Seamless edge-server collaboration for real-time digital twin in machining process

Abstract

With the rise of smart manufacturing systems aimed at creating efficient and cost-effective environments for mass production, various process data acquisition techniques and data models have been developed for CNC manufacturing to enable the creation of Digital Twin (DT) models. This article proposes an edge-server collaborative data architecture collecting crucial CNC machine data, including servo motor parameters, cutting tool information, and sensor data such as vibration, pressure, and temperature. The architecture is designed to process real-Time data during mass production with minimal latency, considering the high-speed nature of machining processes, while also storing historical data in a Data Lake for the development of AI models. An infrastructure for Digital Twin of the brake disc machining process is created, a particularly challenging task due to the complex geometry of the disc and the unique material characteristics of cast iron. During machining, sudden tool failures or even brake disc breakages can occur due to the heterogeneous nature of cast iron. Given that brake discs are critical safety components in automobiles, monitoring process data linked to the cast iron and cutting tool supply chain during mass production is essential. Digital Shadow serves as a foundation for real-Time anomaly detection, predictive maintenance, and tool wear prediction models. This paper also proposes several deterministic modeling approaches for real-Time anomaly detection, predictive maintenance, and remaining useful life predictions for cutting tools. These models leverage machine data such as spindle and feed-drive motor currents, load, and positional errors during brake disc machining, in combination with sensor data, including temperature, pressure, electrical power, and vibration, to enhance the monitoring and optimization of the machining process.