Seismic inversion problems often involve nonlinear relationships between data and model usually have many local minima. Linearized methods been widely used to solve such problems. However, these kinds of strongly depend on the initial are easily trapped in a minimum. Global optimization methods, other hand, do not require very good can approach global exhaustive search techniques that be time consuming. When dimension or space becomes large, slow converge. In this paper, we propose new...

Seismic full-waveform inversion (FWI) aims to build high-resolution images of the physical properties subsurface. However, ill-posedness and nonlinear problems pose a great challenge reconstruction. Although problem can be mitigated by matching subset observation data, resulting are generally low-resolution background structures. Regularization-based techniques mitigate FWI, but iterative method suffers from cycle-skipping computational burden problems. To overcome these problems, we develop...

In this letter, a new differential evolution (DE) algorithm is proposed and applied to waveform inversion. The traditional strategy of not efficient because it treats the individuals in population equally evolves all them each generation. order overcome shortcoming, we propose (PES) decrease size based on differences among during an process. We embed into cooperative coevolutionary DE (CCDE) obtain highly (HEDE). apply inversion experiments both synthetic real seismic data test its...

Seismic high-resolution processing plays a critical role in reservoir target detection. As one of the most common approaches, regularization can achieve inversion result. However, performance depends on settings associated parameters and constraint functions. Further, it is difficult to solve an objective function with complex constraints, requires designing optimization algorithm. In addition, existing algorithms have high computational complexity, which impedes large data volume. To...

Seismic inversion problems are well-known to be nonlinear and their misfit functions often involve many local minima. Global optimization methods capable of converging the global minimum a function, thus, they promising in seismic inversion. As method, multimutation differential evolution (MMDE) has been proven effective solving high-dimensional problems. However, it is challenging choose optimal parameters for MMDE achieve best performance In this paper, we propose new deep network based on...

Time–frequency analysis is an important tool used for the processing and interpretation of non-stationary signals, such as seismic data remote sensing data. In this paper, based on novel short-time fractional Fourier transform (STFRFT), a new modified STFRFT first proposed which can also generalize properties (STFT). Then, in domain, we derive instantaneous frequency estimator chirp signal present type synchrosqueezing (FRSST). The FRSST presents many results similar to those STFT (FSST), it...

Full-waveform inversion (FWI) attempts to find optimal models of subsurface by using full information the observed data. One difficulty in conventional FWI is that misfit function has many local minima because cycle skipping. Envelope (EI), which uses envelope operator (EO)-based function, been proven be effective mitigating skipping and recovering long-wavelength velocity model. However, EI ignores fact within different frequency bands plays roles inversion. In this paper, a controllable...

Seismic inversion problems often involve strong nonlinear relationships between model and data so that their misfit functions usually have many local minima. Global optimization methods are well known to be able find the global minimum without requiring an accurate initial model. However, when dimensionality of space becomes large, will converge slow, which seriously hinders applications in large-dimensional seismic problems. In this article, we propose a new method for based on machine...

In this letter, a new adaptive differential evolution (DE) for high-dimensional waveform inversion is proposed. conventional DE algorithms, individuals are treated as whole and share the same fitness function parameters. However, algorithms have ignored huge difference among subcomponents in an individual not effective problems. Therefore, problems, we expand unit of crossover rate from to its propose adaption algorithm by adjusting each subcomponent. our algorithm, both kinds rate,...

Subsurface scatters are sometimes masked by reflectors in seismic migration images, because the diffractions much weaker energy than reflections. We propose a novel imaging method, named reflection-damped plane-wave least-squares reverse time (RD_PLSRTM), to enhance image. formulate as an inverse problem that minimizes weighted residual between modeled and observed data. In proposed approach, we use destruction filter separate from reflections data residual. A weighting matrix is then used...

As a rock-physics parameter, density plays crucial role in lithology interpretation, reservoir evaluation, and description. However, can hardly be directly inverted from seismic data, especially for large-scale structures; thus, additional information is needed to build such model. Usually, well-log data used model through extrapolation; however, this approach only work well simple cases it loses effectiveness when the medium laterally heterogeneous. We have adopted deep-learning-based...

Summary Full waveform inversion (FWI) is often formulated as an optimization problem to derive the best model that can minimize difference between a field data and simulated one. Usually, iterative gradient based methods are employed for problem. These require accurate initial avoid cycle-skipping, which cannot be described by Born approximation. Many used build such FWI, like reflection tomography migration-based velocity analysis. In this paper, we propose another way FWI using global...

ABSTRACT The elastic reverse time migration approach based on the vector‐wavefield decomposition generally uses scalar product imaging condition to image multicomponent seismic data. However, resulting images contain crosstalk artefacts and polarity reversal problems, which are caused by nonphysical wave modes angle‐dependent reduction of amplitudes, respectively. To overcome these two we develop an amplitude‐preserving vector‐decomposed P‐ S‐wave records. This includes key points. first is...

The convolutional model, which describes the relation among poststack seismic data, wavelet, and reflectivity, is foundation of deconvolution (SD). However, this model only an approximation wave equation, it may not work in complex cases especially when medium anelastic, heterogeneous, anisotropic. In article, we propose a generalized for data. A deep-learning-based data correction term added to characterize ingredients that cannot be characterized by model. new realized using long-short...

Seismic full waveform inversion (FWI) is able to build high-resolution velocity model based on the information carried by seismic wave. However, FWI requires an accurate enough initial ensure convergence. In this paper, we propose a new nonlinear method mitigate dependence problem. Specifically, firstly operator within hybrid model- and data-driven framework frequency controllable envelope (FCEO) deep learning architecture U-Net. FCEO used obtain of band-limited data U-Net realizes mapping...

Multiple-point geostatistics (MPS) is a competitive algorithm that produces set of geologically realistic models. Viewing training image (TI) as prior model, MPS extracts patterns from the TI and reproduces which are compatible with hard data (HD). However, two challenges within framework complex simulation evaluation. With objective to achieve high-quality simulation, we explore way address these issues. First, correlation-driven direct sampling (CDS) proposed realize geostatistical...

In this letter, we propose a new global optimization method for nonlinear seismic inversion problems. The proposed is development of the existing MMDE-Net by introducing learnable strategy choosing problem-dependent basis vectors and regularization parameters that are considered to be fixed in MMDE-Net. We name as optimized (OMMDE-Net) investigate its performance through both synthetic field data examples. experimental results demonstrate OMMDE-Net has advantages over effectiveness efficiency.

The inadequate sampling of seismic data in the spatial dimension results migration artifacts. Conventional least-squares reverse time (LSRTM) could improve image quality. However, even LSRTM will not work some inadequately situations. To mitigate impact artifacts, we have developed a new method with sparse regularization, which takes advantage effective representation subsurface reflectivity model 2D undecimated wavelet transform (UWT) domain. Different from other regularizations, sparseness...