| Summary: | 碩士 === 朝陽科技大學 === 工業工程與管理系碩士班 === 93 === Dry film photoresist is an important raw material for transforming precise circuit layout design into producing a high quality printed circuit board. The coating operation of the dry film process plays a key role in producing high quality products. Due to soft property of the dry film photoresist, it is concave by using contact instrumentation and may result in measurement errors. Thus, how to control the thickness of photoresist in the coating operation is very important.
To solve the process control problem of the coating operation in the dry film process, first, the thickness image of the dry film is captured by a high resolution CCD, and computer vision techniques are applied to thickness measurement of the films. Second, these thickness data are analyzed by MO-VL BP and CG BP models to predict process trends. Third, the proposed model combines SPC-EPC techniques and finite feedback control model. According to SPC monitoring and construction of the disturbance model, then applied EPC feedback adjustment for coating operation. Finally, sensitivity analysis and DOE are conducted to find the best parameter settings of the optimal ANN-SPC-EPC model that can provide predictable, monitoring and feedback control for coating operation.
This research implements the proposed model of combining CG BP and MCEWMA-EPC techniques. Experimental results show that the adjustment times of the proposed model can reduce over 50%, accurately forecast the process trends, and have less almost 20% of the process variance when comparing with the traditional EWMA-EPC method. Obviously, the CG-MCEWMA-EPC model provides forecasting process trend function and process improvement scheme.
|
|---|
