Closed Loop Formulations of Optimal Control Problems for Minimum Sensitivity,
Abstract
A new formulation of the trajectory sensitivity problem is developed to reduce the effects of modeling errors in optimal control systems. Necessary conditions for minimum sensitivity are obtained from a measurable quasiconvex family of direction fields. These techniques are applicable to a large class of nonlinear systems that could not be handled previously by standard sensitivity methods. The principal result is a complete theory for the practical design of minimum sensitive linear feed back compensators. Sufficient conditions are developed from new theorems relating conjugate points to the positive definiteness and controllability of the accessory minimum problem. The advantages of the minimum sensitive compensator relative to least square parameter estimators are discussed. An example illustrates the improved sensitivity characteristics of the compensator as compared to model following and regulating controls. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1971
- Accession Number
- AD0728330
Entities
People
- Robert Nelson Crane
Organizations
- University of California, Los Angeles