DataProphet is a Cape Town, Western Cape based company that provides consulting and product development services to a wide range of industries. As machine learning specialists they are comprised of a diverse team of skilled computer scientists, statisticians, actuaries, engineers and mathematicians that are poised to deliver something which has previously been unavailable to entities – actionable artificial intelligence solutions.
Atlantis Foundries have partnered with DataProphet in a pilot project using Artificial Intelligence that aims to predict sub-surface defects currently detected only after machining, and to identify the optimum process parameters to prevent the sub-surface defects occurring in the first place.
Dr. Michael D. Grant, the Chief Technical Officer at DataProphet, who has been very instrumental in implementing the Artificial Intelligence pilot project at Atlantis Foundries explains how your foundry can benefit from a similar project and move you towards becoming a Smart Foundry.
“The casting process is so phenomenally complex: A cascade of parallel processes, each with their own control responses, set points, and transients. Defects are also only a symptom of the combination of everything in these complex cascade processes, where the unique geometry of the mould also influences the location and type of defect. Foundries globally have an elegant solution to this problem: The great foundryman. This mythical beast can look at a defect and, based on some intrinsic feel, make a small corrective action, which is able to correct the problem… some of the time.”
“Great foundrymen are a rare breed (although there is no shortage at Atlantis), and they are quite literally fashioned over decades in the many foundry operations around the world. The intrinsic feel for how to correct a problem is the expression of their experience. And as we struggle to retain or transfer these skills, the expertise becomes confined to textbooks.”
“Modern foundry operations are generally well connected, with the PLC systems operating machinery recording many process variables. These data systems are usually tied together, with the measurement results often stored on some server (somewhere!). Some modern efforts are now using many small transducers on discrete measurement points to gather plant-wide data. This is referred to as the Internet of Things (IoT) and this distributed architecture has resulted in a proliferation of measurements within industrial plants. However, such IoT devices are not required, and many PLC systems are more than sufficient to gather the required data.”
“However, the servers on which this operational data resides are expensive to maintain and most of the time there is no holistic strategy for the retention and exploration of process data. It’s not a trivial problem, and many folk would rather just hang onto the data in the hope that some future project will unlock the supposed hidden wealth contained therein. This is the perfect recipe for Artificial Intelligence (AI)!”
“Coded into your data is the perfect set of conditions – all the combinations of your plant and how they lead to defects – but discovering these conditions from the process archives is complex and messy. Our AI is able to learn all the nuances of your system and then – like the experienced foundryman – guide an operator to make the smallest set of changes to eliminate that defect. The AI is also able to determine which change to make first, and second, and so forth. These results are communicated to the plant operations through the generation of a control plan, weekly process reports and a predictive model to enable engineers to test new parameter combinations.”
“The modern foundry remains the highly complex process it has always been, except now – inline with Industry 4.0 objectives – all measurements are connected and automatically recorded, process parameters are recorded for every component produced, and the operator becomes the master of this domain using the results of the AI to operate the plant at maximum efficiency. Further improvements are achieved through deeper integration of robotic process automation and the heavy lifting becomes automatic which helps the production line run smoother.”
DataProphet secures Knife Capital funding
DataProphet has announced that it has now received a multi-million dollar funding round from Knife Capital – a leading venture capital firm with offices in Cape Town and London. Knife Capital invests via a consortium of funding partnerships, including SARS section 12J Venture Capital Company KNF Ventures and Draper-Gain Investments.
The funding will be applied to boost the company’s innovation capabilities and accelerate global expansion. DataProphet has already successfully delivered on many local and international projects. Key clients and revenue drivers are largely international and the company has earned a reputation of competing at the highest level, outperforming blue-chip corporates on project deliverables in many instances.
For further details contact DataProphet on TEL: 021 300 3555 or visit www.dataprophet.com