The Three Questions to ask when shopping for Industry 4.0 Solutions

Predictive vs. Proactive Control Of The Plant Floor

The process of deciding which ai solution is best for your company is becoming

increasingly more difficult as the number of new tech companies offering new solutions

increases. Nevertheless, the sooner you decide, the better. Typically, the biggest hurdle to

overcome is the decision to get started! There are different approaches to applying ai to

the day-to-day production operation, so it is important to understand the options before

deciding. One of the fundamental criteria when evaluating Industry 4.x solutions is the

difference between predictive analytics and proactive analytics.

The primary focus of many ai applications is currently predictive maintenance, or the

ability to predict, based on a complex array of variable, when a machine is likely to fail.

The ability to predict accurately enables maintenance organizations to schedule required

maintenance in advance of machines failing during scheduled production hours. With

accurate predictions, maintenance activity can be scheduled during non-production hours

dramatically improving machine uptime and production throughput. The challenge

however is including enough inputs for the mathematical probability of failure to be

trustworthy. For example, if feed-rate, speed, and time are the inputs available for

calculating probabilities when in reality temperature and pressure are the critical

characteristics, accurately predicting when a failure is likely to occur becomes highly

suspect at best. Replacing a $30,000 motor that hasn’t failed yet based on ai that is

probability-based is a highly stressful decision! With the worst part being no way to

verify whether or not the prediction is accurate and true. Proactive analytics on the other

hand takes an entirely different approach and yields a more certain outcome.

Proactive analytics use a fact-based, data-driven approach to predict machine failures.

Available data gathered from sensors mounted to machines or directly from the PLC

(programmable logic controller) running the machine is used by both approaches. The

difference being one approach uses a highly complex statistical engine to calculate the

probability of failure, while the proactive approach monitors a real-time stream of data

from each of the inputs individually. Each individual machine parameter can be

compared to its design target for optimal performance with ai-driven algorithms that

determines when it’s time to schedule a proactive maintenance activity and notifies

appropriate maintenance personnel. Knowing the difference between these two

approaches will equip buyers and decision-makers to challenge assertions made by sellers

about their predictive maintenance and machine-learning capabilities, and depending on

the answers they receive, be in a much better position to compare solution providers

capabilities leading to a much more intelligent defensible decision once a selection is

made. When implementing any new technology, always be sure to include change-

management as an essential requirement for each deployment.

You can find an example of a proactive maintenance solution at www.trumbleinc.com or

get answers to your questions by emailing [email protected] to explore these

topics further.

Jeffrey Trumble is the creator of REVEAL, a patented enterprise software solution

developed by Trumble Inc. offering proactive control of product quality, and machine

performance, through a single application.