Study Of Fracture Property Identification Based On Modified Particle Swarm Algorithm

With the rapid development of society and economy, the conventional energy is still in great demand before the new energy to get popularized, especially for big rise in demand for oil and gas resources. Thus the pace of exploration and development of oil and gas resources is also greatly accelerated. However, conventional structural oil and gas reservoir development has been almost exhausted. Therefore the current main research direction of oil and gas exploration and development turn into lithologic oil and gas reservoir and complex fractured oil and gas reservoir. However, geological conditions in the reservoir area are relatively more complex. So the exploration and development is more difficult. It is difficult to use conventional P-wave exploration method to solve. Multi-wave and multi-component seismic data can provide more abundant stratigraphic and lithologic information than single wave component, especially to carry the information of formation fracture. It can be more reliable to predict lithology and oil and gas distribution and provide an important basis for refined reservoir description to utilize the information synthetically.In the multi-wave and multi-component seismic exploration, when converted shear-wave propagates in anisotropic media containing cracks, it will produce fast shear-wave which is parallel to fracture strike, and slow shear-wave which is perpendicular to the fracture strike. This phenomenon is called shear-wave splitting phenomenon. Shear-wave splitting phenomenon can carry a lot of information of cracks. According to this phenomenon, the orientation and density of fracture can be extracted. Due to the polarization orientation of fast shear-wave is closely related to the fracture azimuth, the calculation of the fracture azimuth can be converted to calculate the polarization orientation of fast shear-wave. Because the propagation time delay of fast and slow waves reflects the fracture density, the predicting issue of fracture density can be converted to calculate the propagation time delay. At present, it is the more direct and reliable detection method that uses the characteristics of converted shear-wave splitting in multi-component seismic to study direction and development degree of fracture.In recent years, crack attribute recognition has new ideas and hope, with the method of genetic algorithm, particle swarm algorithm and the simulated annealing method are introduced into the nonlinear inversion of shear wave splitting fracture property identification. The particle swarm algorithm(PSO) is derived from the research on the foraging behavior of birds. It is an evolutionary computation technique based on swarm intelligence. Because the algorithm itself is characterized by simple implementation, parameter setting, fast convergence speed and so on, it has been applied in many fields.This paper first introduces the fractured anisotropic medium theory and principle of shear wave splitting. It transforms the fracture property identification into a highly non-linear complex optimization problem to use the mathematics in the Pearson correlation coefficient formula on the basis of the principle of shear-wave splitting. It is similar to the optimization of PSO algorithm. Then this paper briefly describes the theoretical basis and improved strategy of particle swarm optimization algorithm PSO, and analyzes the principle of PSO and its implementation flow. For the basic PSO has slow convergence and is easy to fall into local optimal solution, two improved PSO: Joined controlled operator shrinkage factor PSO and shrinkage factor PSO based on simulated annealing have been developed. Both algorithms expanded in the initial algorithm search space by shrinking factor, and enhance the convergence of particles. Joining controlled operator can maintain the particles swarm diversity and better convergence rate by introducing two operators: "absorption" and "diffusion". It make algorithm can accept optimization solution, at the same time with a certain probability to accept the deterioration of the solution according to Metropolis criterion to introduce simulated annealing algorithm. Diversity of the population is increased, and the algorithm is avoided falling into local optimal solution.On the basis of full understanding of particle swarm optimization theory, particle swarm algorithm is combined with Pearson correlation coefficient to apply in fracture property identification. The feasibility of algorithm is proved by processing the single trace seismic data. And two improved algorithm recognition results are compared with basic PSO algorithm recognition results to verify the effectiveness of the improved algorithm. In order to verify the effectiveness of the suggested method in noisy environments, the multi-trace seismic records added random noises are selected as the experimental data to identify fracture properties by applying two improved algorithm. The experimental results show that two proposed methods can identify fracture properties successfully in noisy environments and obtain accurate results. At the same time, the recognition results are compared with the recognition of the basic PSO, and it is found that the anti-noise performance and the stability of the improved algorithms are better than the basic PSO.