Based On GA Algorithm Fuzzy Nerve Network In Wave Impedance Inversion Application

Regarding the fine oil gas exploration, the earthquake wave impedance inversion achievement plays more and more the role in research works and so on in oil deposit description and oil gas crosswise forecast, it can earnest reflect on the spot the underground geologic structure and the lithological character, and can enhance the well drilling greatly the success ratio. Earthquake wave impedance inversion itself belongs to the multi parameter non-linearity question. Because the neural network non-linear inversion is in a non-linear earthquake inversion theory part, simultaneously the genetic algorithm is the multi parameter non-linearity optimization question which one kind of now time develops, Therefore this article through absorbs and digests the domestic and foreign scholars fully about the non-linear optimization research results, in carries on in the fuzzy neural network research process to unify the genetic algorithm to carry on the earthquake wave impedance inversion.This article has first analyzed the ANFIS fuzzy neural network and the genetic algorithm development and the application present situation, as well as the genetic algorithm basic concept, the integrant part, the operation process, the characteristic, the rationale and the basic genetic algorithm involve content research.. And unifies the GA algorithm and ANFIS, applies it in optimizes the ANFIS fuzzy neural network the premise parameter and the conclusion parametric problem; Next, this article has studied the fuzzy neural network correlation theories as well as the ANFIS fuzzy neural network characteristic, the ANFIS fuzzy neural network study rule and the performance; Once more, this article proposed one kind the genetic algorithm and Tabu Search (taboo search algorithm) will unify the auto-adapted mix study algorithm, and practiced two aspects from the theory to prove its feasibility and the usability; Finally, has designed the GA-ANFIS auto-adapted mix algorithm with the matlab language and applies it in approximation of function and questions and so on processing actual earthquake material, had proven from the practice the auto-adapted mix algorithm fuzzy neural network has the good search speed and the restraining precision.