Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/55252
Title: System identification for RTPCVD process
Authors: Yu, Feng.
Keywords: DRNTU::Engineering::Electrical and electronic engineering
Issue Date: 2013
Abstract: Semiconductor manufacturing industry is pursuing a higher degree of automation recently. However, currently the supervisory control of nitride process is based on experts' experiences and manual works. Although there are multiple control schemes that can be implemented to the process, due to the cost for testing control schemes on the real plant is too high, accurate models of process are required to conduct simulations. In this project, multiple models of the Rapid Thermal Processing Chemical Vapor Deposition (RTPCVD) process and their modeling, updating schemes are developed. Owing to the lack of technique details, the modeling is purely based on process data. And after deep analysis of the data, in this project, two feature selection algorithms are used: the Genetic Algorithm (GA) and Artificial Neural Network (ANN), and these two algorithms are specially modified for this project. These modifications include: applied binary encoding, two point crossover, bit string mutation and elitism strategies for GA; and optimized the architecture of RBF and MLP respectively for ANN. Based on the two modeling schemes, different datasets are used for modeling of the process. Comparisons are conducted based on the accuracy, generality and computational efficiency of the model generated. Furthermore, based on the analysis of the models obtained, merits, defects and usage of the models are given. Comments and recommendations are also proposed so that more research can be done focused on improving the performances of the modeling algorithms and the models.
URI: http://hdl.handle.net/10356/55252
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:EEE Theses

Files in This Item:
File Description SizeFormat 
Yufeng2013.pdf
  Restricted Access
Main Report3.69 MBAdobe PDFView/Open

Google ScholarTM

Check

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.