I blogged last week about the need for more intelligence in edge devices and smart sensors. Additionally, I commented in an earlier blog include how completed operational and executed work records, quality assurance records, work flow histories, operational deviations and variations, engineering changes, and many other records related to the production process represented the real big data for manufacturing. The point being that this big data for manufacturing was the real treasure trove of information that would allow advanced analytics applications to actually optimize and determine best
practices for the production processes. In other words, if you want to actually implement continuous process improvement one has to examine the complete production process record history to discover both the flaws (risks) and the best methods in the design/build lifecycle.
In order to follow up on this concept, I would like to take a bit of a deeper (but not too deep) dive into the topic of performance analytics, operational intelligence, or closed-loop PLM, all apt descriptions of this notion of process improvement and validating as-built to as-designed. This is where I see the real payback of advanced analytics, that is, going beyond predictive to prescriptive analytics, where we bring together big data, statistical sciences, rules-based logic, and machine learning to empirically discover and reveal the origins of the complex problems, and then determine decision-based options to resolve them.
According to the Bureau of Labor Statistics, manufacturing, both discrete and process, have the most stored data (well over 1500 petabytes) of any industrial or business sector. One could make the case that this represents a digital brain trust or the primary source of basically unstructured data that needs to aggregated, analyzed, and converted into actionable information.
Moreover, when this information is placed in the context of a design/build lifecycle, it becomes a closed-loop mechanism that is connected by a digital thread that includes product development, manufacturing, and services in the field. In essence, the data that is held in a repository which is the result of manufacturing execution operations records becomes a source for operational intelligence, product performance and production process improvement.
Because a typical large OEM manufacturer can accumulate such extremely large repositories of data the only effective way to analyze the sum total of these records is through large scale pattern matching technology enabled by current machine learning algorithms. What ML already does exceeding well—and will get even better at—is relentlessly processing any amount of data and every combination of variables. Eventually, pattern matches are made and predictable outcomes form that, in the case of product process records, can result in a set of rules and best practices that lead to overall process improvement and better product designs. The process is straightforward and relatively simple in theory, but can involve some rather complex algorithms and access to a lot of data.
One area of research currently receiving a lot of attention and associated with ML is deep learning. Deep learning itself is a revival of an even older idea for computing: neural networks. One characteristic of deep learning is that it gets better and better as you feed it more data, which is clearly the case with accumulated manufacturing execution data. In this sense,
deep learning achieves what computer scientists were trying to do with rules-based inferencing engines years ago. Since deep neural nets are really good at finding patterns in unstructured data, the bigger the data is in volume and density, the more the system will learn. One of the significant benefits of deep learning is that it can be applied generally to many industries and problems and does not require extensive domain expertise to create effective solutions. These technologies applied to manufacturing are spawning an emerging trend toward operational intelligence and product performance platforms that will truly improve both products and production.
Reprinted with permission, original blog was posted here
About ARC Advisory Group (www.arcweb.com): Founded in 1986, ARC Advisory Group is a Boston based leading technology research and advisory firm for industry and infrastructure.
For further information or to provide feedback on this article, please contact firstname.lastname@example.org