Research On Particle Size Logging Interpretation Method Of Low Permeability Oilfield Reservoir Based On Quantum Neural Network

With the intensification of my country’s oil development efforts,there are more and more undeveloped reserves with low permeability and difficult to exploit,and the low permeability production accounts for a relatively low proportion of the total production.One of the key technical difficulties in how to effectively develop low-permeability oilfields is how to master the particle size distribution of the reservoir.If sufficient particle size distribution of the reservoir can be obtained in an economical way,long-term production of a large amount of oil and gas in the reservoir may be realized.In this paper,a particle size logging method for low-permeability oilfield reservoirs based on quantum neural network is designed,respectively,from the optimization of logging parameters,the basis of quantum neural network,the quantum neural network algorithm of improved bee colony,and the interpretation of particle size logging of quantum neural network in low-permeability reservoirs.The application of the in-depth research,the research content is as follows:1.Research on optimization method of reservoir granularity logging interpretation parameters.Increase the number of performance metrics by sorting each attribute in the dataset,split by attributes in a single decision tree,and use nodes to measure and record time to calculate feature importance.Finally,the feature results of all boosted trees are weighted and summed,and then averaged to get the importance score,and the feature importance score is ranked high to prove that it is important.Using the gradient boosting algorithm,the importance score of each feature can be obtained relatively directly.2.Quantum neural network foundation.The quantum state replaces the 0s and 1s in the classical bits.The state of the quantum neuron consists of the input of the states of the remaining neurons.The quantum neuron model includes two parameters,the phase parameter in the form of connection weight and threshold,and the inversion parameter.A quantum neural network has a three-layer architecture,where the first layer is the input layer,the second is the hidden layer and the last is the output layer.3.Quantum neural network algorithm based on improved bee colony.A new Adaptive Average Artificial Bee Colony Algorithm(AMABC)is introduced,which is based on an improved search equation based on the mean information of the best solution to achieve a balance of search behavior.(pbest)and the mean value of the positive direction of the global best position(gbest)are compared to update the search equation in the proposed AMABC algorithm.Compared with the test function,it is verified that the adaptive average artificial bee colony algorithm is stronger than the original artificial bee colony algorithm in terms of global convergence speed,solution quality and robustness.The improved bee colony algorithm is applied to the training process of quantum neural network to improve the learning ability and convergence speed of the network training process.4.Application of quantum neural network in granular logging interpretation of low permeability reservoirs.First,based on the combination of reservoir geological characteristics and analysis of logging data,the modeling wells and test wells in the block wells are determined to be used for the training set and the test set respectively;And the data is normalized.Finally,it is proved by experiments that the quantum neural network model based on the improved bee colony algorithm has higher learning ability and prediction ability for logging data.Experiments show that the particle size logging scheme for low-permeability oilfield reservoirs based on quantum neural network can meet the requirements of low-permeability reservoir particle size prediction accuracy,and has strong theoretical significance and broad application prospects.