Optimal sampling in a noisy genetic algorithm for risk-based remediation design

A management model has been developed that predicts human health risk and uses a noisy genetic algorithm to identify promising risk-based corrective action designs. Noisy genetic algorithms are ordinary genetic algorithms that operate in noisy environments. The noise can be defined as any factor that hinders the accurate evaluation of the fitness of a given trial design. The noisy genetic algorithm uses a type of noisy fitness function called the sampling fitness function, which utilizes sampling in order to reduce the amount of noise from fitness evaluations in noisy environments. This Monte-Carlo-type sampling provides a more realistic estimate of the fitness as the design is exposed to a wide variety of conditions. Unlike Monte Carlo simulation modeling, however, the noisy genetic algorithm is highly efficient and can identify robust designs with only a few samples per design. For complex water resources and environmental engineering design problems with complex fitness functions, however, it is important that the sampling be as efficient as possible. In this paper, methods for reducing the computational effort through improved sampling techniques are investigated. A number of different sampling approaches will be presented and their performance compared using a case study of a risk-based corrective action design.