Andreas Lundell: Transformation techniques for signomial functions in global optimization
November 20, 2009 Andreas Lundell PhD theses
This thesis introduces some new transformation techniques for global optimization of nonconvex mixed-integer nonlinear programming (MINLP) problems containing so-called signomial functions. Signomials are a general class of function including, e.g., polynomials, bilinear and trilinear terms.
In the thesis the signomial global optimization (SGO) algorithm is introduced, in which reformulated convexified problems are iteratively solved to find the global optimal solution.
The thesis can be downloaded here.