This grant provides funding to develop statistical methods for improved quality control in the manufacture of parts. Such methods are of special importance, for example, in the automotive and electronics industries. The research concerns the appropriate development of feed-forward and feedback adjustment methods to supplement presently used monitoring techniques and to further minimize product variation. The new techniques will take into account the costs of making process changes and of the costs of necessary delays that can occur in getting the appropriate data. The methods will also be designed to work well even when departures from the underlying assumptions occur. Other research is planned on surveillance of systems. New methods will be designed that can quickly determine the occurrence of process malfunctions while minimizing the occurrence of false alarms that can be used while the process is being systematically adjusted. These methods are being tested in cooperation with the Low Emission Partnership (LEP) of General Motors, Chrysler and The Ford Motor Company. Appropriate process adjustment techniques can produce maximum process capability at minimum cost so that an existing industrial process can most closely approach its potential to manufacture items that meet the stringent specifications needed for present day competitiveness. The work on process surveillance can greatly extend and illuminate applications of process monitoring. In the past industrial quality control has been discussed almost exclusively in terms of process monitoring using Shewhart and quality control charts. This is evidenced by the content of the dozens of published books and of many of the courses taught in industries and universities. This research will put into the hands of a very wide community appropriately tailored systems of feed-forward and feedback control and of system surveillance at an accessible level and will augment the techniques already in use.
