Thursday, March 31, 2016

Data Visualization and Smart Manufacturing

There is a lot of buzz about the topic of data visualization in the manufacturing industry. Much of the discussion is centered around the necessity of data and the ability to visualize that data in a meaningful way to achieve “Smart Manufacturing.”

But what is Data Visualization?

To some, it is a technology. Software that enables a person to turn copious amount of data into a graphical representation with the premise that it will improve comprehension of a point of view.

To others, it is an art form. Using artistic talent to represent data.

Yet to some, it is a process that one follows to gather and analyze data to inform decision-making.

To me, it is the “Ah Ha” moment, the confluence of all of the above. That moment when the data has been mined, the story has been built, the graphical and pictorial representation has been created, and the exercise has revealed an answer or outcome you could not reach before.

So, data visualization is key to helping manufacturers to use data and analytics in smart decision-making.  Right?

Yes, but here’s the problem: the house of cards hinges on an important concept before anyone can begin to visualize data… context. Visualizing data, or presenting it in a pictorial or graphical format, without context and analysis provides marginal value. It generally provides a one dimensional view without the complete picture.

To better utilize the art, craft and process of data visualization we need to start by setting the scene and determining what question we are trying to answer, or better yet what story we are trying to tell. If you simply ask, “which of my manufacturing lines is performing best?” there are many ways you can answer that question with different data. Is speed the best measure of performance? Cost? Accuracy? The answer is none of the above. To truly measure performance and efficiency you need to analyze all available data, with context, together.

Building context is extremely important for downstream clarity. To build effective context, you need:
·       a way to collect multi-dimensional data, not just data streams coming off individual tools or sensors.
·       a means to determine when is the right time to collect the data validation and a place to store that data and provide early visibility.
·       visibility that enables you to act upon that data and alter your approach if needed.

To think about the problem from another perspective, let’s look at the life cycle of data – from the moment it is produced to the moment it is “realized.” Where does visualization fit in? How do tools and technology assist? Where do we build context?

In the early stages data is produced. Decisions are made about what data to collect, i.e. temperature, humidity levels, power draw, up-time, down-time, etc. At the point of collection, we have the opportunity to provide additional context. What else was happening in the environment when the data was collected? Who was running the machine, what recipes of batches were being made? Now, this collected data starts its transformation into actionable business intelligence.

As we are collecting data, we have the opportunity to look out for other unique events – are we picking up fluctuations in operating parameters, what is happening with other machines running the same recipes or batches, what is going on outside the plant like an abrupt change in the weather? These unique events provide us yet another level of dimensionality to our data.

Now we store the data and begin to analyze. What trends are we seeing, what anomalies are beginning to surface? We start to prove or refine our ideas or hypotheses. We test our assumptions and validate our approach.

Do we need to make changes? If so, let’s adapt our platform or approach and incorporate those changes quickly and efficiently, with minimal impact to operations. Let’s begin the process again, executing the new plan – make the feedback, analysis, and action loop continuous.

Where does contextualization fit in? It is in every stage of this life cycle - helping ensure we have the right data. Setting up visualization enables decision makers to see analysis so they identify new patterns, drill down into charts and graphs for more detail and interactively analyze variable data. When used this way, data visualization can inform almost any decision related to the manufacturing process, from inventory management to maintenance scheduling to human resource allocation.

Together, effective data visualization, and a Smart Manufacturing technology platform can get the right data, in the right format, in the right hands, at the right time, to take the right actions.

Tuesday, March 15, 2016

What Industrial Artificial Intelligence Means for Your Operations

For years there has been a lot of buzz around intelligent computer systems and their capability to replace human-operated jobs. What does this look like in manufacturing and other industrial settings?

We are not talking about robots – like the one’s we see in Hollywood blockbusters ­– running factories and replacing humans. The fact is, the amount of data that we are processing in manufacturing is higher than ever, with faster production rates than ever – while at the same time, the total number of employees has decreased in manufacturing… this isn’t about loss of jobs. Humans simply can’t keep pace with the data. We need a system to triage the information and respond automatically (or if needed, escalate to humans) when there is a fault. Artificial Intelligence makes this possible.

Artificial Intelligence is already being used by manufacturers who deploy “Smart Manufacturing” technologies including manufacturing operations management software. Systems and solutions utilize Artificial Intelligence to allow software to perform actions that previously required human analysis and calculation. This automation of manufacturing processes can be applied at nearly every stage of the product life cycle: design, production planning, production and distribution.

Decision complexity is a common issued faced by manufacturers throughout production. A system that not only collects incoming production data, but also analyzes that data in real-time alongside context and historical data can automate decision making.  Artificial Intelligence can be used to look at manufacturing data from a holistic approach – looking for errors that might not be readily visible during the manufacturing process (i.e. general trends and possible causal/effects correlations). If the system can detect patterns and react accordingly equipment downtime, faults and wasted materials are decreased.

The bottom line? Artificial Intelligence should not be feared by manufacturers, but embraced as a tool to optimize operations and raise profitability.