Firefly Algorithm With Disturbance-Factor-Based Particle Filter for Seismic Random Noise Attenuation

Particle filter (PF) has been proven to be an effective method for seismic random noise attenuation. However, the sample impoverishment caused by resampling is an inherent problem, resulting in serious loss of valid seismic information and poor estimation accuracy. To solve the problem and to further improve the particle quality, we propose the firefly algorithm with disturbance-factor-based PF (FA-PF). In this method, we introduce the strategy of FA to simulate the movement process of particles instead of the resampling and optimize the FA method. We optimize the FA from three aspects. First, we adopt the global optimal value to guide the particle brightness update, improving the ability of global optimization and decreasing the probability of the local optimum. Then, the attraction decreases as the number of iterations increases, reducing particle oscillation. Finally, we apply a disturbance factor in the location update period, increasing the proportion of meaningful particles and further optimizing the distribution of particles. The experimental results show that particle diversity is well-maintained and the particles can represent the valid signal more accurately, which indicates that the FA-PF method performs better in suppressing the random noise and preserving the valid signal than the PF.