PROCESS IMPROVEMENT EXPERT SYSTEM (PI-XPERT)

摘要

This paper describes an expert system that is designed for use by aerospace companies to monitor and improve processes and products. The software, Process Improvement Expert (PI-Xpert), was developed through a NASA SBIR for use in shuttle maintenance operations at Kennedy Space Center. PI-Xpert has the ability to import data from a variety of sources, including Excel, Access, and Oracle. Prefiltering of the data is performed to ensure consistency of data format. Redundant variables are identified. The expert engine proceeds to build an appropriate model involving classification or prediction, or both, depending upon whether a dependent variable is available. Preliminary results are presented to the user for review and input. Output is presented to the user in graphical and narrative formats with the ability to edit the report. Statistical techniques that are available include correlations, control charting, regression, and process capability analysis. Additionally, data visualization techniques are employed to enable the user to discern patterns in the data. Techniques employed to identify appropriate grouping of the data include rough set theory and ID3. Optimization strategies employed include mathematical programming and multiple criteria decision-making techniques. If inadequate data sets are available to solve the problem, advice is offered to the user to design a set of experiments to collect the additional data. Use of PI-Xpert enables companies to identify opportunities to improve processes by identifying unnecessary data collection, by identifying process settings that generate higher process yield and/or quality, and by identifying out-of-control process conditions.