00050 �40334d33ef8dd4883254e43d14c441

A Rule-Based Approach to Multiple Statistical Test Analysis

94.57% for the baseline. For processes measured in defects per million, the gain in detection speed without losing detection accuracy is extremely valuable.

This research provided for new approaches to the use of Artificial Intelligence (Al) in statistics. The system was a very simple rule-based application, and did not include any complex Al methods. The study shows that it is feasible to apply these techniąues and achieve good results, hopefolly opening the door for further study in this field.

Further Research

Further recommended research falls primarily in four areas: 1) further rules to investigate and/or implement, 2) morę involved Artificial Intelligence applications, 3) different methods to explore the multiple statistical test problem, and 4) further statistical and testing avenues to explore.

In summary, the system utilized a set of rules to measure incoming detection signals. The system evaluates the signals based on their initial values, their values five observations later, and the number of individual test firings during the additional runs. The system uses these points to backup signals received from a sensitive lead procedurę. The system gains from the speed of detection of the leadtest, and the accuracy the backup procedures offer.

References

Brownston, L., Farrell R., Kant, E., and Martin, N. (1986). Programming Expert Systems in OPS5: An Introduction to Rule-Based Programming. Addison-Wesley Publishing Co., Reading, MA.

Coleman, D. E. (1989). "Generalized Control Charting," Chapter in J. B. Keats and N. F. Hubele editors: Statistical Process Control in Automated Manufacturing, Marcel Dekker, New York, NY, 155-191.

Hunter, J. S. (1986). ‘The Exponentially Weighted Moving Average,” Journal ofQuality Technology, 18, 4, 203-210.

Kapur, K. C., and Lamberson, L. R. (1977). Reliability in Engineering Design. John Wiley and Sons, New York, NY.

Keats, J. B., Datę, S., and Kim, D. (1989). "Sequential Statistics for Binary Data Application in a Printed Circuit Board Manufacturing Facility," Proceedings of the Quality in Electronics Conference, San Jose, CA, 10/5-7, 1989.