Looking Into the “Future”
In the context of manufacturing, looking into the future or in simpler terms forecasting is an art entirely driven by the experience of the individual.
In this blog, I will talk about how right forecasting tool powered by advanced ML and AI help manufacturing organisations tackle the issue of aging workforce.
Let us consider a fictitious company PSM global Inc. into refinery business for this blog.
Consider a relatively less experienced Line Supervisor running a continuous production line ( say a refinery ). There is some line carrying the fluid and the a pump is used to generate the required pressure. The pump is designed to operate in a particular pattern ( see the image below ). As can be seen from the image, there are some cyclic patterns and also the cycles are trending upwards steadily.
If the Line Supervisor now need to understand what would be the pressure look like, if he/she continues to operate the pipe line for next 2–3 days. This is where his/her experience and intuition plays a major role. The supervisor can make a scientific guess based on the past experience of running the line him/herself or will reach out to an experience line supervisor / manager to seek recommendations. Based on the recommendation then he/she can decide in next steps.
Now fast forward to the age of Industry 4.0 and Machine Learning. With the past pressure data already archived in the database of the Industrial IoT platforms, the supervisor can run predictive models ( read forecasting models) and get a view of how the pattern will look like in next 2–3 days as shown below.
As seen from the image below, the prediction model has predicted the next 2–3 days of pressure profile looking at the past data.
A zoomed version of the forecast is shown below:
with this information in hand and more importantly without any involvement of the senior line supervisor the supervisor can now plan the activities accordingly.
All these happened like magic and the line supervisor seamless able to look days into the future without any help from the senior line supervisor still there is a lot which has gone in the background for making this look so simple. In future blogs, I will talk about the role of data collection and storage, quality of data, algorithms and most importantly the role “senior line supervisor” had to played in order make this look like ridiculously simple.
Note : The model is built using synthetic data simulating the pressure like a real world pump would have produced
[ The views expressed in this blog is author’s own view and it does not necessarily reflects the views of his employer, A&M]