A Genetic Algorithm Method to Assimilate Sensor Data for a Toxic Contaminant Release

摘要

—Following a toxic contaminant release, eitheraccidental or intentional, predicting the transport anddispersion of the contaminant becomes a critical problemfor Homeland Defense and DoD agencies. To produceaccurate predictions requires characterizing both the sourceof hazardous material and the local meteorologicalconditions. Decision makers use information oncontaminant source location and transport prediction todecide on the best methods to mitigate and prevent effects.The problem has both observational and computationalaspects. Field monitors are likely to be used to detect therelease and measure concentrations of the toxic material.Algorithms are then required to invert the problem in orderto infer the characteristics of the source and the localmeteorology. Here, a genetic algorithm is coupled withtransport and dispersion models to assimilate sensor data inorder to characterize emission sources and the wind vector.The parameters computed include two dimensional sourcelocation, amount of the release, and wind direction. Thismethodology is demonstrated for a basic Gaussian plumedispersion model and verified in the context of an identicaltwin numerical experiment, in which synthetic dispersiondata is created with the same dispersion model. Errorbounds are set using Monte Carlo techniques androbustness assessed by adding white noise. Algorithm speedis tuned by optimizing the parameters of the geneticalgorithm. Specific GA configurations and cost functionformulations are discussed.