Masters' Theses A Study on the Implementation of the Speech Recognition System for the Real-Time Decision Support of Manufacturing Cells
2015.07.29 12:30
| year | 1999 |
|---|---|
| author | Eoksu Sim |
| Keyword | manufacturing cell management, speech recognition, real-time data collection, decision support |
| Abstract | In recent competitive business environment, it is by the fast and exact data that the company can take the advantage over the others. The correctness of the input data is important for manufacturing automation and enterprise resource management. However, there has not been enough study on the management of manufacturing cells. It is necessary that without time delay the manufacturing cell status data can be collected and the worker´s information retrieval can be executed. Study on the above issues should be carried out. There have been many problems such as the time delay and static information input in the conventional manufacturing cell data input method. Moreover, it was not possible to retrieve the information in real time by the worker. This research proposed and implemented the system through the integration of the manufacturing cell management and the speech recognition technology. Database was built for the shop floor data and Dragon Systems´s speech recognition engine was used. Workspace was built with the worker who can use the headset wireless microphone and can operate with Turning Center. This research, also, conducted an experiment to show system´s applicability in noisy manufacturing cell by measuring the recognition ratio of the registered speech commands. Using the headset microphone and limited speech commands minimized the effect of the noise on the recognition ratio. And the recognition ratio was measured at 90% on the average. This thesis implemented the proposed system that enables the fast and exact data input and real-time information retrieval. By the experiment, it is proved that the integrated system can be applied to the manufacturing environment. |
| c | MS |
Download :