A leading South African diamond-mining consortium is deploying Artificial Intelligence-based software tools for diagnosing the root cause of failures in diamond sorting machines used around the world. Using a technique called model-based reasoning two software programs called ICAL and ICAT, enabled the diamond sorting machine maker to drive expert knowledge to field engineering staff in remote locations where the machines are used to solve and repair problems in the field.
According to a key project engineer, "Our decision to follow this route was based on the remote location of the equipment, limited skills levels in those areas and our push to increase maintainability of our products." He added, "It is like having the design engineers and the field technicians side by side at all times through the troubleshooting procedure, fixing the fault, learning from failure and reporting back to the design engineers, all integrated within the workflow."
The project
The project saw its first light in 2000. The idea was to make the design engineers' knowledge of the diamond sorting machines available to field technicians at the mines. Using a different expert system, the company started phase 1 of the project. According to one of the development engineers, "Gaps were discovered in phase 1 of the project which we could rectify with Automated Reasoning's ICAL program."
In 2003 ICAL and ICAT software was adopted for the second phase of development. The ICAL development software enabled the development team to construct a software model of the sorting machine, link subsystems and add technical documentation. He went on to say, "Using ICAL we were able to develop our own customised Web front-end with the availability of an Automated Reasoning supplied API, then install the ICAT diagnostic program onboard the machine itself. Additionally, ICAL uses a model-based approach with the added benefit of using existing cases and rules, which allowed us to reduce modelling time by more than 50%, so the cost of deployment is markedly reduced."
The final system monitors the machine's sub-systems in realtime and takes action in the event of malfunctions. Information is gathered from the machines and from various test and measurement instruments to measure heat, vibration, voltage and pressure. All this event and alarm information is incorporated into the ICAL model for decision making and troubleshooting optimisation by the ICAT AI engine.
The final system is Web-based, as are the diamond sorting machines, so this critical knowledge can be accessed and sent back to engineering for reporting and learning purposes regardless of the location of the machines.
Self-learning
The cause of failure from each event can be learnt in order to update the models, and to optimise the operating efficiency of the ICAT program. The optimised runtime can then be redeployed over the intranet or Internet to distribute that knowledge back to the machines and the field engineers where it belongs. Fault finding increases, and the knowledge gained can be shared globally; with a keystroke - and because ICAL and ICAT software are model-based reasoning programs, they can be used for almost any equipment, system or process. So they are surprisingly adaptable to a wide variety of applications. The programs also support multiple deployment options to adapt to a broad range of end use requirements.
For more information contact Quintin Els, Automated Reasoning, 011 453 3992, www.automated-reasoning.com
© Technews Publishing (Pty) Ltd | All Rights Reserved