Issue link: https://iconnect007.uberflip.com/i/948150
62 SMT007 MAGAZINE I MARCH 2018 not. The clearer meaning the elements of the stored data have, the easier it is to create value, which is the politically correct way of saying "Garbage in, garbage out". Lightening the Cloud There are many applications and solutions that are ideally placed in the cloud, especially where potentially huge amounts of data stor- age will be processed. Examples could be capacity plan reports, long term quality and productivity metrics, traceability data analy- sis. The question is then to decide what soft- ware should exist in the cloud to provide these functions. There are many enterprise-grade pack- ages available, which provide easy ways to configure data search criteria and the format for reporting, such as business intelligence (BI). Unfortunately, the practical bottleneck rules come into play. For such off the shelf analytic systems to work effectively, the data itself must be highly organized and meaning- ful. Data obtained directly from manufactur- ing processes is usually quite the opposite. For example, a machine that reports events such as "stop other than error" or "waiting for PCB" have very little meaning in isolation. When put into the cloud, there would need to be extensive processing to find out the cause, for example why the machine had to wait for a PCB, which potentially comes from a variety of sources. A massive and complex analysis of a whole set of such simple events, of which there can be thou- sands every hour, is needed to try to figure out the true information for every step of the analysis. A typical BI application is going to be impossi- bly stretched to provide an algorithm that can do this effectively, under- standing the nature of raw produc- tion data. An alternative to the off the shelf approach would be to develop a bespoke "raw production data processor" in the cloud which would be very complex and expen- sive to develop and support. Intelligent processing of data locally at the production site before the data goes into the cloud is critical for the ability for cloud-based solutions to work effectively, exactly like the advanced SEO data associ- ated with web-pages. The first stage is to listen and piece together the multiple disparate data elements in strict time sequence, derived from different kinds of machines, bar-code readers, sensors, material preparation transactions, etc. Qualified events can then be created, which is the conversion of raw data to create informa- tion. These events are meaningful and action- able pieces of information. When out into the cloud, a great deal more value can be obtained with a mere fraction of the processing, and hence cost. The Hybrid Theory Processing data locally is already a part of an advanced MES system that has direct connec- tions and the ability to process data directly from both automated and manual production operations, as well as to manage them. Start- ing with dashboards, which need to be fed meaningful events as they happen, then gener- ating alerts based on a live situation, the real- time data processing requirement is already well established. We are already experiencing the extension of this processing, as we see the use of data in real-time by a new generation for "smart" systems such as those associated with