance services based on the influence between observed features and previously documented diagnostics

Maintenance of a reducer by vibration and wear particles analysis

Purpose - The purpose of the paper is to provide information on wear debris on oil and vibration analysis as predictive maintenance techniques in reducer. Design/methodology/approach - The estate of a reducer is verified by analyzing the vibration and oil conditions of a test rig under well-designed conditions utilizing some predictive variables. Findings - According to the vibration and oil analysis it is found out what it was happening into the reducer without disassembling it. Practical implications - This paper demonstrates the use of oil debris analysis and vibration analysis as a technique that enhances preventive maintenance practices. The paper helps practitioners to utilize these techniques more effectively. Originality/value - This paper gives information about two predictive maintenance techniques with a test rig. © Emerald Group Publishing Limited.

Proposição de um indicador global de funcionalidade operacional para sistemas de processamento primário em instalações marítimas de produção de petróleo: estudo de caso para o sistema de separação e tratamento de óleo

Este texto verificou o tema de monitoramento de processos de produção de petróleo em plataformas no mar de forma a propor indicadores globais para monitoramento da funcionalidade dos equipamentos que envolvem estes processos, permitindo a antecipação da tomada de decisão sobre suas eminentes quedas de disponibilidade. Para tal, buscando conhecer as áreas envolvidas na gestão de uma plataforma de petróleo no mar, foram identificados os principais sistemas relacionados ao processo de produção de petróleo e optou-se pelo Sistema de Separação e Tratamento de Óleo para desenvolvimento e validação dos indicadores propostos. O Indicador Global de Funcionalidade Operacional (IGFO) foi desenvolvido a partir dos dados disponíveis para monitoramento dos processos de produção de petróleo, focado nas visões das principais áreas envolvidas na gestão da plataforma. Este indicador foi elaborado com objetivo de simplificar a análise dos processos, permitindo assim aferir o desempenho das ações de monitoramento de processos em plataformas de petróleo, de modo a atuar de forma antecipativa, contribuindo com a identificação e a disseminação das melhores práticas de manutenção preditiva. Neste aspecto, os indicadores na forma de normalização utilizada permitem comparações em diversos níveis, tanto em relação as plataformas dentro de uma empresa, como plataformas de empresas diferentes. Os indicadores globais propostos obtiveram resultados que permitiram avaliar a queda de funcionalidade dos equipamentos do Sistema de Separação e Tratamento de Óleo durante o período avaliado, servindo de base para identificação das causas das falhas destes equipamentos. Assim sendo, pôde-se perceber que a utilização dos indicadores globais identificados pode responder com sucesso às necessidades propostas

Modelagem e simulação da decomposição térmica do óleo mineral isolante aplicadas à classificação de defeitos em transformadores de potência

A análise de gases dissolvidos tem sido aplicada há décadas como a principal técnica de manutenção preditiva para diagnosticar defeitos incipientes em transformadores de potência, tendo em vista que a decomposição do óleo mineral isolante produz gases que permanecem dissolvidos na fase líquida. Entretanto, apesar da importância desta técnica, os métodos de diagnóstico mais conhecidos são baseados em constatações de modelos termodinâmicos e composicionais simplificados para a decomposição térmica do óleo mineral isolante, em conjunto com dados empíricos. Os resultados de simulação obtidos a partir desses modelos não reproduzem satisfatoriamente os dados empíricos. Este trabalho propõe um modelo termodinâmico flexível aprimorado para mimetizar o efeito da cinética de formação de sólidos como restrição ao equilíbrio e seleciona, entre quatro modelos composicionais, aquele que apresenta o melhor desempenho na simulação da decomposição térmica do óleo mineral isolante. Os resultados de simulação obtidos a partir do modelo proposto apresentaram uma melhor adequação a dados empíricos do que aqueles obtidos a partir dos modelos clássicos. O modelo propostofoi, ainda, aplicado ao desenvolvimento de um método de diagnóstico com base fenomenológica.Os desempenhos desta nova proposta fenomenológica e de métodos clássicos de diagnóstico por análise de gases dissolvidos foram comparados e discutidos; o método proposto alcançou desempenho superior a vários métodos usualmente empregados nessa área do conhecimento. E, ainda, um procedimento geral para a aplicação do novo método de diagnóstico é descrito

Prediction of integral type failures in semiconductor manufacturing through classification methods

Smart management of maintenances has become fundamental in manufacturing environments in order to decrease downtime and costs associated with failures. Predictive Maintenance (PdM) systems based on Machine Learning (ML) techniques have the possibility with low added costs of drastically decrease failures-related expenses; given the increase of availability of data and capabilities of ML tools, PdM systems are becoming really popular, especially in semiconductor manufacturing. A PdM module based on Classification methods is presented here for the prediction of integral type faults that are related to machine usage and stress of equipment parts. The module has been applied to an important class of semiconductor processes, ion-implantation, for the prediction of ion-source tungsten filament breaks. The PdM has been tested on a real production dataset. © 2013 IEEE.

On Regression Methods for Virtual Metrology in Semiconductor Manufacturing

Virtual metrology (VM) aims to predict metrology values using sensor data from production equipment and physical metrology values of preceding samples. VM is a promising technology for the semiconductor manufacturing industry as it can reduce the frequency of in-line metrology operations and provide supportive information for other operations such as fault detection, predictive maintenance and run-to-run control. The prediction models for VM can be from a large variety of linear and nonlinear regression methods and the selection of a proper regression method for a specific VM problem is not straightforward, especially when the candidate predictor set is of high dimension, correlated and noisy. Using process data from a benchmark semiconductor manufacturing process, this paper evaluates the performance of four typical regression methods for VM: multiple linear regression (MLR), least absolute shrinkage and selection operator (LASSO), neural networks (NN) and Gaussian process regression (GPR). It is observed that GPR performs the best among the four methods and that, remarkably, the performance of linear regression approaches that of GPR as the subset of selected input variables is increased. The observed competitiveness of high-dimensional linear regression models, which does not hold true in general, is explained in the context of extreme learning machines and functional link neural networks.

Extreme learning machines for virtual metrology and etch rate prediction

Virtual metrology (VM) aims to predict metrology values using sensor data from production equipment and physical metrology values of preceding samples. VM is a promising technology for the semiconductor manufacturing industry as it can reduce the frequency of in-line metrology operations and provide supportive information for other operations such as fault detection, predictive maintenance and run-to-run control. Methods with minimal user intervention are required to perform VM in a real-time industrial process. In this paper we propose extreme learning machines (ELM) as a competitive alternative to popular methods like lasso and ridge regression for developing VM models. In addition, we propose a new way to choose the hidden layer weights of ELMs that leads to an improvement in its prediction performance.