Research On Parameter Estimation Algorithm For Dangerous Gas Source In Complex Environment

With the development of modern chemical industry,all kinds of dangerous gases have potential safety hazards in the process of production,transportation and storage.Once the dangerous gas leakage accident occurs and can not be timely and effectively warned,it’s bound to cause serious damage to people’s production and living environment,even cause casualties.It is very important for carrying out emergency rescue activities to get fast,accurate and effective identification of dangerous gas source location,source strength and leakage initial time and other parameters information.In this thesis,the research status of gas diffusion model and gas source parameter estimation algorithm is introduced in detail.The general framework of gas source parameter estimation is summarized.The optimization method and Bayesian method of gas source parameter estimation are discussed in depth.The corresponding gas diffusion model is deduced pertinently around the real complex dangerous gas leakage scenario,and the algorithm for estimating the parameters of dangerous gas sources under various scenarios is proposed.Aiming at the situation that the dangerous gas diffusion of continuous leakage gas source has reached a steady state,a steady-state gas diffusion model of continuous leakage is derived based on turbulence theory,and then a parameter estimation algorithm of steady-state gas source based on particle filter is proposed by using this diffusion model.The weighted centroid algorithm is used to roughly estimate gas source parameters,the estimated results are applied to particle filter.Through these operation,the position and source strength of steady-state gas source can be estimated quickly and accurately.The proposed steady-state gas source parameter estimation algorithm optimizes the likelihood function based on multi-sensor measurements,and improves the stability of particle filter convergence.In order to verify the performance of the proposed algorithm,the Cramér-Rao lower bound of the estimated parameters is derived based on the steady-state gas source model,and the root mean square error of the estimated parameters approximates the Cramér-Rao lower bound of the theoretical derivation.Simulation results show that the proposed algorithm has faster convergence speed and stronger robustness than the weighted centroid algorithm,which verifies the good performance of the proposed algorithm.In view of the scenario that the dangerous gas diffusion of a continuous leaking gas source does not reach a steady state,the law of concentration variation with time at the initial stage of gas leakage is considered.According to Fick’s diffusion law,a non-steady-state gas diffusion model of continuous leaking is deduced.Based on this model,under the framework of maximum likelihood estimation,the problem of gas source parameter estimation is transformed into optimization problem,and then a parameter estimation algorithm of unsteady gas source based on multi-population genetic algorithm in two-dimensional environment is proposed.Compared with the traditional genetic algorithm,the proposed algorithm adopts a multi-population co-evolution scheme which is mutually independent and has different evolution mechanism,has the advantages of strong global search ability and fast convergence speed,and can effectively estimate the parameters of unsteady gas source.At the same time,compared with the weighted centroid algorithm,the proposed algorithm has better positioning accuracy of gas source.The simulation results also verify the validity and superiority of the multi-population genetic algorithm for the estimation of unsteady gas source parameters.Considering the dangerous gas diffusion scenario of instantaneous leaking gas source,an instantaneous gas diffusion model is established by using Euler method,and a parameter estimation algorithm of instantaneous gas source based on Gaussian particle filter is proposed.The algorithm uses weighted centroid algorithm to get the initial value of parameter estimation,which improves the convergence speed of Gaussian particle filter.The likelihood function is reconstructed from the measured values of multiple sensors,and the stability of the proposed algorithm is improved.Based on the established instantaneous gas diffusion model,the Cramér-Rao lower bound of instantaneous gas source estimation parameters with different number of sensors is derived.The instantaneous gas source parameter estimation algorithm based on Gaussian particle filter can quickly and accurately estimate the location of the instantaneous gas source,the initial leakage time and other parameters,and can reduce the estimation error.The performance of the proposed algorithm approximates the theoretically deduced Cramér-Rao lower bound.Simulation results show that the proposed algorithm has smaller estimation error than the classical particle filter algorithm,and the performance of instantaneous gas source parameter estimation becomes better with the increase of the number of sensors.