Metaheuristics
Efficiently explore parameter combinations for running all kinds of equipment and systems.
Efficiently explore parameter combinations for running all kinds of equipment and systems.
Metaheuristics is an organic fusion of empirical methods (heuristics) for solving highly difficult optimization problems in a real-world space with an enormous number of possible combinations.
The fundamental idea in this approach is to start with a certain combination, change it little by little, and then adopt it if it is satisfactory or try changing it again if it is not. This cyclic exploration process can be compared to the evolution of living organisms in which genes undergo change, and the organism survives if its genes suit the environment.
For example, when many devices are combined together to form a system, the system owner may wish to derive a combination of equipment parameters that will achieve the highest system performance. Recently, there have been active attempts to solve algorithms like this with quantum computers.
Reducing of the number of calculations through a fusion of our knowledge on equipment and systems.
Metaheuristics can obtain a robust solution by selecting the right method from various different options, but it also requires a huge amount of combinatorial calculations when many parameters need to be optimized.
Mitsubishi Electric has incorporated knowledge of the design and use of equipment and systems into its metaheuristics solution methods. We have developed technologies to acquire optimal parameter combinations and to achieve robust control of systems by selecting the parameters that best match the environment. This makes it possible to obtain optimum performance with a small number of calculations, and to plan system behavior and organizational actions appropriately.