SOFTWARE UPDATE
results. Putting all the data in one place without a way to make sense of it isn’t a result. It’s been fascinating to watch. Twelve to 24 months ago, it seemed every large manufacturer had tasked a team with figuring out how to make use of all their factory data. The companies that moved forward did one of two things. Some took a broad approach and tried everything they could get their hands on, without attacking a specific problem that needed to be resolved. Those companies often failed to make much progress. Other companies took an alternate approach that we’ve found to be much more successful: pick a specific, definable business problem, apply technology to solve the problem, and build up from there. In order for a project to succeed, there needs to be a real business problem to be solved. If the project is properly defined,
the results can be seen in as little as a few weeks. We worked with a major industrial company that had a high scrap rate with the silicon chips used in its pressure sensors. The company’s manufacturing excellence team was tasked with finding the root cause, but analyzing production and quality data at scale was a challenge due to manually-intensive processes. We brought in our analytics platform
in the same way financial software has long been able to let companies zoom in and out at will. With the availability of real-time data and analytics on pro- duction and quality, those silos are breaking down. We’ve even seen companies create new functions that meld OT and IT. ME: Your company was recently cited by analysts at LNS
Research as a key technology player to watch. What’s the fu- ture direction for Sight Machine’s manufacturing technology? Sobel: In traditional structured data software, the bound- aries of data types and analytical processes are built into the software. If a user wants to bring in a new type of data or produce a different type of report, it often requires an IT
New manufacturing analytics features in Sight Machine 2.0 offer users a “digital twin” of plant-floor operations.
to automate the data acquisition, analysis and visualization processes. Within three weeks the company was able to cut the scrap rate by 30%. They also identified a process improvement in a very sophisticated production process they would never have been able to see otherwise. Several management-related factors helped the company achieve rapid results. The project had a cross-functional executive champion with accountability and responsibility, and Sight Machine was given access to engineers and data owners with knowledge of the manufacturing OT [operational technology] and IT.
Once companies embark on this path, they often find a need to break down silos and improve their collaboration across functions. Historically, factory floor data has been the domain of the operations technology team and was rarely integrated with the corporate information systems that track financial, customer, marketing and human re- sources data. There was no good way to roll up that data
22
AdvancedManufacturing.org | December 2016
process involving project managers and offshore developers and timelines of weeks or months. That’s not going to work with manufacturing analytics. It’s quite common for new sen- sors to be put on a machine, and then that new data needs to be incorporated, even if it has an entirely different format than the other data you’re already collecting. In a traditional database structure, that is going to break your model. Modern data analytics software puts much more con-
trol in the hands of users. Big data by definition involves a large variety of data types [along with high volume and high velocity]. To be effective, artificial intelligence and analytics platforms need to support self-serve analytics. That means letting non-specialist users bring in new data and data types and then design and perform the analysis themselves. In the first half of 2017, we’re releasing Sight Machine Commander, which will let our clients and services integrators add new data sources and data types themselves, and will let them control how that data is conditioned and analyzed.
Image courtesy Sight Machine Inc.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84
