Fresh New AI Tool That Can Monitor CNC Machine Tools

NUMai software helps CNC machine tool users predict any deviation from normal machine operating behaviors. The graph shows the prediction of a variable (blue), the measured value (red) and the residual error (yellow). (Image courtesy of num.com.)

Artificial intelligence (AI) is, perhaps without much disagreement, one of the most disruptive technologies that has progressed across many disciplines. NUM, the 50-year-old developer company of computerized numerical control (CNC) systems, has just launched NUMai, a state-of-the art AI software package that provides CNC machine tool users with highly cost-effective condition monitoring capabilities.

NUMai is fully compatible with NUM’s biggest success to date, the Flexium+ CNC system. This means that no additional devices or sensors are required since it is installed and run on the same industrial PC used for the CNC system’s human-machine interface (HMI) that is already in place. Therefore, the software can be implemented once the machine tool has been commissioned and is ready to start the production process, or it can be run on a machine that is already being used for production purposes.

Put simply, the technology behind NUMai is a neural network that learns how a machine behaves under optimum standard conditions. The neural network learning process starts when operational data is collected from a machine that is used for normal production tasks over a certain period to ensure that the data are as exhaustive and accurate as possible. When implemented, the NUMai software application will continuously emulate the machine’s normal behavior. If any deviation from this behavior occurs, the software will help identify the cause behind it.

For those interested in stats, NUM beta tested the technology on CNC milling machine geared with three axes and a spindle. This required a neural network made up of 36 neurons with three hidden layers. The values of 396 parameters had to be precisely defined, which required acquiring more than 2 million “known good” data points and 300 iterations of the teaching phase. In all, the process took approximately four hours per axis.   

During a machine’s lifetime, failures will lead to machine downtimes that can last anywhere from hours to several days. These interruptions can cause enormous disruptions that can have a great impact on service quality and costs, with consequences ranging from delivery delays to penalties levied against companies.

NUMai’s solution provides powerful monitoring, fault prediction and diagnostic capabilities to help maintain production standards and minimize downtime. AI is key for machine derating diagnosis and preventing mechanical or electrical problems, such as aging, backlash, motor efficiency reduction and missing phase. Therefore, having the NUMai software solution running in the background of the industrial PC will detect discrepancies or anomalies outside of user-defined parameters and communicate them to the PLC so that it can take appropriate action. This can range from a simple warning message to an emergency disengagement.

Wasting money and time due to unforseen machine downtime can be prevented by diagnostic and fault prevention functions using AI. This graph shows how the increasing residual error of an estimated variable is affected by a fault condition. This results in the system sending a warning. (Image courtesy of num.com.)

As mentioned above, the NUMai conditioning monitoring software is compatible with any NUM Flexium software version 4.1.10.10 or higher. To find out if the AI-based NUMai option meets your manufacturing needs, check the current Flexium+ CNC System catalog on NUM’s website.