A Study On Soft Computing Integrated Theory And Method Of Reservoir Forcasting Information Management

Nowadays, the oil gas exploration and development has been developed further, the areas which are superficial and simple reservoir condiction have got the comparatively sufficient exploration and development. with the fact that exploration marching towards deep reservoir and complicated collection condition areas, the oil exploration activity is more and more complicated, and the technology of oil exploration should be improved. Oil and gas development will focus on the complex oil and gas reservoirs in the future, but the complicated oil and gas reservoir, a non-uniform, non-linear and the uncertainty of the response characteristics, it is clear that based on the theory of traditional linear uniform statistical methods can not meet this Requirements. On the other hand, reservoir management refered in massive data of seismic, logging and geology, and then the general information processing technology can not be timely and effective treatment. Therefore it should be improve the accuracy and reliability of reservoir description. In order to maximum use of existing mining and logging and seismic information to adapt to the complex reservoir of oil and gas field exploration and development requirements, we must find a new information management technology to meet the oil exploration Development of new challenges.Soft Computing as a intelligent technology, makes full use of inaccuracy, uncertainty and some real information, is easy to handle and robustness of the merits of the problem areas through the use of integrated search method and reasoning in a complex reservoir oil and gas field exploration. It will play an important role in the exploration .In this paper, the National Natural Science Foundation Grant project to the "Oil Reservoir Management in the Integration of Soft Computing Theory and Methods (NO: 70573101)" as the basis, focused on the integrated soft computing technology in the reservoir prediction of information management theory and application. This paper researches on how to effectively integrate Soft Computing in the neural network, fuzzy logic, intelligent optimization algorit(?) (GA, PSO) to establish the link between reservoir parameters and seismic information, and how to extraction corresponding fuzzy rules. we can get the dynamic process of seismic data--parameters information--reservoir cognition through the optimization of seismicattributes and reservoir parameters lateral forecast from seismic reservoir fuzzy rules to achieve the dynamic process seismic data - parameters information- reservoir cognition. This paper studies the following:(1) Discussion the general principles and law among the soft computing technology integrated. This paper made more detailed analysis and discussion on soft computing integrated technology among neural networks, fuzzy systems, evolutionary algorithm on the basis of the summary overview of technological research and development.(2) It's proposed a seismic attribute optimization method based on GA-BP in this paper. The proposed method can adaptively optimize seismic attribute which extract from very complex seismic records by using GA binary coding and BP neural network integration methods. This method can focused on choosing the subset which is the most closely and best represent the characteristics of reservoir from all seismic attributes. The method can reduce redundant information and multiplicity of solution, and improve the ability of the reservoir prediction accuracy.(3)It's proposed an oil wells type recognition method by integrated GA-FCM. the method can optimize seismic attributes according to the relationship between seismic attributes of common deep point(CDP) and oil wells type (dry wells, low-production wells, high-production wells), and A very few seismic attributes which have the strongest relationship with wells type can be got. Used to recognize types of wells, it's not only the high recognition rate, but the results of the identification has strong explanatory by using fuzzy membership function.(4) A dynamic all parameters adaptive BP neural networks model is proposed. the method fused genetic algorithms (GAs), simulated annealing (SA) and error back propagation neural network (BPNN) to offset the demerits of one paradigm by the merits of another. Adopting multi-encoding, the model can optimize the input nodes, hidden nodes, transfer function, weights and bias of BP networks dynamically and adaptively. Under accurate premise, the simple architecture (less input and hidden nodes) of network model is constructed in order to improve the adaptation and generalization ability of networks, and to greatly reduce the subjective choice of structural parameters. The results of application on oil reservoir prediction show that the proposed model with comparatively simple structure can meet the precision request and enhance the generalization ability.(5) A improved the nearest-neighbor clustering center selection method of RBF networks is suggested. This method has been improved the centre Nearest Neighbor learning algorithm of RBF networks centers effectively based on discrete PSO, and then overcome the difficulty of radius selection and the deficiencies of Centre vector depend on input sequence excessively.(6)This paper proposes an integrated learning method, the MPSO-RBF integrated coding method. Integrated with PSO and RBF technique, it could solve the problem of finding the hidden node, optimize both RBF network parameter (center, width) globally and output layer weight. Compared with current RBF learning method thoroughly, this method has relatively less nodes in RBF hidden layer and better performance. It has been proved that this method could be successfully applied in seismic reservoir forecast.(7) Extracting fuzzy rules of seismic reservoir. According to the post-optimization seismic attribute and with neuron network, fuzzy system and evolution calculation joint together, I proposed GA-FNN network. This model has prominent physical meaning and is highly transparent and explainable. It has been applied successfully to the fuzzy rule extraction of relationship between seismic attribute and reservoir thickness, and then generates decision knowledge.In addition, to realize each soft computing integration method of this article, the author has compiled a lot of Matlab programme.The research of this paper is of great theoretical value to soft-computing integration. And it could guide the oil exploration, seismic reservoir information management in practice.