Quantum Neural Network Model And Its Application In Reservoir Identification

The accurate identification of oil and gas reservoir is the most important problem in the exploration and development,most of the recognition depends on the expert experience,but most of human factors lead to the low accuracy of reservoir identification.Thus,it will waste much power and material resources,but we will not achieve the desired results.It's difficult to describe the relationship between the influence factors of reservoir identification and the nonlinear mapping relation.The neural network can express the mapping relationship.However,the traditional neural has some obvious defects,such as poor approximation ability and lo-cal minimum.In recently,it has been shown that the quantum neural network,which is based on the combination of quantum computation and neural computation,it can effectively improve the defects of traditional neural networks.In this paper,we propose a new method to apply the quantum neural network to reservoir identification,which can improve the recognition accuracy of reservoir and improve the recognition accuracy.Quantum neural network is regarded to be a brand new field,it is still far from mature.It is necessary to further study its fusion with other algorithms,in order to improve the performance of neural computing.In this paper,we propose a new idea of the neural network model based on the rotation of the quantum bit around the Bloch sphere.Thus,a new quantum neural network model and algorithm are studied and put to use on identification of oil and gas reservoirs.The main advantage of the model lies in the parallel processing of information and the mufti-dimensional adjustment of network parameters,which improves its ability to approach and predict significantly.This paper mainly studies the following aspects:First,A new three-layer quantum neural network model is designed.The input and output layers consist ordinary neurons,the middle layer occupied by quantum neurons.The input of quantum neurons is the qubit,and its mapping mechanism works like this: First make the input bits rotate about the axis,then calculate the coordinates of the rotation using Pauli matrix.Finally use the sigmoid function to map the coordinate values to the output of the quantum neurons.Second,in the training of quantum neural networks,L-M algorithm and quantum bee colony algorithm are designed in this paper to resolve the problem.However,both algorithms have limitations.L-M algorithm converges faster,but it's easy to fall into local minimum.Although the quantum bee colony algorithm has better global optimization ability,using population optimization will cause low computational efficiency.Therefore this paperpresents a two-period training algorithm that combines the two algorithms.The specific research program comes as follows: First use quantum bee colony algorithm to explore weights in the network;Then the L-M algorithm is used to develop the local network weights.In the end,As for reservoir identification,this paper studies the recognition method based on hybrid quantum derivation neural network.First research the classification and factors which influence it.Then the reservoir identification method based on hybrid quantum derivation neural network is proposed in combination with the actual logging interpretation database of mine.This method shows a new path for reservoir identification.