Adaptive wave impedance inversion method and system based on deep convolutional neural network
By constructing a semi-supervised learning model based on seismic data and low-frequency constraint data, and adaptively adjusting the weights of the loss term, the limitations of training set size and network structure in seismic impedance inversion of deep convolutional neural networks are solved, achieving high-precision and efficient inversion results.
CN116047583BActive Publication Date: 2026-03-17CHINA PETROLEUM & CHEMICAL CORP +1
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-10-27
- Publication Date
- 2026-03-17
Smart Images

Figure CN116047583B_ABST
Abstract
The application provides a kind of adaptive wave impedance inversion method and system based on deep convolutional neural network, belong to seismic inversion interpretation field.The method includes: (1) generating training set;(2) constructing one-dimensional deep convolutional neural network wave impedance inversion model;(3) using training set to train neural network wave impedance inversion model, obtain trained wave impedance inversion model;(4) input the seismic data to be inverted and low-frequency constraint data into trained wave impedance inversion model, output inversion result.The application adopts semi-supervised learning mode, establishes the objective function consisting of seismic waveform loss term and low-frequency constraint loss term, constructs the deep convolutional neural network model containing multiple convolutional layers and full connection layers and suitable for wave impedance inversion, training set does not need label data, in the training process, the model can adaptively learn the weight coefficient combination of two loss terms in the objective function, effectively improve the inversion precision and efficiency.
Need to check novelty before this filing date? Find Prior Art
Citation Information
Patent Citations
Seismic inversion big data generation method based on convolutional neural network
CN110488350A
Small sample learning-based convolutional neural network earthquake-logging joint inversion method
CN110515123A