AI can optimise manufacturing processes; awareness needed

8th June 2018

By: Jessica Oosthuizen

Creamer Media Reporter

     

Font size: - +

Artificial intelligence (AI), data and automation tools can aid manufacturers in enhancing and optimising their processes, says Cape Town-based machine learning start-up DataProphet cofounder and MD Frans Cronje.

The company specialises in developing AI solutions for the manufacturing industry, and was the overall winner of the inaugural Mercedes-Benz South Africa Innovation Challenge in 2017.

Cronje says AI has not yet been implemented in South Africa, but that use and implementation is developing. A major challenge is understanding AI, as many companies are still not sure what it is and are also sceptical to use it because of this uncertainty, he adds.

Cronje notes that there is still much to be done in terms of educating the industry about AI and its value. “There is, to a certain extent, a particular mindset around how data has been analysed traditionally and that needs to change – by informing industry.”

DataProphet’s AI solution is used for dynamic process control. It identifies and prevents defects from occurring. It can provide real-time insights from massive amounts of data, which is extremely valuable, Cronje states. The company’s AI solution, OMNI, uses advanced predictive and prescriptive machine learning capabilities as well as high-quality computer vision to predict defects, faults and quality errors and prescribe the ideal process variables to shift processes to higher yields.

He comments that AI has the potential to facilitate skills transfer by learning from experience contained in historic data, for example, an experienced quality control engineer might make edits to processes and the effect on defects or faults that result from the edits would reflect in the data. AI would learn from this experience and go on to share it, when relevant, with the next control engineer.

“If the engineer had uncovered an improvement in yield, the AI will automatically see it in the data, learn from it and retain the learnings. “What makes this exciting is that AI doesn’t rely on a person’s memory or experience; instead, it becomes an accurate repository of all process data and insights, including the learnings of the most experienced humans using the system, which can be used to augment new staff members’ skills and knowledge.”

He enthuses that AI should be seen as a tool for taking skills, knowledge and experience, and sharing it among new workers entering the world of employment.

However, there is also the danger of over-automating and the value of human capabilities must not be overlooked, Cronje states. He explains that AI observes what it narrowly can observe through a camera or sensor, while humans can observe more broadly and generally. “Contrast the field of view that a camera has compared with that of a person,” he comments.

“You cannot write a programme and instruct it to detect defects if it has no way of ever seeing the defect in the data collected.” He notes that, fortunately, most sites the company has seen are already collecting sufficient data as part of established controlled processes – it is only a matter of bringing the data into one place for AI to learn from it.

Further, DataProphet tries to drive awareness and information regarding AI by writing and sharing content on the topic. Company representatives also attend and speak at events. Cronje will address attendees of the technical workshop on the first day of the Manufacturing Indaba, which will take place on June 19 and 20 at the Sandton Convention Centre, in Johannesburg.

The company will also showcase its OMNI AI solution at the exhibition and will be located at stand C17.

Edited by Zandile Mavuso
Creamer Media Senior Deputy Editor: Features

Comments

The content you are trying to access is only available to subscribers.

If you are already a subscriber, you can Login Here.

If you are not a subscriber, you can subscribe now, by selecting one of the below options.

For more information or assistance, please contact us at subscriptions@creamermedia.co.za.

Option 1 (equivalent of R125 a month):

Receive a weekly copy of Creamer Media's Engineering News & Mining Weekly magazine
(print copy for those in South Africa and e-magazine for those outside of South Africa)
Receive daily email newsletters
Access to full search results
Access archive of magazine back copies
Access to Projects in Progress
Access to ONE Research Report of your choice in PDF format

Option 2 (equivalent of R375 a month):

All benefits from Option 1
PLUS
Access to Creamer Media's Research Channel Africa for ALL Research Reports, in PDF format, on various industrial and mining sectors including Electricity; Water; Energy Transition; Hydrogen; Roads, Rail and Ports; Coal; Gold; Platinum; Battery Metals; etc.

Already a subscriber?

Forgotten your password?

MAGAZINE & ONLINE

SUBSCRIBE

RESEARCH CHANNEL AFRICA

SUBSCRIBE

CORPORATE PACKAGES

CLICK FOR A QUOTATION