中文摘要：

针对粒子滤波（PF）应用到含“开关”过程的资料同化时，在采样过程中次优的重要性函数，为了优化采样过程，将粒子群优化算法（PSO）融入PF中（称为PSOPF方法），并将其用于含“开关”过程的资料同化中。为了检验PSOPF的有效性，利用一个含“开关”的偏微分方程作为控制方程，分别从观测算子为线性和非线性、观测误差为高斯型的情形用PF方法和PSOPF方法进行了数值实验。结果显示，在观测误差为高斯噪声时，无论观测算子是线性的还是非线性的，基于PSOPF的同化方法质量明显好于基于PF方法的同化方法。

英文摘要：

When the particle filter （PF） is applied to data assimilation with physical＂ on-off＂ process, importance functions adopted are suboptimal during importance sampling processes. In order to reduce the impact of this suboptimality, this study attempted to introduce particle swarm optimizations （PSOs） into PFs. To test the effectiveness of the PSOPF, an idealized model of partial differential equation discontinuous＂ on-off＂ switches in the forcing is used in this paper, and numerical experiments are conducted for two cases of linear and nonlinear observation operators, Gaussian observation errors. The assimilation results show that no matter the observation operator is linear or nonlinear, assimilation approaches based on PSOPF in quality is apparently more advantageous than that based on PF.