Optimal input-output variable assignments for multivariable systems

✍ Scribed by L.F. Yeung

• Publisher: Elsevier Science
• Year: 1991
• Tongue: English
• Weight: 557 KB
• Volume: 27
• Category: Article
• ISSN: 0005-1098
• DOI: 10.1016/0005-1098(91)90067-c

✦ Synopsis

The interchannel couplings of a multivariable system can be measured in terms of row and column dominance measures. Permuting rows and columns, which is equivalent to permuting output and input variables of a system, respectively, will structurally alter the original system and hence its cross-couplings. Experience shows that by just interchanging two rows (or columns) at a time one can improve the column (or row) dominance measures but cannot produce an optimal solution. In this paper an optimal solution to this permutation problem can be derived through the application of graph theory, and the permutation problem can be transformed into assignment problems. A recursive bottleneck assignment algorithm is proposed.