MTConnect Challenge For Machine Monitoring & Big Data Prepared by B. Blomquist 763-263-5306 Bill.blomquist@remmele.com May - 2013 Machine Monitoring & Big Data Page 1 1.0 Abstract The objective of this paper is to demonstrate the capability of MTConnect and Big Data analysis to conduct shop floor machine monitoring and analysis, accurately and inexpensively. The availability of collecting machine data via the MTConnect protocol, visual data acquired by cameras and from multiple business systems will enable accurate analysis of machine utilization and shop floor efficiency. Current shop floor machine data collection struggles with finding out why the machine is not running from a plethora of diverse reasons. The conflict resides in requiring the machine operator to manually input the reason for why the machine is not running. When the data used for analysis is dependent of manual input, the resulting report is compromised and cannot be trusted for production decisions. Shop managers want to know the foremost reason why the machine is not running and implement a fix. An application which executes unobtrusively and reports accurate analysis is of high value. 2.0 Application Goal Build an unobtrusive application, which provides accurate reports of why the machine is/was not running, obtained by analyzing a mash-up of data from multiple points. The application will require no interaction with the machine operator. The application will determine who is present at what times at the machine. (Operators, managers and maintenance personnel) The application will collect actual near real time machine data from the machine control using MTConnect, business systems and from multiple low cost vision cameras which will be stationed at various points around the machine. The application will utilize ‘Big Data’ analytics to determine the reason the machine was/is not running and why it is/was running. The application will create a holistic report of what took place at the machine for any given time period, or manufacture of a part. The application will also provide reports that reflect if a specific reason for why the machine is not running is trending better or worse. Machine Monitoring & Big Data 3.0 Page 2 Technical Requirements The application will require that the machine control is provisioned with an MTConnect adapter. Also all accessory sensors, like spindle vibration signal, oil chiller temperature, tooling RFID chip etc. will provide data to the application via the MTConnect protocol. Other optional inputs of data may be obtained from: Materials Resource Planning (MRP)system Manufacturing Execution Systems (MES)system Product Lifetime Management (PLM)system Engineering Resource Planning (ERP)system Product Data Management (PDM)system Emails Phone calls Personnel RFID badges The more data which can be collected, the more accurate the ‘Big Data’ analytic engine will be able to mash-up the data and determine the reason for downtime or uptime. Visual data will be obtained by cameras monitoring key points of the machine/device Spindle or work effector (currently used at RTI Remmele Inc) Operator control interface area Pallet shuttle or work queue area Cutting Tool loading area Chip conveyor or waste removal area Machine/device, maintenance access panel doors Operator work area Other locations of visual data collection: Shipping and Receiving Inspection department work area Coordinate Measuring Machine Tool crib area All the technical requirements needed to realize this application can be met at this time. Visual data will need to be analyzed with vision algorithms There are multiple existing visual data analytical engines. (ref. Sight Machine) A secure cloud server infrastructure will be needed to store very large amounts of scalable data. There are multiple existing cloud infrastructures Amazon cloud services Microsoft Azure services The ‘Big Data’ backend engine will execute on a cloud based infrastructure to maximize processing power scalability. There are existing examples of ‘Big Data’ analytical applications in use. (ref. IBM Big data development platform) Machine Monitoring & Big Data 4.0 Benefits The application will provide multiple benefits: Accurate reasons for machine/device downtime Accurate reasons for machine /device uptime Unobtrusive- No required production personnel input required Ability to see if process improvement is working or not Ability to analyze manufacturing problems in a detailed holistic view Ability to prove actual part process mapping (receipt of process) Ability to provide process simulation software ‘current state’ of process Page 3