Recent improvements to a multiplier-free reduced Hessian successive quadratic programming algorithm

Nonlinear programming (NLP) algorithms form the core of many optimization tools in process systems engineering. Over the past 15 years, the successive quadratic programming (SQP) algorithm has been especially useful for a wide variety of processing applications. In particular, the reduced space SQP...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Computers & chemical engineering Ročník 22; číslo 7; s. 963 - 978
Hlavní autoři: Ternet, David J., Biegler, Lorenz T.
Médium: Journal Article
Jazyk:angličtina
Vydáno: Oxford Elsevier Ltd 01.01.1998
Elsevier
Témata:
Applications of mathematics to chemical engineering. Modeling. Simulation. Optimization
Applied sciences
Chemical engineering
Exact sciences and technology
nonlinear programming algorithms
successive quadratic programming algorithms
successive quadratic programming algorithms
nonlinear programming algorithms
Quadratic programming
Non linear programming
Optimization
ISSN:0098-1354, 1873-4375
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Nonlinear programming (NLP) algorithms form the core of many optimization tools in process systems engineering. Over the past 15 years, the successive quadratic programming (SQP) algorithm has been especially useful for a wide variety of processing applications. In particular, the reduced space SQP algorithm has seen many applications to large-scale engineering models. Designed for large NLP problems with few degrees of freedom, this approach has been implemented and applied to engineering problems ranging from flowsheet optimization, on-line process optimization, dynamic systems and even applications with PDE models. This paper describes recent advances in the development of reduced space SQP algorithms. In particular, we investigate several enhancements that improve the efficiency, performance and reliability of these algorithms. Moreover, all of these enhancements preserve the desirable convergence properties shown in our earlier studies. These enhancements are demonstrated on a wide variety of test problems, including scalable mathematical problems, a library of equation-based NLPs and realistic problems in real-time optimization. Results indicate significant performance improvements due to these enhancements.
Bibliografie:ObjectType-Article-2
SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 23
ISSN:0098-1354
1873-4375
DOI:10.1016/S0098-1354(98)00009-X