Optimal design of generalized multiple model adaptive controllers

Advanced analysis and optimal design techniques that achieve performance improvement for multiple model adaptive control (MMAC) and multiple model adaptive estimation (MMAE) based control are developed and tested. An adjunct area of research yielded modified linear quadratic Gaussian (LQG) control design techniques that also can be applied to nonadaptive control. For the modified LQG (MLQG) controller, the proposed designs remove the assumption that the Kalman filter and the controller gain matrix design are necessarily based on the same model as the best system model. The proposed modified MMAC (M3 AC) architectures use the MLQG controller. The performance improvements for the MLQG controller carry over to the M3 AC architectures as well as to the MMAE-based control architecture. Design approaches developed for the M3 AC are applied to the MMAE-based control with similar performance improvements. Analyses of a representative example of the new design implementations demonstrate the performance improvements of the proposed architectures by comparing the results with those of the typical MMAC and LQG implementations. Though incidental to this research, the performance enhancement of the MLQG controller itself has proven to be significant and the possibilities for application to non-adaptive control transcend this research.