CONTROL SYSTEM SYNTHESIS TO MINIMIZE MULTIPARAMETER SENSITIVITY.
ILLINOIS UNIV URBANA COORDINATED SCIENCE LAB
Pagination or Media Count:
A new multiparameter sensitivity index for control systems is introduced. This measure of the deterioration in the system performance is based either on the integral square error or the mean square error in the system response due to parameter variations. A first order approximation to the change in the system transmission is used to simplify these functions. The index is extended to include situations when the varying parameters are stochastic. The expected value of the index with respect to the parameters is employed under these conditions. Moreover sensitivity analysis techniques for worst case conditions are developed to deal with parameter specifications in the form of tolerances. Synthesis procedures are derived which permit the minimization of the control system sensitivity with respect to the parameters. These procedures are investigated in terms of the control system stability and physical realizability. They also are applied to reduce response errors due to noise input signals and parameter variations, which occur simultaneously. The errors involved in the first order approximation to the change in the system transfer function are explored for a specific example. Numerical examples computed on the CDC 1604 computer are given to illustrate the analysis and synthesis procedures. Author
- Operations Research