Efficient symmetric Hessian propagation for direct optimal control. (February 2017)

• Record Type: Journal Article
• Authors: Quirynen, Rien; Houska, Boris; Diehl, Moritz
• Is Part Of: Journal of process control. Volume 50(2017:Feb.)
• Page Start: 19
• Page End: 28
• Publication Date: 2017-02
• Subjects: Optimal control -- Sensitivity analysis -- Algorithms and software -- Nonlinear predictive control; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2016.11.011
• Languages: English
• ISSNs: 0959-1524

Abstract

Abstract : Highlights: A symmetric Hessian propagation is proposed for explicit or implicit dynamics. The approach is presented for discrete- and continuous-time problems. A three-sweep propagation scheme can reduce the memory requirements. An open-source implementation is presented as part of the ACADO Toolkit. Abstract: Direct optimal control algorithms first discretize the continuous-time optimal control problem and then solve the resulting finite dimensional optimization problem. If Newton type optimization algorithms are used for solving the discretized problem, accurate first as well as second order sensitivity information needs to be computed. This article develops a novel approach for computing Hessian matrices which is tailored for optimal control. Algorithmic differentiation based schemes are proposed for both discrete- and continuous-time sensitivity propagation, including explicit as well as implicit systems of equations. The presented method exploits the symmetry of Hessian matrices, which typically results in a computational speedup of about factor 2 over standard differentiation techniques. These symmetric sensitivity equations additionally allow for a three-sweep propagation technique that can significantly reduce the memory requirements, by avoiding the need to store a trajectory of forward sensitivities. The performance of this symmetric sensitivity propagation is demonstrated for the benchmark case study of the economic optimal control of a nonlinear biochemical reactor, based on the open-source software implementation in the ACADO Toolkit.