一种火山岩浆源参数反演的粒子群算法

A particle swarm optimization algorithm for inversion of volcanic magma source parameters

针对最小二乘(LS)方法、总体最小二乘(TLS)方法在用于火山Mogi模型反演压力源参数时,在线性化的过程中容易产生偏差,导致求得的参数解偏离真值的问题。该文分析了Mogi模型的非线性特点,同时考虑了粒子群算法在非线性问题求解的优势,将两者结合得到了一种适用于Mogi模型参数反演的粒子群算法(PSO)。通过模拟算例与真实火山反演的验证,表明该文方法所得结果在模拟算例上相对其他方法所得结果在精度上提高了1~2个数量级且更接近真值,在真实火山反演中其拟合结果更接近地表观测值,表明该文方法在火山Mogi模型反演中具有适用性与有效性。

When the least squares(LS)method,total least squares(TLS)method are used to invert the pressure source parameters of the volcano Mogi model,the deviation is easy to occur in the linearization process,which leads to the problem that the obtained parameter solution deviates from the true value.This paper analyzed the nonlinear characteristics of Mogi model,considered the advantages of particle swarm optimization in solving nonlinear problems,and combined the two results in a particle swarm optimization algorithm for Mogi model parameters inversion.The simulation results and the verification of real volcanic inversion showed that the results obtained by the proposed method were improved by 1 to 2 orders of magnitude and closer to the true value in the simulation example compared with other methods.In real volcanic inversion,the fitting results were closer to the surface observations.It showed that the proposed method had applicability and effectiveness in the volcanic Mogi model inversion.