Predictive manufacturing system

The predictive manufacturing system (PMS) is a manufacturing paradigm wherein the system is programmed with certain amount and type of intelligence that allows assets to estimate their own condition, detect the presence of a fault or an anomaly, infer future fault events and even diagnose potential root cause of the problem. The goal of predictive manufacturing system (PMS) is to enable machine and systems with “self-aware” capabilities.[1] The core technology of PMS in order to deliver “self-aware” capabilities is the smart computational agent that will contain advanced analytics to conduct predictive modeling functionalities.[2]

Background

Transparency in an organization allows unraveling and even quantifying sources of uncertainty. It is an objective estimation of the company’s manufacturing capability and readiness.[3] Most manufacturing strategies (such as mass production, lean manufacturing, flexible manufacturing, and reconfigurable manufacturing) haphazardly assume continuous equipment availability and constant maximum performance, which may never occur in a real factory. Manufacturing machines, like any other mechanical engineering assets, are prone to wear and tear that inhibit uninterrupted uptime. In order to achieve transparence within a facility, manufacturing industry is then asked to metamorphose into predictive manufacturing system so that data can be systematically processed into information that can explain the uncertainties and thereby enable users of the manufacturing assets to make more informed and evidence-based decisions.

The foundations of predictive manufacturing have already been laid out with the aggressive adoption of “Internet of Things” (IoT) principles wherein essential structures such as smart sensor networks and smart machines have become more prevalent.[4] Such scenario enables for seamless aggregation of data from assets (or asset fleet) on to a central location for appropriate processing. With the discovery and development of more advanced analytics, such data can then be transformed into information that can provide clarity or transparency, eventually addressing the uncertainties in the factory.

Framework

An illustration of the conceptual framework of a predictive manufacturing system is shown in Figure 1. [5] The system starts out with a data acquisition system that can either be built-in by the original equipment manufacturer (OEM) or a third party provider. Using appropriate sensor assemblies, various signals such as vibration, pressure, temperature, etc. can be recorded. The types of signals and data acquisition parameters are determined by the application and the failure modes of the asset being monitored. Communication protocols, such as MTConnect and OLE-DB Process Control or OPC, can help users to acquire process or controller signals. Such data can provide context as to the type of action/function the machine was performing when sensor data was being collected. Imagine when all data are gathered and combined with all the assets in a facility. When all the data are aggregated, this phenomenon is called “Big Data” because of the volume of data collected, velocity by which data is being received and variety of data that are being collated. Such phenomenon requires new paradigms and approaches for analysis instead of statistical process control or other traditional statistical analysis techniques. The new transforming agent then consists of several components: an integrated platform, predictive analytics and visualization tools. The deployment platform is naturally selected based on several factors such as speed of computation, investment cost, ease of deployment for scaling purposes and update, etc. The actual processing or transformation of big data into useful information is performed by utilizing predictive analytics such as the tools found in the Watchdog Agent® toolbox that has been developed by researchers at the National Science Foundation (NSF) Industry/University Research Cooperative Center (I/UCRC) for Intelligent Maintenance Systems (IMS) since 2001.[6][7] There are also other predictive analytic providers such as IBM, Hadoop, SAS, SAP, etc. The algorithms found in the Watchdog Agent® can be categorized into four sections namely, signal processing and feature extraction, health assessment, performance prediction and fault diagnosis. By utilizing visualization tools, health information such as current condition, remaining useful life estimation, root cause, etc., can be effectively conveyed using radar charts, fault maps, risk charts and even health degradation curves. The calculated health information can then be forwarded or made available to existing company management systems such as enterprise resource planning system (ERP), manufacturing execution system (MES), supply chain management system (SCM), customer relation management system (CRM), product lifecycle management system (PLM) to achieve overall enterprise control and optimization. With manufacturing transparency, management then has the appropriate information (such as actual condition and state of machines, not just cycle times) to determine factory-wide overall equipment effectiveness (OEE). By knowing when assets will fail, equipment can then be managed more effectively with just-in-time maintenance. Finally, if historical health and failure modes can be compiled, such information can then be fed back to the equipment designer for closed loop lifecycle redesign

See also

References

  1. Lee, Jay; Lapira ER; Bagheri B; Kao HA (2013). "Recent advances and trends in predictive manufacturing systems in big data environment". Manufacturing Letters. 1 (1): 38–41. doi:10.1016/j.mfglet.2013.09.005.
  2. Lee, Jay; Lapira E; Yang S; Kao Ha (2013). "Predictive manufacturing system trends of next generation production systems". the Proceedings of the 11th IFAC workshop on intelligent manufacturing systems. 11 (1): 150–156. doi:10.3182/20130522-3-br-4036.00107.
  3. Lee, Jay; Lapira E. "Predictive factories: the next transformation". Manufacturing leadership journal. Frost Sullivan.
  4. Chui M, Loffler M, Robert R. "The internet of things". McKinsey Quarterly No. 2. Retrieved 2010. Check date values in: |access-date= (help)
  5. Lee, Jay; Lapira E (2014). "Recent advances and trends in predictive manufacturing in Industry 4.0 environment". Uptime Magazine: 16–21.
  6. Djurdjanovic, Dragan; Lee J; Ni J (2003). "Watchdog Agent – an infotronics-based prognostics approach for product performance degradation assessment and prediction.". Advanced Engineering Informatics. 17 (3-4): 109–125. doi:10.1016/j.aei.2004.07.005.
  7. Lee, Jay; Djudjanovic D; Qiu H; Liao H (2006). "Intelligent prognostics tools and e-maintenance. Computers in Industry". Computers in Industry. 57 (6): 476–489. doi:10.1016/j.compind.2006.02.014.
This article is issued from Wikipedia - version of the 6/5/2016. The text is available under the Creative Commons Attribution/Share Alike but additional terms may apply for the media files.