Non-Darwinian evolution for source estimation

A non-Darwinian evolutionary algorithm is introduced for the problem of source characterization of atmospheric pollutants from limited ground concentration measurements. Non-Darwinian evolution differs from traditional evolutionary algorithms because solutions are not generated through pseudo-random operators, but rather through a reasoning process based on machine learning rule generation and instantiation.

The proposed algorithm was tested on a synthetic case and on the Prairie Grass field experiment. To further test the capabilities of the algorithm to work in real-world scenarios, the source identification of all Prairie Grass releases was performed with a decreasing number of sensor measurements.

The proposed methodology has a wide domain of applicability, not restricted only to the source detection problem. It can be used for a variety of optimization problems, and is particularly advantageous for those problems where the fitness function evaluation involves a computationally expensive operation.