Industries Applications

Panel and Semiconductor industry

Panel and the Semiconductor industry Profet AI supports domain experts in different functions to build predictive models with the manufacturing data they have , enabling manufacture parameter optimization , quality root cause analysis , NPI parameter simulation and virtual metrology prediction.

Overview

The semiconductor industry is one of the most important industries in the 21st century.
Semiconductors are widely used in our lives, including computers, mobile phones, industrial applications, in-vehicle communications, self-driving, and electric vehicles, AIOT, 5G, etc. and the World Semiconductor Trade Statistics Organization (WSTS) estimated that the global semiconductor output value is expected to increase by about 8.4% annually in 2021.

Challanges

The semiconductor industry's high requirements for manufacture process precision cost control, and delivery time, as well as well-controlled production environment in the factory, sufficient equipment controllable conditions, and data, results in good application opportunities and business value for AI. How to improve internal quality and efficiency through AI introduction will be a key issue for Semiconductor and TFT-LCD companies.

Applications

Application - Glass cleaning, PR coating, Exposure, developing, Etching, Stripping

Virtual Measurement Metrology

Quality, efficiency, and cost are the focus of the manufacturing industry. Quality testing has detection costs, but using predictive models through AI to predict product quality can decrease this cost burden.

Assist on product cost analysis, product research and development

The semiconductor manufacturing process requires the use of a large number of polishing equipment. How enterprises use AI to build models through big data in a short time to assist product development is a new method. Using models to simulate the required physical or chemical properties can reduce the need for manufacturing numerous test samples, resulting in a shortened product development timeline.

機台異常

廠區大量設備於高頻率保養現況下，有固定的人力與維保成本，是否存在過度保養或備品庫存過多所導致人力時間與成本上的浪費，目前這議題已開始被正視。越來越多的工廠管理者，開始使用 AI 機器學習技術找尋設備內可能導致異常或故障的重要特徵值，並透過該異常診斷機制，對相關重要設備進行預防性的異常診斷作業。

Quickly adjust parameters and accelerate production capacity

The semiconductor industry has a short life cycle, but whether it is the front-end wafers or the back-end packaging and testing, whenever a new product is introduced, all engineers often spend a lot of time on parameter adjustment when new materials and new wafers come in. The key competitive advantage among various factories is how to quickly enterprises can adjust parameters efficiently to stimulate production capacity.

Al for Everyone strategy

From the discovery of valuable AI topics to the corporate proliferation of AI adoption (from a single team to corporate level activities), domain experts hope to upgrade their analytic capabilities from BI to AI in order to explore more opportunities of AI implementation in different businesses processes.