Sound Monitoring and Applications for Smart Manufacturing Using MTConnect Framework and Artificial Intelligence

Abstract

With the new era of Industrial Revolution 4.0, adopting human skills and knowledge is getting more valuable to smart manufacturing and monitoring. Moreover, cognitive ability that comes from human senses and intuition is able to facilitate smart sensing and collaboration between humans and machinery. Smart manufacturing is the collaborative outcome of information technology, industrial manufacturing technology, and human intelligence and creativity, leading to a rapid evolution of the manufacturing system. Adopting human skills and expertise is getting imperative to expedite the revolution from conventional to smart manufacturing regime. Auditory sensing is one of the basic sensory information from humans. Ambient sound is prevalent on every shop floor and in every facility area. Although novice operators have basic and minimum knowledge without long-term training or experience with a machine, they can detect the basic information such as running state and productivity from the emitted sound. Experts in manufacturing processes or systems can diagnose the health by auditory sensing. In the machining field, for instance, well-experienced machinists can detect tool conditions as well as cutting quality by listening to sound under various cutting conditions and in an environment where a coolant is flooding in which visual inspection is impossible. If these human skills can be digitized and utilized to train artificial intelligence (AI) using standard communication methods, the impact on the manufacturing industry will be significant, especially in small and medium-sized companies. In this presentation, sound monitoring and applications for smart manufacturing are proposed based on artificial intelligence created by human knowledge. To make a seamless sound stream with machine information, MTConnect, which is an opensource royalty-free manufacturing information standard for machine tools. To accommodate the high sampling frequency of the sound sensors, the DiscreteTimeseries format of MTConnect was utilized. Not only digital twin for state-of-art computer numerical control (CNC) machine monitoring applications but also various sound monitoring cases for legacy manufacturing machines in real shop floors will be presented. As one of the CNC machine applications, a digital twin for 3D visualization with cutting sound monitoring by Three.js library and MTConnect was created. On the other hand, for legacy manufacturing machines such as pressing, plasma cutting, and custom-hydroforming machines, productivity and runtime predictions by Grafana interface and MTConnect were applied.