Authors:

Everett Mason

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

The implementation of Internet of Things (IoT) and Artificial Intelligence (AI) techniques in manufacturing has become increasingly significant for improving yield and reducing expenses. Edge computing holds great potential to monitor machine productivity and predict performance anomalies in real time. With most manufacturing equipment being electrically powered, the electrical current consumption pattern can indicate the equipment's state and condition. However, conventional IoT and AI solutions that rely primarily on cloud computing encounter issues with bandwidth, latency, economics, reliability, and privacy (BLERP). This project utilized edge data analytics to convert current consumption signals into operation and condition information for modern manufacturing equipment. Four current transformers using the IO-Link protocol were deployed on a plasma etching machine for processing semiconductor devices. The monitoring model consists of three steps: 1) operation state, 2) operation type (recipe), and 3) anomaly detection. Pattern matching of time-series data and Machine Learning (ML) algorithms were developed and implemented on an edge computer. Electrical usage was calculated using the monitoring system. The collected data and operation history indicate a periodic pattern in the idle state due to subcomponent operation. Additionally, when the machine operates a recipe, the data suggests an increase in electrical current magnitude that aligns with operation start times. The study demonstrates the application of edge data analytics for monitoring the operation of manufacturing equipment through electrical current consumption. The implementation of edge computing of electrical current analysis provides real-time insights into equipment's operational state, type of operation, and potential anomalies. Keywords: Industrial IoT; Edge Computing; Artificial Intelligence; Real-Time Monitoring; Pattern Machining

Source:

Purdue University / 2024

Topics:

No topics listed

Co-authors:

Everett Mason