Study On Nonlinearinversion Methods For Elastic Impedance Of Seismic Prestack Data

The concept of elastic impedance (El) was formally proposed by Connolly in 1999, and then elastic impedance inversion spread quickly all over the world. Elastic impedance is a combination of acoustic impedance (AI) and AVO. As compared with AI, El overcomes the shortcomings of vertical incidence assumption and losing valid information sensitive to the oil and gas, and can make full use of a variety of logging curves, which makes it more comprehensive and more intuitive to distinguish the oil and gas information and can effectively reduce multi-solution of AI inversion. El can be used for not only stratigraphic inversion, but also for reservoir characteristics inversion. As compared with AVO technique, El overcomes the shotcoming of wavelet not changing with the offset in AVO inversion, and is a more robust inversion method. Elastic impedance inversion has become one of the development direction of seismic prestack inversion. Now the seismic inversion is moving towards the development of AI and El combination, AI and AVO combination.The studies on elastic impedance are mainly focused on the new elastic impedance formula derivation and the qualitative analysis in practical application, otherwise, there is a lack of systematic study and discussion for the accuracy of El formula and the quantitative calculation of El inversion. This paper discusses the basic principles of El and current eleven PP-wave El formula. According to the introduction of ray-path parameter, these El formula can be divided into two categories. With the forward modeling of different models, the main factors, such as K value, incident angle and the convolution model assumptions impacting on the accuracy of first Class El formula are discussed. Meanwhile, according to different forward modeling result, the accuracy of the two major categories of El formula are compared. These analysis will effectively provide a theoretical basis for the next El inversion.With regard to El inversion method, because the concept of El is very simple and its reflection coefficient expression is similar to AI's, current El inversion method also uses the linear method or the generalized linear inversion method, just like the post-stack inversion of AI. The linear method has inevitably led to its low accuracy, strong dependence on the initial model, easily falling into a local defects. To overcome these shortcomings, this paper introduces non-linear methods into El inversion process, which are two new non-linear inversion method-ant colony algorithm (ACA) and particle swarm optimization (PSO).The principle and status quo of ant colony algorithm are presented in this paper. Through the introduction of anti-S function and chaotic operator, a new hydrid ACA is raised, which improves the efficiency and precision of search. Four benchmark functions are used to test the performance of improved ant colony algorithm, and finally the improved ant colony algorithm is applied to the El inversion process of forward modeling data, which has achieved good results.After introducing the basic principle, parameter selection question and research status of particle swarm optimization algorithm, a new hydrid PSO is raised, which greatly enhances the global search ability and search accuracy of basic PSO by introducing of simulated annealing operator into discrete particle swarm optimization algorithm. The same four benchmark functions test its performance, and then the improved hybrid particle swarm algorithm is used in El inversion, and achieved good results.After the above discusses, a new El nonlinear inversion method based on ant colony algorithm and particle swarm algorithm with EI-Fatti formula is raised. To test the anti-noise ability of this new method, different degrees of random noise are added to the original data. The inversion results show that the new El inversion method has strong anti-noise ability. Even with the random noise up to 10% of the seismic angle gathers data, this new method can still achieve the reasonable elastic parameters. Besides, unlike the conventional El linear inversion methods, the new nonlinear inversion method does not depend on the initial model, does not depend on the interpretation of horizon.The final part of this paper is the application of new EI nonlinear inversion of prestack data in Shengli Oilfield. Although the goal layer buries deeper and AVO anomaly is not obvious, which make the reservoir prediction very difficult, the results of new EI inversion method raised in this paper still reflect the AVO anomaly. The well logging data interpretation also verifies the correctness of this method, which means that the new EI nonlinear inversion method has good potential and application prospects for the reservior exploration.