Statistical study of tank model identification by genetic algorithm

摘要

The tank model is useful for runoff analysis since it can represent a non-linear stream flow behavior. It is difficult to properly identify a lot of model parameters from observed data. The genetic algorithm (GA) is a search procedure based on the mechanism of natural genetics and is effi- cient for global optimization. Uncertainty of the identified parameters, which may significantly affect the prediction of stream flow, should be taken into account. This paper describes the statistics of the identified parameters using GA.