Browse Prior Art Database

Method for Automated Control of Multi-Variable Processes

IP.com Disclosure Number: IPCOM000115650D
Original Publication Date: 1995-Jun-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 2 page(s) / 57K

Publishing Venue

IBM

Related People

McKay, JM: AUTHOR [+2]

Abstract

A programming method is disclosed that allows continuous optimization of process settings in production processes. The method is novel in that it uses inaccurate response maps generated from production data, rather than accurate maps produced by experimental data.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 59% of the total text.

Method for Automated Control of Multi-Variable Processes

      A programming method is disclosed that allows continuous
optimization of process settings in production processes.  The method
is novel in that it uses inaccurate response maps generated from
production data, rather than accurate maps produced by experimental
data.

      In the classic linear programming problem, a set of hyperplanes
form the boundary of an allowed solution space, and the optimal point
within the space is one of the hyperplane intersections.  The
hyperplane intersection that is optimal is determined by the
"objective function".  In a practical application to a multivariable
manufacturing problem the hyperplanes correspond to boundaries on the
allowed values for process settings and resulting product parameters.
The objective function to be maximized usually corresponds to
variables related to total cost such as material cost, process
thruput, and defect rates.

      The conventional approach to solving the manufacturing problem
is to first carry out experiments on the manufacturing system to
obtain accurate estimates of the relation between process settings
and product parameters (forming the boundaries of the solution
space).  These relations, with an objective function, can be solved
to obtain the optimal vertex in the solution space.

      Since the optimum solution point is a vertex of the solution
space, an identical solution point will be discovered regardless of
cha...