Biomimetic Methods for Power Plant Optimization

Friedrich Biegler-König


Evolutionary methods, neural networks, mixed integer optimization


The energy production optimization of a fleet of power plants involves the solution of a mixed integer problem. Usually this is done by linearization of the continuous and non-linear parts of the problem and subsequent application of a Simplex-type algorithm. Alternatively, heuristic “biomimetic” optimization methods can be used which can also handle nonlinearity. As an example we are proposing an approach to first model power plant blocks with fast Neural Networks and then optimize the operation of multi-block power plants over a period of time using Simulated Annealing and Artificial Bee Colony Algorithm.

Important Links:

Go Back