Abstract
A useful approach to the calculation of optimal controls is to take a piecewise constant approximation to the control and to solve the resulting nonlinear program using available techniques. There is no way of specifying the required number of control intervals a priori, but this paper shows that the adjoint system used to calculate gradients for the optimization provides at each iteration sufficient information to assess the gain from increasing the number of intervals and to indicate the best locations for the appropriate switching times. An example is presented which shows the potential computational savings that can be realized when the number of control intervals is progressively increased until the desired accuracy of the approximation is achieved.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Sargent, R. W. H., andSullivan, G. R.,The Development of an Efficient Optimal Control Package, Proceedings of 8th IFIP Conference on Optimization Techniques, Würzburg, Germany, 1976; Springer-Verlag, Berlin, Germany, 1977.
Davison, E. J., andMonro, D. M.,A Computational Technique for Finding Time Optimal Controls of Nonlinear Systems, Proceedings of the 10th JACC Conference, pp. 270–280, 1969.
Horwitz, L. B., andSarachik, P. E.,A Computational Technique for Calculating the Optimal Control Signal for a Specific Class of Problems, Proceedings of the 2nd Asilomar Conference on Circuits and Systems, IEEE, pp. 537–541, 1968.
Mellefont, D. J., andSargent, R. W. H.,Calculation of Optimal Measurement Policies for Feedback Control of Linear Stochastic Systems, Proceedings of 8th IFIP Conference on Optimization Techniques, Würzburg, Germany, 1976; Springer-Verlag, Berlin, Germany, 1977.
Author information
Authors and Affiliations
Additional information
Communicated by C. T. Leondes
Rights and permissions
About this article
Cite this article
Mellefont, D.J., Sargent, R.W.H. Calculation of optimal controls of specified accuracy. J Optim Theory Appl 25, 407–414 (1978). https://doi.org/10.1007/BF00932902
Issue Date:
DOI: https://doi.org/10.1007/BF00932902