Researchers at UCI have developed a non-intrusive method for building a virtual replica of manufacturing machine, which allows for accurate diagnostics of the state of the system. This provides manufacturers with real-time information on quality control and immediately identifies any malfunctions in the system.
Digital twins are virtual replicas of physical systems that are extensively used to model machines and products, and provide information on machine operation, design, and diagnostics. This allows manufacturers to maintain quality control of the resulting products and receive early indicators of device malfunction. Typically, digital twins rely on a large network of built-in, physically interconnected sensors to provide data: each of these sensors monitors one parameter of the system, such as vibrations or energy consumption. However, many existing machines do not have such sensor networks built-in by default, making the creation of a digital twin for such systems time-consuming, expensive, and sometimes altogether impossible.
To this end, researchers at UCI have developed a living digital twin which instead utilizes various external, wireless Internet of Things (IoT) sensors (acoustic, vibration, magnetic, power). The signals collected from these sensors are fed into a novel clustering algorithm which builds an accurate virtual representation of the physical state of the system. Unlike current state-of-the-art, this digital twin is “alive,” able to update itself and identify faults in manufacturing systems, for example, additive manufacturing systems, in real-time.
For real-time quality interference and fault localization in manufacturing systems and machines
· Non-intrusive: Unlike standard state-of-the-art, which require an extensive network of sensors to provide data for the digital twin model, this methodology relies on informative and non-invasive IoT sensors which are external to the device.
· Accurate: This methodology provides >83% accuracy in localizing malfunction.
· Living: The IoT network is “alive”, able to adapt and learn from its environment in real-time.
The invention is currently in the proof-of-concept stage.