Innovative Design and Integrated Manufacturing Laboratory
創新設計與集成製造實驗室

With the development of modern industrial manufacturing and the demand of reducing product development life-cycle, reverse engineering becomes to play an important role in product design and industrial manufacturing. Reverse engineering refers to the process of obtaining the engineering design data from existing parts. Recently, 3D scanning technologies and devices have been widely applied in product redesign by scanning existing parts to acquire 3D point cloud and obtaining the 3D models by some subsequent procedures which are criterion to design new products. The process of innovation design is first scanning the real part in the world to generate a 3D point cloud set (PCD), then the preprocess of de-noising and data reduction is performed to optimize the PCD; secondly the PCD is deformed under boundary constraint with an innovative algorithm to generate some new models to help increase the speed of product design; thirdly the procedure of similarity estimation is performed to choose the suitable elements to design a new product.
The following works are in progress:

1. Extensions to K-Means Algorithm based on Random Sampling and Similarity Measure of Area Density for 3D Point Cloud Massive Data
2. A Novel Point Cloud Data Reduction and Regularity Framework for Design Reuse
3. Constraint-based Adaptive Shape Deformation Technology for Customised Product Development
4. An Investigation on 3D Shape Similarity Assessment for Design Re-usage

The rotating machinery is the one of significant equipment in modern industrial applications, such as power plants, vehicle, aircraft, and so on. Any abnormal situations of the rotating machinery may interrupt normal machine operation as well as hazard to personnel to cause enormous economic loss.

The traditional manual inspection on the rotating machinery faces two main chanlleges. One of challenges in rotating machinery diagnosis is that the existence of simultaneous-faults, that is, multiple single-faults appeared concurrently. Another challenge is a typical multi-signal fusion problem; it involves the use of multi-signal such as vibration and sound to simultaneously detect and identify faults. To solve these challenges, development of a reliable and accurate intelligent system for fault diagnosis of rotating machinery is therefore a promising research topic.

Manufacturing Execution System (MES) is recently been introduced to leverage the competitive edge of manufacturing enterprises where the shop-floors are under dynamic and mixed-product assembly environment. However, current MES practices for wireless manufacturing still suffer from two kinds of difficulties: (1) inefficient information acquisition technique; (2) lack of reliable and fast signal-positioning conversion method for real-time monitoring of manufacturing objects. Traditional indoor positioning algorithms cannot process massive signal data and predicate the positions of manufacturing objects in accurate and efficient manner. This project handled the first challenge by adopting the RFID technology which could constantly capture the dynamic wireless signals sent from the tags mounted on manufacturing objects. The second difficulty could be solved by applying online sequential extreme learning machine (OS-ELM), which inherits the elegant properties of ELM (extremely fast learning speed and high generalization performance) and could avoid retraining for new arrived objects and disturbance existed in dynamic shop-floor environment. With the RFID enabled framework based on OS-ELM, the proposed method upgrades the manufacturing objects as smart manufacturing objects (SMOs) for real-time signal processing and intelligently tracing of MES. The experimental results verify that the proposed RFID enabled indoor positioning method based on OS-ELM outperforms than other prevailing approaches in terms of accuracy, efficiency and robustness.