Seismic Wavelet Estimation Based On Adaptive Chaotic Embedded Particle Swarm Optimization Algorithm

How to get high-precision seismic wavelet extraction is an important research project inseismic data processing field. Accurate seismic wavelet is the base for seismic interpretationsuch as wave impedance inversion. In recent studies, high-order cumulant are widelyconcerned for its Gaussian noise suppression characteristics. Based on high order statisticalcumulant and the ARMA model, the group applied the combination of linear method (matrixmethod) and nonlinear method (cumulant matching method).Aimed at solving inaccuracy and inefficiency of the nonlinear method, a new algorithm-Adaptive Chaotic Embedded Particle Swarm Optimization Algorithm (ACEPSO) has beenpresent by studying on standard particle swarm optimization (SPSO). Firstly, ACEPSOadjusted the inertia weight and learning factors nonlinearly and adaptively, so the localsearching ability and the global searching ability can be balanced. Secondly, ACEPSOembedded chaotic variables in SPSO, and built a mechanism to observe the particles byjudging the population fitness variance of particle swarm and average distance amongst pointsto avoid the search being trapped in local optimum. Thus if the algorithm falls into the localsupreme, chaotic researching will be applied to jump out of local optimum to overcome thepremature.To verify the algorithm performance, ACEPSO had been used in the optical tests ofstandard test functions, synthetic seismogram process and real seismic seismogram process.Simulation results showed that the convergence precision and similarity of the waveletextracted by ACEPSO are better than chaos genetic algorithm (CGA) and improved particleswarm optimization algorithm (IPSO). The robustness and performance indicators showedACEPSO has good stability and high efficiency. In conclusion, ACEPSO is an algorithm of high accuracy, quick convergence speed and good robustness in extracting seismic wavelet.