Research On Model And Application Of Oil Field Development Indexes Prediction Based On Fuzzy Computation

Dynamic predictive control of oilfield development is the important way of realizing oilfield scientific and reasonable exploitation and guaranteeing the oilfield stable production. Currently most of our oilfields have continued in the mid-late period development phase. It is increasingly prominent that water raises sharply, natural decline increases, and stable production is difficult. How to control water stability and stable production is a serious problem faced by these oilfields.In order to solve these problems, except for the application of new technology, it is more important that to implement optimized control measures on the base of predict the dynamics of oilfield development so that both the oilfield development follow its own internal mechanism and the optimal along the expectations as possible development trajectories,making the complex oilfield development process control in the best operating condition.This paper studied the prediction problem of oilfield development indexes which based on fuzzy logic, evolutionary computation, neural network and quantum computing and so on. Combined with main development target forecast examples of Daqing Oil Field Co., Ltd. No.7 Oil Extraction Factory putaohua oilfield, several aspects of research as following.Firstly, a prediction method which based on T-S model is proposed.Aming at the problem of T-S model has many parameters and the conventional method not easily construct, a kind of quantum genetic algorithm which based on phase encoding (PQGA) is proposed and then a kind of T-S model construction scheme which based on PQGA is proposed. Experimental results verify the effectiveness of the proposed method.Secondly, a prediction method which based on fuzzy neural network is presented. At first, a fuzzy neural network (FNN) model is presented by combining fuzzy computation with neural networks. Then, a quantum particle swarm algorithm based on phase encoding of quantum bit is presented, which is applied to optimize parameters of FNN models. Taking prediction of the moisture content of oilfield development as an example, the effectiveness of the model is demonstrated.Finally, a prediction method which based on T-S reasoning network is proposed. At first, a T-S reasoning original model is proposed by fusing T-S model and neural network. Then, T-S reasoning original network model is constituted by a number of reasoning element. The model includes fuzzy set parameters and T-S consequent parameters. These two kinds of parameters are optimized and determined by the improved quantum particle swarm algorithm.The simulation results show that the model is effective and feasible.