Evolutionary algorithm enhancement for model predictive control and real-time decision support

Model predictive control of urban drainage systems: A review and perspective towards smart real-time water management

“…There is a trade-off between the accuracy in the description of the system dynamics in the internal MPC model and the computation time it takes to solve an optimization model that encloses these dynamics. The two extremes in the reviewed literature are the papers by Zimmer et al (2015) and Joseph-Duran et al (2014c). In the high end with respect to computation time, Zimmer et al (2015) reported computation times in the order of 0.5 to 1.5 hr for solving a 1-hr control horizon problem using genetic algorithms.…”

“…The two extremes in the reviewed literature are the papers by Zimmer et al (2015) and Joseph-Duran et al (2014c). In the high end with respect to computation time, Zimmer et al (2015) reported computation times in the order of 0.5 to 1.5 hr for solving a 1-hr control horizon problem using genetic algorithms. In the low end, Joseph-Duran et al (2014c) solved a mixed-integer linear program for a 40-min control horizon in, on average, less than 1 sec.…”

Empirical investigation and analysis of the computational potentials of bio-inspired nonlinear model predictive controllers: success and challenges

“…To design a powerful MPC, it is necessary to seek for the implementation of highly efficient control-oriented models as well as powerful real-time optimisation techniques to cope with any nonlinear objective function. Despite of such a fact, there exist rare reports on using BICs as an optimisation module of MPC (Naeem et al, 2005;Zimmer et al, 2015;Sarailoo et al, 2015). In this study, a comparative investigation is carried out considering different types of controlling problems.…”

Thematic issue on Evolutionary Algorithms in Water Resources

“…As part of the analysis, the performance of seven different EAs (or similar heuristic search methods) for use in the outer loop is compared, including a simple genetic algorithm, an improved genetic algorithm, particle swarm optimisation, simulated annealing, dynamically dimensioned search, dynamic coordinate search using response surface models and the stochastic radial basis function method. Zimmer et al (2015) explore an approach to reducing the size of the search space and increasing the computational efficiency of real-valued EAs in the context of model predictive control for timevarying systems with moving decision windows. This is achieved by investigating the impact of a range of modifications to standard genetic algorithms, including gene shifting, use of a reduced alphabet, application of the compact genetic algorithm and use of the micro genetic algorithm.…”