In the past 2 years, ProSensus has witnessed continued investment in Industry 4.0 and analytics, with many of our clients implementing dedicated digitalization teams.
The abundance of literature, conferences, and general buzz, in combination with recent interaction with our clients, leaves no room for doubt: leading manufacturers are acting on this initiative.
These front-runners have approached ProSensus with a common theme of challenges and frustrations, including:
“We have data, we need to learn to use it better.”
“We need to up our game.”
ProSensus has responded to these needs by tailoring consulting services to progress our clients’ digitilization journey. Regardless of the clients’ stage in this journey, ProSensus gears our services towards achieving quick-wins, in order to provide informed, data-guided decisions for the path forward.
Why Should Digitalization Groups Embrace Multivariate Analysis?
“I have no doubt that the use of the massively increasing amounts of data that we are collecting on processes will be the most important factor influencing control engineering practice over the next few decades.” – Dr. John F. MacGregor, Founder, ProSensus (CONTROL magazine, March 2013)
ProSensus believes that properly executed multivariate analysis should be the first step in the majority of manufacturing analytics projects.
1. MVA’s intuitive plots can be used to quickly identify data integrity issues that commonly plague datasets (such as inconsistent units, unexpected trends, high percentage of missing data, etc.).
2. MVA can readily be used for the critical task of validating the existence of expected correlation in the dataset (relationships between input and output variables) with process experts.>
Extensive modeling should not be performed until these 2 steps are complete.
The Digitalization Journey
These manufacturers are often producing discrete components, which is typically a manually intensive processes where little data is being collected or historized. ProSensus accelerates the digital journey by providing guidance on what existing measurements should be collected, where additional sensors (such as machine vision) could be beneficial, and by consulting on data structure. We have seen many limitations in customer datasets over the past 15 years of consulting projects. Avoiding these in early stages can greatly simplify future analysis and significantly improve the value that can be extracted.
Most manufacturers fit into this category, where process measurements and quality attributes are presently being logged. However, they may still face challenges such as: insufficient measurements, inability to trace product through multiple production units, uncertain analytics objectives, etc. In these cases, ProSensus typically starts with a data-audit to highlight any deficiencies in the data, suggest improvements, and establish what can be achieved with the data in the current format. With most of our clients, even an imperfect dataset can be analyzed to deliver valuable insights and provide guidance on the path forward.
These clients have highly automated processes, with a large amount of informative measurements presently being stored in an appropriate manner that can be extracted for analysis. These manufacturers are typically looking to do more with the massive amounts of data that they already have available, but are not yet regularly analyzing. Our first step in these cases is typically an accelerated modeling session to obtain quick-wins on a few key datasets, provide training, and establish further analyses that could be performed.