Yield Forecast of Great Size Polarizer under Learning Effect

碩士 === 國立成功大學 === 工業與資訊管理學系專班 === 96 ===   Due to the TFT-LCD product price drops continuously, many companies think any ways to reduce the production and the stock cost all. One of the Components Polarizer of manufacturing industry is belong to industry of high material cost, it is very important t...

Full description

Bibliographic Details
Main Authors: Chien-Cheng Chen, 陳建誠
Other Authors: Der-Chang Li
Format: Others
Language:zh-TW
Online Access:http://ndltd.ncl.edu.tw/handle/21178113994582228384
Description
Summary:碩士 === 國立成功大學 === 工業與資訊管理學系專班 === 96 ===   Due to the TFT-LCD product price drops continuously, many companies think any ways to reduce the production and the stock cost all. One of the Components Polarizer of manufacturing industry is belong to industry of high material cost, it is very important that how to reduce raw material input-cost. At present, Polarizer manufactory still use the way of ordinary-firm-yield in the production quantity plan. Because various systems of the yield change can affect the entire production quantity plan rationality. Therefore, we must integrate the yield with the consideration. It can respond the actual production condition and the production cost, and it is expected that in the consideration study effect situation, by the past production experience of the small-size, forecast change to the future that great-size of yield. Causes the company resources to make the more perfect arrangement and exploitation, and then reduces the original material investment-cost and the stock-cost.   Along with TFT-LCD technology and contrast promotion, great-size monitor demand also growth, however, great-size material cost relative high. If you has not been able to consider the yield change in the production plan, but invests excessively many original materials, which very easy to create the material and the stock cost increase. Therefore, it even more demonstrates importance of the construction production study mode. For this reason, this research use production information for the past year, and permeate the way of Support Vector Regression (SVR), which constructs production mode of the small-size and the great-size. By the way of Leave-one-out Cross-Validation to confirm the forecast result whether it can be accepted in the scope or not. Expected in the study effect situation, it can reduce so many nonessential inputs that decrease disbursement of the original material cost. It develops that the yield forecast mode through research and analysis, In the future, the tool which might provide referral about the polarizer manufacturer when to produce great-size product.   For this reason, this research through confirmation proposed four kind of different production modes. According to the different production situation and the background, and suggested that apply separately to different production mode. It enables the industry control to produce yield change, and achieved the goal which reduces effectively the production cost.