首页>
外国专利>
SEISMIC DATA-BASED OIL AND GAS RESERVOIR DISTRIBUTION-ORIENTED CONVOLUTIONAL NEURAL NETWORK LEARNING AND PREDICTING METHOD
SEISMIC DATA-BASED OIL AND GAS RESERVOIR DISTRIBUTION-ORIENTED CONVOLUTIONAL NEURAL NETWORK LEARNING AND PREDICTING METHOD
展开▼
机译：基于地震数据的油气储集卷积神经网络学习与预测方法
展开▼
页面导航
摘要
著录项
相似文献
摘要
A seismic data-based oil and gas reservoir distribution-oriented convolutional neural network learning and predicting method. Said method comprises firstly taking original seismic data as a basis, extracting seismic attributes which can represent oil and gas characteristics; then designing a convolutional neural network model, taking several preferred seismic attributes as an input layer of a network, extracting seismic attribute values at a well location, taking the void ratio, permeability and oiliness of the well location as a training sample, performing back propagation by means of a BP neural network, so as to continuously correct parameters such as the convolution kernel, weight W and bias b until the model training is completed; and then testing data of an area having no wells, so as to implement lateral prediction of reservoirs from an area having wells to an area having no wells. Said method directly performs convolutional neural network learning on oil and gas sensitive attribute bodies, being able to implement lateral prediction of seismic information characteristics of unknown seismic reservoirs in the same block and even cross regions.
展开▼
