Enhancing Situation Awareness of Chemical Release Through Source Inversion

Situation awareness is an important function module contained in the information system for regional emergency rescue. To improve the existing emergency information system, a situation awareness model of chemical release is constructed containing three levels, i.e. Data Acquisition, Intelligent Analysis and Simulation & Prediction, in which Intelligent Analysis was designed as an independent functional item for unknown source detection. Combining the receptor data with the atmospheric dispersion model, the source items estimation problem was then converted to a standardized concentration field fitting problem. Particle Swarm Optimization (PSO) was introduced to optimize the combination solution of multi-parameters including source strength, location, height and release time. The effectiveness and applicability of this method was verified through dozens of simulated tests. Further performance comparison with Nelder Mead Simplex Method and Genetic Algorithm show the results: in terms of estimation accuracy, computational efficiency and algorithm robustness, PSO are all superior to the other two algorithms. It can be flexible with different atmospheric dispersion models for fast source inversion in integrated situational awareness system.