Intelligent maintenance system

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search

An intelligent maintenance system (IMS) is a system that utilizes the collected data from the machinery in order to predict and prevent the potential failures in them. The occurrence of failures in the machinery can be costly and even catastrophic. In order to avoid them, there needs to be a system which analyzes the behavior of the machine and provides alarms and instructions for preventive maintenance. Analyzing the behavior of the machines has become possible by means of advanced sensors, data collection systems, data storage/transfer capabilities and data analytic tools developed for such purpose. More information on the related data analytic tools can be found in prognostics. The aggregation of data collection, storage, transformation, analysis and decision making for smart maintenance is called an intelligent maintenance system (IMS).

Definition[edit]

An intelligent maintenance system is a system that utilizes the history data by facilitating data analytics and decision support tools to predict and prevent the potential failure of the machines. With the growing complexity of manufacturing processes and machinery, the role of monitoring systems and intelligent maintenance is becoming more crucial. The recent advancements in information technology, computers and electronics have facilitated the design and implementation of such systems.

Based on the industries growing demand in sustainability of the assets, different maintenance practices have been designed and implemented. The maintenance practices generally include condition-based maintenance (CBM), reliability-centered maintenance (RCM), corrective maintenance, scheduled or planned maintenance and predictive maintenance. An intelligent maintenance system enhances the performance of predictive maintenance systems by utilizing the advancements in computer science, electronics and information technology. The key research elements of intelligent maintenance systems consist of:

  1. "Transformation of data to information to knowledge and synchronization of the decisions with remote systems;
  2. Intelligent embedded prognostics algorithms for assessing the degradation and predicting the performance in future;
  3. Infotronics software and hardware platforms that make the products capable of thinking, linking, reconfiguring and sustaining in a networked and tether-free environment;
  4. Embedded product services and life cycle information for closed-loop product designs"[1]

Center for Intelligent Maintenance Systems[edit]

Center for Intelligent Maintenance Systems (IMS Center)[2] is an Industry/University Collaborative Research Center (I/UCRC) that consists of the University of Cincinnati, the University of Michigan, and Missouri University of Science and Technology. Since 2001, IMS has been a frontrunner in advancing methods, tools and technologies for enabling the products and systems to achieve and sustain near-zero breakdown. The techniques and technologies developed by the IMS Center have been validated in over 70 projects conducted with research and industry partners across the globe. These projects include various applications from a wide range including manufacturing, energy, rotating machinery etc. [www.imscenter.net]. A complete list of publications by IMS Center can be found in its website.[3] In 2012, a feasibility study conducted at North Carolina State University[4] showed that among the NSF I/UCRC research centers across the U.S., IMS Center has had the biggest economic impact with the benefit of $238.3 for every dollar invested on it (238.3:1). Berkley Sensors and Actuators Center (BSAC) at UC Berkley and Industry-University Center for Surfactants (IUCS) at Columbia University were ranked second and third with economic impacts of 36.2:1 and 2.8:1 respectively.

IMS procedure and methodology[edit]

The methodology for developing intelligent maintenance systems consist of finding the critical assets within a machine or process, instrumentation for collecting the suitable data, pre-processing and analyzing the collected data and extracting indicative features, applying the relevant machine-learning algorithms for health assessment, predicting the performance of the assets, and finally devising the appropriate maintenance action based on the obtained knowledge of the assets.

5S approach[edit]

The "5S" approach was devised by the IMS Center in order to address the needs of future maintenance services. This systematic approach consists of five key elements: Streamline, Smart Processing, Synchronize, Standardize and Sustain.

Watchdog Agent Toolbox[edit]

Watchdog Agent is the IMS Center's collection of tools and techniques for Prognostics and Health Management (PHM). For developing intelligent maintenance systems, a major step is to properly select such tools for data analysis and facilitate a decision support system. This toolbox can be customized and reconfigured for nearly any application – from products and assets, to complex systems, processes or manufacturing lines. The Watchdog Agent® includes four categories of analytical tools that can be used to assess and predict the performance or degradation of machines and processes by extracting the performance-related features from inputs such as measured sensor data, controller signals, and also expert knowledge, etc. Prediction results are then used for maintenance decision-making infrastructure operations IMS Brochure.[5] The Watchdog Agent® conducts its performance assessment based on the readings from multiple sensors that measure the critical properties of a process, machine or component. Since the degradation process alters the sensor measurements, the Watchdog Agent® is capable of quantitatively describing such changes in the sensor readings by using analytical tools and extracting those changes by means of appropriate tools and methods.[1][6][7]

Reconfigurable prognostics platform[edit]

An intelligent maintenance system should have the flexibility of being reconfigured for different platforms and software languages. The platforms are divided into three categories: stand-alone, embedded and cloud-based. RPP is a designed platform that can be used for heath assessment and performance prediction. RPP can be installed on equipment and is capable of converting the data to information related to the performance. Such information can then be integrated into an asset management system proper maintenance decision-making.[8]

Closed-loop life cycle design[edit]

The concept of closed-loop life cycle design refers to one of the aspects in the future of maintenance where an intelligent maintenance system can lead to redesigning the products for increased reliability and serviceability.

Smart maintenance scheduling[edit]

There are several important issues to be addressed for the effective maintenance of production systems. They include (1) how to assess the impact of a machine breakdown on the factory throughput and determine what to do first, (2) if an unscheduled machine failure occurs, or if several events occur simultaneously, which reactive maintenance job has the highest priority (3) which machine failure is most seriously endangering the production schedule, (4) where are the opportunities for maintenance without affecting production throughput, and (5) how to efficiently utilize the factory resources (e.g., maintenance crews) on the critical sections of the systems.[9] IMS center has developed maintenance policies to answer these questions. Moreover, the developed policies have been successfully implemented to several automotive manufacturing plants, which have also won three "BOSS" Kettering awards given by General Motors.[10][11][12][13]

The future of maintenance[edit]

The goal of intelligent maintenance systems is to achieve and sustain near-zero breakdown. This is defined as the future of maintenance in which an intelligent system can equip the machines and systems to achieve highest performance and near-zero breakdown with self-maintenance capabilities. Such goal can be achieved by the transformation of raw data to valuable information regarding the current and future condition of the asset or process being monitored. Such vision has been introduced by Center for Intelligent Maintenance Systems (IMS) at the University of Cincinnati, and is summarized as a transformation of fail and fix maintenance practices to predict and prevent by focusing on frontier technologies in embedded and remote monitoring, prognostics, and intelligent decision support tools. The center also has trademarked Watchdog Agent prognostics toolbox[14] and device-to-business (D2B) infotronics platform[15] for e-maintenance systems. The "unmet needs" to be addressed in the scope of future maintenance rely on three aspects:

  • Self-assessment: In future, the assets need to have embedded intelligent agents for nonstop monitoring that can guarantee the sustainability.
  • Only handle information once (OHIO): The flow of information from the field (user) to service/business systems is referred to as Only Handle Information Once (OHIO).
  • All-time readiness: The obtained information can be used for predicting, optimizing and planning maintenance scheduling for achieving near-zero breakdown.[16]

E-manufacturing and e-maintenance[edit]

With evolving applications of tether-free communication technologies e.g. Internet, the e-intelligence is gaining bigger impact on industries. Such impact has become a driving force for companies to shift the manufacturing operations from traditional factory integration practices towards an e-factory and e-supply chain philosophy. Such change is transforming the companies from a local factory automation to a global business automation. “It transforms companies from a local factory automation to a global enterprise and business automation. The technological advances for achieving this highly collaborative design and manufacturing environment are based on multimedia-type information-based engineering tools and a highly reliable communication system for enabling distributed procedures in concurrent engineering design, remote operation of manufacturing processes, and operation of distributed production systems.”[17] The goal of e-manufacturing is to plant floor assets, predict the deviation of the quality of the products and possible loss of any equipment. This brings about the predictive maintenance capability for the machines. Utilizing such e-maintenance capability with a dynamic rescheduling of maintenance and operation can lead to achieving lower downtime.[18]

The major functions and objectives of e-manufacturing are: “(a) provide a transparent, seamless and automated information exchange process to enable an only handle information once (OHIO) environment; (b) improve the utilization of plant floor assets using a holistic approach combining the tools of predictive maintenance techniques; (c) links entire SCM operation and asset optimization; and (d) deliver customer services utilizing the latest predictive intelligence methods and tether-free technologies”.[17] The e-Maintenance infrastructure consists of several information sectors:[19][20]

  • Control systems and production schedulers
  • Engineering product data management systems
  • Enterprise resource planning (ERP) systems
  • Condition monitoring systems
  • Maintenance scheduling (CMMS/EAM) systems
  • Plant asset management (PAM) systems

Annual PHM data challenge[edit]

As an effort to promote the science of PHM and contribute to the development of intelligent maintenance systems, the Prognostics and Health Management Society (PHM Society) sponsors an annual competition data challenge to both professional and student participants. Participants use the available data and their prior experience and algorithm knowhow for developing their health monitoring algorithms for the given application. The accuracy of the health monitoring algorithm output is compared with the true health state; the true health state is unknown to the participants but is known by the organizers of the data challenge.

2008: jet engine RUL estimation[edit]

The 2008 data challenge was the Remaining Useful Life (RUL) prediction of jet engines. The simulated data set consisted of multivariate time series that were collected from multiple engine units. There were three operational settings that have a substantial effect on unit performance. The data for each cycle of each unit included the unit ID, cycle index, 3 values for the operational settings and 21 values for 21 sensor measurements. Each unit started with unknown degrees of initial degradation and manufacturing variation. The units were operating normally at the start of each time series, and developed a fault at some point during the series. The data set was further divided into training and testing subsets. In the training data set (218 units), the fault grew in magnitude until system failed. There is no "hard" failure in the data set; however, the remaining useful life of the last operational cycle of each unit in the training data was considered as zero. In the testing data set, the time series ended some time prior to system failure. The objective of the problem was to predict the number of remaining operational cycles before failure in the testing data set. A portion of the testing data set (218 units) was provided first to assist algorithm development and the rest (435 units) was released towards the end of the competition as the validation data set to score the algorithm.

Winning team in student category and overall: Tianyi Wang from Center for Intelligent Maintenance Systems at the University of Cincinnati. A publication related to this study can be found in.[21] Winning team in professional category: Felix. O. Heimes from BAE Systems, Electronics and Integrated Solutions, Johnson City, NY. A publication related to this study can be found in.[22]

2009: gearbox health assessment[edit]

The 2009 data challenge application dealt with gearbox health monitoring and diagnosis. The objective of the competition was to develop accurate health monitoring and diagnostic methods for gearbox components. The data provided consisted of 560 data sets, in which the gearbox was tested under 5 different speeds, 2 different loads, and two different gear types. The 560 data sets consisted of data in which the gearbox was under different conditions of the mechanical components. No training data set was provided for the data challenge and fault detection was to be performed by analyzing the available vibration and speed signals and comparing the results against the known gearbox signatures in the literature.[23] Winning team in professional category and overall: Fangji Wu from Center for Intelligent Maintenance Systems, University of Cincinnati. A publication related to this study can be found in.[24] Winning team in student category: Hassan Al-Atat and David Siegel from Center for Intelligent Maintenance Systems at the University of Cincinnati. A publication related to this study can be found in.[23]

2010: RUL estimation of CNC milling machine cutters[edit]

The 2010 challenge was focused on RUL estimation for a high-speed CNC milling machine cutters using dynamometer and accelerometer data. The provided data consisted of collected force and acceleration signals in three directions along with the acoustic emission signals. The data was provided from six cutters and the participants were expected to estimate the remaining useful life of the cutters as they cut the work-piece. The actual wear measurement for three of the cutters was already provided along with the data to be used for training the models. A full description of this competition including its dataset can be found in.[25] Winning team in professional category and overall: Sreerupa Das and Gregory Harrison at Lockheed Martin Corporation, in Orlando, FL[26] Winning team in student category: Gang Liu at the University of New Orleans in New Orleans, LA[26]

2011: fault detection of anemometers[edit]

The 2011 data challenged dealt with anemometer sensor health assessment for wind energy applications. The objective of this challenge was to develop accurate sensor health monitoring methods for specific applications in anemometers. The data set provided for this challenge consisted of data from anemometers from two different configurations. The two configurations consist of the “paired” data set in which two anemometers are positioned at the same height, and the “shear” data set, which includes an array of anemometers at different heights. Various wind speed statistics, wind direction, and ambient temperature information are provided, in which the objective is to classify the anemometer health status during a set of samples from a 5 day period.[27] More information about this data challenge can be found in.[28] Winning team in student category and overall: David Siegel from Center for Intelligent Maintenance Systems at University of Cincinnati[29] Winning team in professional category: Danny Parker from Miltec Research and Technology[29]

2012: RUL estimation of bearings[edit]

In 2012, no data challenge was organized by [www.phmsociety.org PHM Society] but instead, IEEE Reliability Society and FEMTO-ST Institute organized a data challenge focusing on the RUL estimation of bearings.[30] The data was collected from 3 different loads (rotating speed and load force). Participants were provided with 6 run-to-failure datasets in order to build their prognostics models, and were asked to estimate accurately the RUL of 11 remaining bearings. Monitoring data of the 11 test bearings were truncated so that participants were supposed to predict the remaining life, and thereby perform RUL estimates. Also, no assumption about the type of failure to be occurred was given. The challenge datasets were characterized by a small amount of training data and a high variability in experiment durations (from 1h to 7h). Thereby, performing good estimates was quite di cult and this made the challenge more exciting. Note also that theoretical framework (L10, BPFI, BPFE, etc.) mismatches the experimental observations.[31] Winning team in industrials category: Sergey Porotsky from A.L.D Ltd.[31]

Winning team in academic category: Arvind Sai Sarathi Vasan from Center for Advanced Life Cycle Engineering (CALCE), University of Maryland[31]

2013: remote monitoring diagnosis and maintenance action recommendation[edit]

The 2013 data challenge focused on maintenance action recommendation based on historical cases and the algorithms were evaluated on their ability to recommend confirmed problem types.[32] This data was generated from an industrial piece of equipment with a sensor network to measure several parameters and an onboard condition monitoring system. The measured data goes through a control logic in order to monitor the equipment's operating regime. At any time instant when some of these parameters meet a specific condition, the control system generates a unique event ID/code. Each case is described by a set of event codes which characterize the atypical operating condition of the equipment. Some of these cases with specific event code combinations may be operationally significant and could be indicative of problem types, some of which are assumed to be known to the subject matter experts.[33] More information regarding this data challenge can be found in.[34] Winning team: Santanu Das from NASA Ames Research Center, CA[35] Runner-up: Vassilis Katsouros, Vassilis Papavassiliou, and Christos Emmanouilidis from Athena Research and Innovation Center[35]

NASA's PHM data repository[edit]

The Prognostics Data Repository is a collection of data sets that have been donated by various universities, agencies, or companies. The data repository focuses exclusively on prognostic data sets, i.e., data sets that can be used for development of prognostic algorithms. Mostly these are time series of data from some nominal state to a failed state. The collection of data in this repository is an ongoing process. The contributors to this collection of data sets include Center for Intelligent Maintenance Systems (IMS), BEST Lab at UC Berkley, Prognostics CoE at NASA Ames and more. More information regarding this data repository including the data sets can be found in.[36]

Most cited papers[edit]

The most cited papers about industry 4.0 are in the following lists:

See also[edit]

References[edit]

  1. ^ a b Lee, Jay; Jun Ni; Dragan Djurdjanovic; Hai Qiu; Haitao Liao (August 2006). "Intelligent prognostics tools and e-maintenance". Computers in Industry. 57 (6): 478. doi:10.1016/j.compind.2006.02.014. Retrieved 23 January 2014. 
  2. ^ "Center for Intelligent Maintenance Systems". IMS Center. Retrieved 23 January 2014. 
  3. ^ "List of published, accepted or submitted journal and conference papers". IMS Publications. Retrieved 23 January 2014. 
  4. ^ Gray, Denis O.; Drew Rivers; George Vermont (May 2012). "Measuring the Economic Impacts of the NSF Industry/University Cooperative Research Centers Program: A Feasibility Study" (PDF): xiii. Retrieved 23 January 2014. 
  5. ^ http://www.imscenter.net/Resources/brochure_2012_red_final.pdf
  6. ^ Djurdjanovic, Dragan; Jay Lee; Jun Ni (July–October 2003). "Watchdog Agent—an infotronics-based prognostics approach for product performance degradation assessment and prediction". Advanced Engineering Informatics. 17 (3-4): 111. doi:10.1016/j.aei.2004.07.005. Retrieved 23 January 2014. 
  7. ^ Takata, S; F. Kirnura; F. J. A. M. van Houten; E. Westkamper; M. Shpitalni; D. Ceglarek; J. Lee (2004). "Maintenance: Changing Role in Life Cycle Management". CIRP Annals - Manufacturing Technology. 53 (2): 652. doi:10.1016/S0007-8506(07)60033-X. 
  8. ^ Liao, Linxia; Jay Lee (January 2010). "Design of a reconfigurable prognostics platform for machine tools". Expert Systems with Applications. 37 (1): 241. doi:10.1016/j.eswa.2009.05.004. 
  9. ^ Ni, Jun; Xiaoning Jin (2012). "Decision support systems for effective maintenance operations". CIRP Annals-Manufacturing Technology. 61 (1): 411–414. doi:10.1016/j.cirp.2012.03.065. Retrieved 31 January 2014. 
  10. ^ Chang, Qing; Jun Ni; Pulak Bandyopadhyay; Stephan Biller; Guoxian Xiao (2007). "Supervisory factory control based on real-time production feedback". Journal of Manufacturing Science and Engineering. 129 (3): 653–660. doi:10.1115/1.2673666. 
  11. ^ Chang, Qing; Jun Ni; Bandyopadhyay Pulak; Biller Stephan; Guoxian Xiao (2007). "Maintenance opportunity planning system". Journal of manufacturing science and engineering. 129 (3): 661–668. doi:10.1115/1.2716713. 
  12. ^ Chang, Qing; Jun Ni; Pulak Bandyopadhyay; Stephan Biller; Guoxian Xiao (2007). "Maintenance staffing management". Journal of Intelligent Manufacturing. 18 (3): 351–360. doi:10.1007/s10845-007-0027-7. Retrieved 31 January 2014. 
  13. ^ Yang, Zimin; Qing Chang; Dragan Djurdjanovic; Jay Lee; Jun Ni (2006). "Maintenance Priority Assignment Utilizing On-line Production Information". Journal of Manufacturing Science and Engineering. 129 (2): 435–446. doi:10.1115/1.2336257. Retrieved 31 January 2014. 
  14. ^ "Development of Smart Prognostics Agents (WATCHDOG AGENT ® )" (PDF). IMS Center. Retrieved 23 January 2014. 
  15. ^ "Device-to-Business (D2B)™ Platform" (PDF). IMS Center. Retrieved 23 January 2014. 
  16. ^ "Advanced Prognostics for Smart Systems" (PDF). IMS Center. Retrieved 23 January 2014. 
  17. ^ a b Koc, Muammer; Jun Ni; Jay Lee; Pulak Bandyopadhyay. "Introduction to e- Manufacturing" (PDF): 97-2. 
  18. ^ Lee, Jay (December 2003). "E-manufacturing—fundamental, tools, and transformation". Robotics and Computer-Integrated Manufacturing. 19 (6): 505. doi:10.1016/S0736-5845(03)00060-7. Retrieved 23 January 2014. 
  19. ^ Moore, W. J.; A. G. Starr (August 2006). "An intelligent maintenance system for continuous cost-based prioritisation of maintenance activities". Computers in Industry. 57 (6): 596. doi:10.1016/j.compind.2006.02.008. Retrieved 23 January 2014. 
  20. ^ Baldwin, Robert C (2001). "Enabling an e-Maintenance infrastructure". Maintenance Technology. 12. 
  21. ^ Wang, Tianyi; Jianbo Yu; David Siegel; Jay Lee (6–9 October 2008). "A similarity-based prognostics approach for Remaining Useful Life estimation of engineered systems". International Conference on Prognostics and Health Management: 1–6. doi:10.1109/PHM.2008.4711421. Retrieved 23 January 2014. 
  22. ^ Heimes, F. O. (6–9 October 2008). "Recurrent neural networks for remaining useful life estimation". International Conference on Prognostics and Health Management: 1–6. doi:10.1109/PHM.2008.4711422. 
  23. ^ a b Al-Atat, Hassan; David Siegel; Jay Lee (2011). "A Systematic Methodology for Gearbox Health Assessment and Fault Classification" (PDF). International Journal of Prognostics and Health Management. 2: 2153–2648. 
  24. ^ Wu, Fangi; Jay Lee (2011). "Information Reconstruction Method for Improved Clustering and Diagnosis of Generic Gearbox Signals" (PDF). International Journal of Prognostics and Health Management. 04. 
  25. ^ "2010 PHM Society Conference Data Challenge". PHM Competition 2010. PHM Society. Retrieved 23 January 2014. 
  26. ^ a b "2010 PHM Data Challenge Closure Announcement". PHM Society. Retrieved 23 January 2014. 
  27. ^ Siegel, David; Jay Lee (2011). "An Auto-Associative Residual Processing and K-means Clustering Approach for Anemometer Health Assessment" (PDF). International Journal of Prognostics and Health Management. 2 (2): 50–61. 
  28. ^ "2011 PHM Society Conference Data Challenge". PHM Data Challenge 2011 Announcement. PHM Society. Retrieved 23 January 2014. 
  29. ^ a b "2011 Data Challenge Closure Announcement". PHM Society Data Challenge 2011. PHM Society. Retrieved 23 January 2014. 
  30. ^ "IEEE PHM 2012 Data Challenge". IEEE PHM 2012 Announcement. FEMTO-ST. Retrieved 23 January 2014. 
  31. ^ a b c "IEEE PHM 2012 Prognostic challenge Outline, Exp eriments, Scoring of results, Winners" (PDF). IEEE Reliability Society and FEMTO-ST. Retrieved 23 January 2014. 
  32. ^ Katsouros, Vassilis; Vassilis Papavassiliou; Christos Emmanouilidis (2013). "A Bayesian Approach for Maintenance Action Recommendation" (PDF). International Journal of Prognostics and Health Management. 4 (2). Retrieved 23 January 2014. 
  33. ^ Das, Santanu (2013). "Maintenance Action Recommendation Using Collaborative Filtering" (PDF). International Journal of Prognostics and Health Management. 4 (2). Retrieved 23 January 2014. 
  34. ^ "2013 PHM Society Conference Data Challenge". PHM Society. Retrieved 23 January 2014. 
  35. ^ a b "PHM Data Challenge Winners Announced". PHM Society. Retrieved 23 January 2014. 
  36. ^ "Prognostics Data Repository". Prognostics Center of Excellence. NASA. Retrieved 23 January 2014. 

Further reading[edit]

  1. K. A. H. Kobbacy et al., “Towards an intelligent maintenance optimization system"
  2. M. J. Ashby et al., “Intelligent maintenance advisor for turbine engines”, The Journal of the Operational Research Society, vol. 46, No. 7 (July 1995), 831-853.
  3. J. Lee et al., “Informatics Platform for Designing and Deploying e-Manufacturing Systems”, Collaborative Design and Planning for Digital Manufacturing. Springer London, 2009. 1-35.
  4. A. K. S. Jardine et al., “A review on machinery diagnostics and prognostics implementing condition-based maintenance”, Mechanical Systems and Signal Processing 20 (2006) 1483–1510.
  5. R. C. M. Yam et al., “Intelligent Predictive Decision Support System forCondition-Based Maintenance”, Int J Adv Manuf Technol (2001) 17:383–391
  6. A. Muller et al., “On the concept of e-maintenance: Review and current research”, Reliability Engineering and System Safety 93 (2008) 1165–1187
  7. A. Bos et al., “SCOPE: An Intelligent Maintenance System for Supporting Crew Operations”, AUTOTESTCON 2004. Proceedings. IEEE, 2004.