Towards an innovative virtual optimisation machine for the power industry (Q979522)

Summary: The present work reports the development of a fully automated, innovative, optimisation environment (optimisation machine) based on evolutionary algorithms. The evolutionary algorithm uses the same mechanisms (selection, recombination, and mutation) as the natural evolution to identify fittest (=optimised) individuals (=operational settings and designs). The optimisation machine is capable of automatically determining potential better scenarios, evaluating the achieved results, and working out future development strategies. The applicability of this machine for industrial problem solving is demonstrated by the computer-aided repetition of successful optimisation exercises in power plants. The optimisation machine is able to identify the same optimised settings that can be found experimentally. A comparison of the required time and manpower effort clearly shows the advantages of the computer-aided approach.