Smart manufacturing in machining process using industrial edge device

In modern manufacturing, smart machining solutions leverage advanced data analytics, real-time monitoring, and adaptive control strategies to optimize production efficiency, minimize costs, and ensure the quality of the product with remarkable precision and reliability. This thesis, titled "Smart Manufacturing in Machining Process Using Industrial Edge Device," explores innovative approaches for integrating industrial Edge devices into machining operations, enabling real-time monitoring, predictive maintenance, and cost-effective process optimization. The research is structured into four main items, each addressing a critical aspect of smart manufacturing in machining processes, specifically focusing on tool wear prediction, sensorless cutting forces and torque measurement, tool chipping detection, and runout monitoring. First, an AI-assisted digital shadow was developed to predict tool flank wear using data acquired from an industrial Edge device in a drilling operation. The study combines high-frequency data acquisition from an industrial Edge device and a rotary dynamometer, followed by extensive feature engineering. A recurrent neural network (RNN) model utilizing bidirectional long short-term memory (Bi-LSTM) and bidirectional gated recurrent unit (Bi-GRU) architectures is employed to analyze tool wear regions. The developed digital shadow enhances cost efficiency by reducing the dependency on expensive multi-sensor systems and mitigating unnecessary tool replacements, aligning with the principles of smart manufacturing. Second, a novel sensorless method is introduced for real-time measurement of cutting forces and torque in the milling process. Accurate measurement of cutting forces and torque is critical for process monitoring, tool condition assessment, and optimization in machining operations. However, conventional dynamometers, while precise, are costly and introduce complexities in experimental setups. This study leverages an industrial Edge device to extract spindle current and torque data in real time. Experimental validation during the milling of Titanium alloy Ti6Al4V demonstrates strong correlations between dynamometer and Edge device cutting data, achieving mean errors of less than 12%. This sensorless approach significantly reduces setup time and cost while ensuring precise force monitoring, making it a viable alternative to traditional measurement techniques in smart machining environments. Third, tool condition monitoring was investigated with a focus on detecting cutting-edge chipping in the milling of titanium alloys. As a critical factor, tool chipping significantly impacts machining efficiency and tool life. The study analyzes spindle current and torque signals obtained from an industrial Edge device operating at a high-frequency sampling rate of 500 Hz in the time and frequency domains. Fast Fourier Transform (FFT) analysis reveals significant spectral differences in Edge device signals before and after tool chipping occurrences. These findings establish industrial Edge computing as a reliable tool for tool condition monitoring, minimizing the need for expensive multi-sensor systems and enabling early detection of chipping events to prevent catastrophic tool failure. Finally, an innovative sensorless approach for in-process runout detection was proposed. Runout, a key factor influencing machining accuracy, surface finish, and tool wear, is traditionally assessed using contact-based sensors. This study demonstrates that spindle current and torque signals, captured through an industrial Edge device, are highly correlated with dynamometer measurements, enabling real-time frequency-domain analysis for runout detection. The proposed method is experimentally validated during milling of Ti6Al4V, presenting a cost-effective and efficient approach to improve machining precision and stability in smart manufacturing environments. In summary, this thesis contributes to the advancement of smart manufacturing in machining processes by harnessing the potential of industrial Edge devices for real-time monitoring, predictive maintenance, and sensorless process optimization. The proposed methodologies reduce reliance on expensive sensors, improve machining efficiency, and enhance cost-effectiveness, thereby paving the way for the next generation of intelligent and autonomous manufacturing systems.