- Factory 5.0 — AI boosted manufacturing
Just now·5 min read
Digital change (dx) is making an impact on our life and work. Behind this trend, a tremendous amount of data are generated. This phenomenon created the need for data-related specialization. From the Data Engineers to Machine Learning Experts. I want to show the story how to advance from 3.0 to 4.0 and how to use the potential of well done 4.0 company to be even 5.0 (My vision of the 5.0 company)
Photo by Markus Spiske on Unsplash
For the needs of the story, I have created a fictitious company — SuperX. Right now SuperX is well prospering 3.0 factory but the CEO is inspiring to be 4.0 or even higher.
During the technological advance from manufacturing 3.0 — to 4.0 the company accumulate many IoT devices, network and server infrastructure is starting to grow. The main problem is that IoT device is used only for special-purpose at the desired location place (usually close to the production unit). From the engineering point of view, this advance is brilliant. Now SuperX controls the crucial parameters, the yield is going up and manual processing is substituting by robots. Everyone is happy. Production capacity is increasing and it seems that no one will stop the growth. Everything went well until …. Boooom!
Photo by Jens Johnsson on Unsplash
The process now has dozen of parameters to control. Each IoT device has its own interface and working well only for one purpose. At a small scale, everything seems fine but in the big picture the problems start to be unsolvable, and testing any hypothesis takes ages.
Now the company is starting to understand the role of data. Unfortunately, it is a little late.
SuperX heard about something called databases. SQL/NoSQL or even Hadoop. Let`s load everything there and we will be a data-driven company! asked CEO Unfortunately SuperX data environment looks like a swamp!
Photo by Anastasia Zhenina on Unsplash
You will identify the problem when the first query appears. Let me show you an example of the date standards
Sensor 1 — Epoch Time (count seconds from 1 January 1970)
Sensor 2 — DD/MM/YYYY
Sensor 3 — MM/DD/YYYY
Sensor 4 — What is a timestamp?
Sensor 5 — SuperX ordered a device from china all data labels are in Chinese how to write the column name in Chinese? (pro tip — Unicode Escape )
There are many more examples but basically, data is anyhow standardized.
Photo by Denys Nevozhai on Unsplash
What SuperX should do?
Option A — Stay at the current state of the art. No one understands the production but we like to be firefighters making ASAP’s ASAP.
Option B — Let`s be true 4.0 Factory and bring the chill-out.
If you choose option “A” the best part of the article will be not interesting for you. Otherwise, go ahead and check how to go forward and make it correctly.
Photo by Aaron Burden on Unsplash
The milk was spilled, SuperX needs the data expert.
The first step is to create the true data lake and use its potential for the decision-making process. The Data Lake development needs to be planned instead of “Let`s be a Data Lake”. The power of data standardization and well-done ETL processes will lead the company into a new era, the era of Manufacturing Intelligence.
In the data world, people usually know what is BI — Business Intelligence.
In brief, BI is the process of data transformation to get the knowledge to make companies more competitive. The most important aspect is the “tase of data”.
At many branches of business eg e-commerce, loans, insurances. All this data is somehow related to humans. The churn rate, credit or insurance scoring, etc.
Photo by Ulises Baga on Unsplash
In my case, I have no idea how to make credit or insurance scoring models. It is something called Business Domain. For the machine-based data, the BI equivalent is MI. Manufacturing intelligence is also the process of data transformation to get knowledge for manufacturing processes optimization.
Photo by Ameer Basheer on Unsplash
The Indicators of 4.0 manufacturing
- IoT Data is well prepared in the data lake,
Data is curated by Data Curators,
Executives are making the decision processes basing on data,
Citizen Data Scientists are showing up,
Tools for Big Data analysis are easy to use,
Company is optimizing the workforce RPA (robotic process automation) robots are created,
Processes are highly automated,
Fewer operators are needed,
Right now factory 4.0 is the goal of many companies. It looks like the final form of technological advance. But at well done 4.0 company there is the field of opportunity.
As a true Science Fiction fan, I like to make kind of “Reverse Engineering” of unrealistic solutions. I am often debating (mostly with myself) what has to be done to make it real?
As Senior Data Scientist at LG Energy Solution (Factory 4.0 producing the Li-ion Batteries), I have asked the question myself. Why did we need to analyze the data by ourselves?
I would like to share the example of incorporating AI into the manufacturing environment. Due to confidentiality, I am not able to dive into detail, but the project outcome — Using the AI model the root cause of the problem was found and eliminated.
Lean Six Sigma Black Belt project achieved >$28M / year
The list of the ideas:
- Predict the defect before they appear
Virtual DOE — real tests no longer needed
Score product safety — New Level of Quality
Automatic abnormality detection
Automated Production Planning
Thank you for reaching this point. If you like this story please let me know I will consider writing more often.
Hire me to your project
After my work in LG Energy Solution. I can offer you up to 10h/weekly as a freelance consultancy. If you want to hire me as a consultant just catch me on LinkedIn.
Data Science and Industry 4.0
Let`s date with data
- Date of publication:
- Tue, 02/23/2021 - 13:26
Click on the link - it will be copied to clipboard