Large‐scale DAE optimization using a simultaneous NLP formulation

Abstract

The differential-algebraic equation (DAE)

Prevention of Emergency Voltage Collapses in Electric Power Networks using Hybrid Predictive Control

“…To deal with control and optimization of systems modeled by DAEs, several strategies have been proposed, such as simultaneous strategies [6,9,21], multiple shooting strategies [11,12], and direct search methods [43]. The particular structure of power network models (see Section 3.3.4) can especially be used advantageously to set up tractable models for model-based control approaches, such as model predictive control (MPC) [31].…”

Process optimization of reactives systems by partially reduced SQP methods

“…These methods are particularly appropriate for the problem class at hand, since they profit from the fact that, although the total number of variables in the discretized system of equations is large, the number of real process influence variables is small. The partial reduction approach is similar to the reduced SQP methods described in Biegler, Nocedal and Schmid (1995); Cervantes and Biegler (1998) and Cervantes and Biegler (2000), but it differs from them in so far as it provides a wider family of methods, of which the latter methods are only one member. Its mathematical convergence theory can be found in Schulz (1996).…”

An Evolutionary Approach to Solve a Novel Mechatronic Multiobjective Optimization Problem

“…This approach is also known as control vector parameterization. In the simultaneous approach, the state and control variables are discretized resulting in a largescale NLP problem which requires of special algorithms for its solution [1]. Because of the diversity of mathematical programming algorithms already established, the transformation of the MDOP into a NLP problem was done adopting the sequential approach.…”

“…a set of solutions is randomly generated). We selected Differential Evolution [17] because of several reasons: (1) it is an EA which has provided very competitive results when compared with respect to traditional EAs such as genetic algorithms and evolution strategies in real-world problems [6], (2) it is very simple to implement [17] and (3) its parameters for the crossover and mutation operators generally do not require a careful fine-tuning [13].…”