City Research Online

Abstract

Nonlinear systems that can be described by ordinary differential equations without pure delays are considered. The difference in information contents between the model and the real system it is intended to represent is taken into account by allowing variations in the model.

The "goodness" of these models and the performance of the control applied are evaluated by cost functionals. Gradient algorithms are employed to calculate piecewise constant solutions that minimize the cost functionals.

Piecewise constant parameter trajectories in models may be preferred to continuous ones due to the reduction in the amount of information required to store them.

The dependence of the performance index of a control policy on model variations is estimated by its sensitivity to parameter variations in the model. The suboptimal piecewise constant solutions with cost close to the one corresponding to the continuous optimal control are convenient due to their low implementation cost.

Increase in accuracy and/or reduction in memory requirements are obtained by using spline functions to approximate the trajectories.

The algorithms developed are applied for parameter trajectory identification in a modelling problem, to the startup of a chemical process, the control of an economic process, and the control of an R&D project.

The sensitivity of the performance index to parameter variations provides an indication on possible improvement of the performance of the system.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science > QA Mathematics T Technology > TK Electrical engineering. Electronics Nuclear engineering T Technology > TP Chemical technology
Departments:	School of Science & Technology > Department of Engineering School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads