Request a Quote

Please note - we are closed for summer holidays and will reopen on Tuesday August 3

Message / Order details:


Keep up to date with us on the latest industry news as well as what's going on at True Gear & Spline Ltd. We also post articles for insider tips and tricks, so make sure to check back frequently.

The Challenges of Introducing AI and ML in Gear Manufacturing

July 27, 2018

Artificial intelligence concepts are at a relatively nascent stage in the gear manufacturing industry. AI is reliant on machine learning and for machine learning to be effective it relies heavily on data input. The more data that can be crammed into the program, the more accurate the insights that can be gleaned. Gear manufacturing is not something one would consider a tremendous source of data generation.


Gear manufacturing generates enormous amounts of data, which is ripe for implementing machine learning and artificial intelligence concepts.


In fact manufacturing industries are veritable treasure chests of data. With each project involving hundreds, if not thousands, of minute calculations, adjustments and incremental actions, the milling, lathing, shaping and sanding (speaking broadly) are all creating a wealth of information about the manufacturing process.


Expertise in gear manufacturing, in particular, is an asset clients look for. Artificial intelligence offers a great opportunity to deconstruct what is meant by this expertise and how it manifests itself in the manufacturing process. We list out some general conceptions that are likely to challenge artificial intelligence implementation and shape its use in the industry.


Ready source of data – Much of the gear manufacturing done today is done using computer numerical control. Virtually all machine tools are CNC controlled and even in everyday cutting programming can patterns be determined. Whether this data can be captured by AI tools will depend on the CAD/CAM software providers.


Sensing minuteness – But the real challenge will be getting up close with the production process. How a cutting and milling is performing is something that cannot be captured at present. Whether optical, weight-based or other, sensors must capture cutting performance much more finely to offer insights on production efficiency.


Production planning – The first port of call for artificial intelligence implementation will most likely be assisting in production planning. From allocating production time to ordering replacement parts, AI will be ideally poised to contribute, at last initially, to bring about macro level efficiencies.


An education – Shop floors are a great place for insight – by the human; much less so for gathering and inputting data. Machine learning algorithms are heavily reliant on the accuracy and frequency of data. Such vigilance towards data will require a learning curve. Engineers and technicians will have to be taught how to accurately collect data.