Research On Source Inversion Problem In The Leakage Of Toxic Gases Based On Hybrid Genetic-Pattern Search Algorithm

With the development of chemical industry and related fields, some of the leakages of hazardous chemicals, especially toxic gases occurred from time to time. Due to the bursties and uncertainties, it often leaded to a large number of casualties and property losses. In view of the specificities of toxic gases, quickly estimating the locations and strength were critical in the emergency response. We must calculate the concentration and determine the scope of the affected region in a short time in order to reduce the losses to the minimum.According to the source strength, meteorological data, as well as the dispersion model, it would be able to determine the downwind concentrations of pollutants, and delimit the affected areas, thus quickly identification of the source strength was the basis for emergency rescue. In the paper, we used Gaussian puff model to simulate the instantaneous release. And then took the sum of the squared errors between measured and calculated concentrations, which were produced by the same dispersion model, as the objective function. Having posed the source inversion problem to an optimization problem, we used pattern search, genetic algorithm, as well as the hybrid GA-pattern search methods to optimize the objective function. The parameter value, leading to the minimum function value, was served as the optimal source strength. To analysis the stability of the modeling method, we added Gaussian white noise to measured concentrations to analyze the accuracy of the data on the impact of search results. We also considered the influence of the parameters of the algorithm, including the initial values, population sizes, crossover rates, mutation rates, and so on. The method was validated using the simulated concentrations; the results showed that it was feasible to use the pattern search, genetic algorithm, and the hybrid method to obtain the source strength. Meanwhile, taking the results of the genetic algorithm as the initial value of pattern search would speed up to converge to the global optimal solution.For multiple sources, the text studied the source apportions; simulating in a circular distribution of the source field, the genetic algorithm could obtain a correct apportion values in a reasonable time.Using the source inversion model, the true source strength could be obtained; with the results and the dispersion model, downwind concentrations estimated. They could guide to delimit the contaminated region and evacuation region, and also was helpful in the determination of the rescue plans and tactics in the emergency rescue of the leakage of toxic gases.