Masters Abstracts (1991)

SHIH, WURONG
(December 1991), Associate Professor, Nan-Tai Institute of Technology, Taiwan

Email: wurong@mail.stut.edu.tw

A Knowledge Based Advisor for Surface Mount Component Placement Sequence Identification A Knowledge Based Advisor for Surface Mount Component Placement Sequence Identification

The pick and place process is a critical factor in determining the throughput of a manufacturing line. The need for high speed, repetitive, accurate, and repeatable pick and place processes enhances the requirement to identify acceptable (possibly optimal) component placement sequences. The component placement optimization problem has been analyzed using several human approaches and computationally intensive approaches. An approach that mimics the rational thought process of a human expert can consider and effectively deal with the dynamic situation on the shop floor and incorporate the subtle heuristic approaches used by humans in arriving at placement sequences.

The global objective of this research was to design and develop a knowledge based expert system which can yield an acceptable (possibly optimal) placement sequence while considering facility, machine, and process constraints in a batch type production environment. The prototype system considers factors such as the machine capacity and characteristics, component information, types of placement heads on machine, optimization of the tooling and nozzles changes, and the configurations of feeders. Some of these factors are interrelated.

A knowledge based expert system that uses a tandem system architecture was designed and developed. This structure permits the integration of an expert system and an optimization (computation) module in attempting to deduce an acceptable placement strategy for PCB assembly. The qualitative knowledge based expert system provides inputs to and request outputs from the optimization module. The output from the optimization module is evaluated by the expert system prior to actual use in the reasoning process.

The rule base of the system is designed in a modular fashion , where each module is developed independent of the other. The modules were developed using PROLOG, a popular language used in artificial intelligence applications. Furthermore, the declarative predicate formats of PROLOG can properly represent the domain knowledge of the SMC placement process. The system can run as a stand alone application on a personal computer with a minimum of 640 Kbytes of Random Access Memory (RAM). This design enables the user to setup this knowledge based system on the same computer where the operating system of the selected pick-and- place machine is installed.