An investigation of a deposit feature based screen print control system

Tools

Zhuang, W., 2000. An investigation of a deposit feature based screen print control system. PhD, Nottingham Trent University.

Preview

Text
10290136.pdf - Published version
Download (436MB) | Preview

Official URL: http://gateway.proquest.com/openurl?url_ver=Z39.88...

Abstract

This thesis investigates the complex control of screen printing machines, to enable the flexible production of a wide range of products. To meet the requirements of different products, these machines have controllable parameters, which should be set carefully to balance all the factors involved. In addition, to cope with the uncertainties involved in the printing environment and to achieve prints of consistently good quality, continuous adjustment of machine parameters during a print operation is also required. However, a large number of material and process factors and a lack of knowledge of the printing process make the set-up procedure complicated and identification of the control parameters difficult.

To solve the problems caused by the difficult set-up and control of the screen printing process, a lot of effort has been made by many researchers to build analytical models relating the process factors to the critical process output characteristics, including deposit thickness and deposit geometry quality. It was expected that if the requirements of the products had been specified and process factors, such as screen mesh and squeegee type had been decided, the machine parameters could be calculated based on equations derived from the models. By monitoring the process outputs and some process factors on-line, it would also be possible to develop a closed-loop control system based on these models. However, the literature review of existing screen analytical models has shown that the practical use of these models is constrained by the following problems: 1) These models usually represent some specific aspects of the printing process and some of them have not been fully validated by experimental results. 2) The models contain unknown quantities. 3) Some vital parameters used in the models are rarely measured during printing.

The aim of this thesis is to develop a new control strategy for the screen printing process. The fundamental idea of this strategy is to integrate inspection techniques with deposit image analysis algorithms to provide process state information for a control system to automatically and efficiently set up and control the machine variables. This is known as Deposit Feature- based Control System (DFBCS). The outcome of the experimental work confirms that the printed deposits contain rich information about the process and that the deposit's geometrical characteristics do represent the current state of the printing process. This thesis is focused on extracting the useful process information from the deposit image data and identifying the inverse process relationships. Based on the observation of deposit geometrical attributes, the deposit characterisation algorithm and the sequential forward selection algorithm (SFSA) are proposed to transmit raw inspection image data into useful process behaviour features. To establish the link between deposit features and machine controllable parameters, this thesis proposes the use of Adaptive Network-based Fuzzy Inference Systems (ANFIS) to implement a Deposit Feature-based Control System (DFBCS). The result of a simulation test illustrates that the proposed control strategy is effective and the resulting Deposit Feature- based Control System is robust.

This thesis provides a successful representation of screen printing process control. The algorithms proposed for deposit characterisation, feature selection and quality classification can be integrated into an inspection system to realise on-line process monitoring and supervision. The developed DFBCS possess a learning capability using its knowledge representation feature. It can be applied easily to other similar processes by manual tuning or on-line learning. The learning process has also become much more efficient, since important process structure knowledge is encoded in the developed DFBCS.

Item Type:

Thesis

Creators:

Zhuang, W.

Date:

2000

ISBN:

9781369323856

Identifiers: