Artificial Fish Swarm Algorithm Improved Tikhonov Regularization Inversion Of Non-spherical Aerosol Particle Size Distribution

When the size distribution of atmospheric aerosol particles is obtained by using Mie scattering method,the obtained results often have some errors with the actual size distribution of atmospheric aerosol.This error is mainly due to the treatment of non-spherical aerosol particles in the atmosphere as spherical particles when Mie scattering is used to simulate aerosol particles.In order to solve this problem,a discrete dipole approximation method is proposed to simulate non-spherical aerosol particles.Firstly,the simulation of spherical aerosol particles is performed and compared with Mie scattering method to verify the applicability of the discrete dipole approximation method.The discrete dipole approximation method is used to analyze the anisotropy of aerosol particles with different shape parameters and to study the error of light scattering characteristics caused by anisotropy.Then the effects of the real and imaginary parts of complex refractive index on the light scattering characteristics of non-spherical particles are discussed.Then,the light scattering characteristics of non-spherical aerosol particles with different shape parameters in continuous wavelength range are discussed by setting fixed effective radius and ambient refractive index and comparing with Mie scattering method.Under the conditions of setting the ambient refractive index and incident wavelength,several groups of non-spherical aerosol particles with different shape parameters are simulated and the corresponding extinction,absorption and scattering efficiency factors are obtained.In this paper,Tikhonov regularization method is used to construct the inversion equation of aerosol particle size distribution.Aiming at the problem of parameter instability in Tikhonov regularization equation,artificial fish swarm algorithm is introduced to optimize the parameters.After experiments on the effects of step size,field of view,crowding factor and other conditions on artificial fish swarm algorithm,the optimal algorithm conditions are obtained and the corresponding regularization equation parameters are calculated.Under the condition of 0-3μm particle size,the optical data of extinction,absorption and scattering of aerosol simulated by discrete dipole approximation method and regularization parameters are substituted into the inversion equation,and the particle size distribution of non-spherical aerosol is obtained.In order to study the applicability of Tikhonov regularization method for artificial fish swarm optimization to particle size distribution,this experiment compared the inversion results by changing the environmental refractive index and the given lognormal distribution parameters.It is found that the particle size distribution of 1.50+0.02i refractive index fits the given distribution better than the particle size distribution of 1.33+0.01i refractive index under the condition that the lognormal distribution parameters remain unchanged and the ambient refractive index changed.When the ambient refractive index is constant and the lognormal distribution parameter is changed,the particle size distribution corresponding to r_mod=0.22 and lnσ=0.58 fits the given distribution better than that corresponding to r_mod=0.18 and lnσ=0.50.When two methods are used to invert the number concentration size distribution of non-spherical aerosol with different shape parameters,the maximum inversion error of GCV is 32.40%and the minimum inversion error is 10.24%,while the maximum inversion error of artificial fish swarm method is 30.06%and the minimum inversion error is 9.93%.The maximum inversion error of GCV is 38.41%and the minimum inversion error is 13.06%,while the maximum inversion error of artificial fish swarm method is 36.24%and the minimum inversion error is 12.85%,when two methods are used to invert the volume size distribution.The mean square error of spectrum distribution obtained by Tikhonov regularization method optimized by artificial fish swarm algorithm is less than that obtained by GCV method proving it is feasible to invert aerosol particle spectral distribution by Lagrange multiplier optimized by artificial fish swarm algorithm,and the artificial fish swarm method improves the accuracy of GCV method to a certain extent.