Multi-scale Grids Research Articles

MuSic-UDF: Learning Multi-Scale dynamic grid representation for high-fidelity surface reconstruction from point clouds

Surface reconstruction for point clouds is a central task in 3D modeling. Recently, the attractive approaches solve this problem by learning neural implicit representations, e.g., unsigned distance functions (UDFs), from point clouds, which have achieved good performance. However, the existing UDF-based methods still struggle to recover the local geometrical details. One of the difficulties arises from the used inflexible representations, which is hard to capture the local high-fidelity geometry details. In this paper, we propose a novel neural implicit representation, named MuSic-UDF, which leverages Multi-Scale dynamic grids for high-fidelity and flexible surface reconstruction from raw point clouds with arbitrary typologies. Specifically, we initialize a hierarchical voxel grid where each grid point stores a learnable 3D coordinate. Then, we optimize these grids such that different levels of geometry structures can be captured adaptively. To further explore the geometry details, we introduce a frequency encoding strategy to hierarchically encode these coordinates. MuSic-UDF does not require any supervisions like ground truth distance values or point normals. We conduct comprehensive experiments under widely-used benchmarks, where the results demonstrate the superior performance of our proposed method compared to the state-of-the-art methods.

Multi-scale coupled attention for visual object detection.

The application of deep neural network has achieved remarkable success in object detection. However, the network structures should be still evolved consistently and tuned finely to acquire better performance. This gears to the continuous demands on high performance in those complex scenes, where multi-scale objects to be detected are located here and there. To this end, this paper proposes a network structure called Multi-Scale Coupled Attention (MSCA) under the framework of self-attention learning with methodologies of importance assessment. Architecturally, it consists of a Multi-Scale Coupled Channel Attention (MSCCA) module, and a Multi-Scale Coupled Spatial Attention (MSCSA) module. Specifically, the MSCCA module is developed to achieve the goal of self-attention learning linearly on the multi-scale channels. In parallel, the MSCSA module is constructed to achieve this goal nonlinearly on the multi-scale spatial grids. The MSCCA and MSSCA modules can be connected together into a sequence, which can be used as a plugin to develop end-to-end learning models for object detection. Finally, our proposed network is compared on two public datasets with 13 classical or state-of-the-art models, including the Faster R-CNN, Cascade R-CNN, RetinaNet, SSD, PP-YOLO, YOLO v3, YOLO v5, YOLO v7, YOLOX, DETR, conditional DETR, UP-DETR and FP-DETR. Comparative experimental results with numerical scores, the ablation study, and the performance behaviour all demonstrate the effectiveness of our proposed model.

A general modeling scheme for spatiotemporal DGGS with emphasis on encoding and operating multiscale time grids

AbstractOne of the basic scientific problems concerning geographic information science is how to rapidly organize, query, and compute spatiotemporal big data. The spatiotemporal discrete global grid system (DGGS) provides a homogenized discrete structure for processing multiscale and multitype spatiotemporal data. To date, most research in spatiotemporal DGGS has focused on spatial discretization while neglecting temporal discretization. Here, we propose a general modeling scheme for spatiotemporal DGGS with emphasis on encoding and operating multiscale time grids. We subdivide continuous time into multiscale temporal grids, which are then encoded as integers. Moreover, we designed integer code operations, including hierarchical traversal, neighborhood finding, and temporal relationship calculations. Compared to the multiscale time segment integer coding (MTSIC) approach, the proposed method resulted in 22% higher encoding efficiency, 10.92 times faster decoding, 2.81 times better parent code finding efficiency, 41% improved efficiency, 100% accuracy in finding children codes (compared to less than 100% with MTSIC), and a 62% enhancement in temporal relationship calculation efficiency. The application of querying spatiotemporal trajectory data validates the feasibility and practicality of substituting conventional string‐based time and floating‐point location coordinates with spatiotemporal integer codes to query data. The time encoding and operation methods proposed here indicate high efficiency, superior accuracy, and broad application prospects.

BIM Data Model Based on Multi-Scale Grids in Civil Engineering Buildings

The construction of digital twin cities is a current research hotspot; GIS technology and BIM technology are widely used in the field of digital twin cities. However, BIM is still subject to major limitations in its applications, mainly due to huge amounts of model data, low query efficiency and accuracy, non-uniform marking systems, etc. The reason is that the BIM model itself focuses more on the expression of visual effects and lacks spatial calculation ability and the utilization of spatial location information. Secondly, the current lightweight processing methods for BIM models are mostly based on geometric transformation and rendering optimization, focusing more on the data compression and visual quality of the model, which essentially does not change the data structure of the BIM model, and it is difficult to establish the mapping relationship between spatial location and spatial data, information, and resources. In addition, current coding methods proposed for BIM models are mostly based on the line classification method, which realizes the identification of components based on the classification of their attributes, and the location information is stored according to the attributes or natural language descriptions, which need to be parsed and translated when they are used, and this procedure ignores the importance of spatial location in daily management and emergency management. The importance of spatial location in daily management and emergency management is also ignored. Based on this kind of identification code, it is impossible to directly analyze and apply spatial location data. Therefore, this paper takes the combination of GIS technology and BIM technology as the starting point and proposes a BIM data modeling method based on the BeiDou grid code, based on the efficiency of its underlying data organization and the accuracy of its real geographic location expression on the one hand and the completeness of the information expression by BIM and fine three-dimensional visualization on the other hand. Finally, a series of experiments are carried out based on the method. Through visualization modeling and efficiency experiments, different feature models are meshed to verify the feasibility and efficiency of the model. Through coding and information query experiments, the model′s data organization capability, data dynamic carrying capability, and efficient spatial computation capability and practical application capability are verified.

CSRL-Net: contextual self-rasterization learning network with joint weight loss for remote sensing image semantic segmentation

ABSTRACTSemantic segmentation plays a vital role in the intelligent comprehension of remote sensing images (RSIs). However, research on semantic segmentation of RSIs still faces the following challenges: 1) The complexity of ground object structures, including variations in scale and shading environments, poses difficulties for current methods in capturing global context. 2) In long-tail distributed remote sensing data, the scarcity of tail classes makes it difficult for their features to be effectively learned, as features of head classes often overshadow them. To address these issues, we propose a contextual self-rasterization learning network (CSRL-Net) for the semantic segmentation of RSIs. Our approach comprises the following two key components. Firstly, a grid context perception mechanism is developed to collaboratively establish context dependencies within and among multi-scale grids, capturing long-range spatial correlations. Secondly, a joint weight loss function is designed to convert the prior knowledge into weight factors. This loss function combines re-weighting and logit adjustment, giving more attention to tail classes and integrally balancing learning bias. To evaluate the effectiveness of our proposed method, we apply it to the Potsdam, Vaihingen, and GID datasets and compare its performance with current advanced models. Experimental results demonstrate that our method achieves excellent performance in terms of MIoU, mean F1 and OA, with improvements ranging from 0.364% to 1.764% compared to the 17 comparison models. Notably, the proposed joint weight loss significantly improves IoU and F1 for tail classes, resulting in increases of 2.909% (IoU) and 2.085% (F1) on the Vaihingen dataset and 4.697% (IoU) and 4.043% (F1) on the GID dataset.

Landscape Ecological Risk Evaluation Study under Multi-Scale Grids—A Case Study of Bailong River Basin in Gansu Province, China

To solve grid-scale problems and evaluation indicator selection in landscape ecological risk index (LERI) evaluation, this paper takes the Bailong River Basin in Gansu Province (BLRB) as an example. The LERI evaluation formulae and optimal grid scales were determined by screening landscape indices and area changes in the LERI at different grid scales. The evaluation indices were finally obtained according to the landscape characteristics and the correlation analysis of the landscape index value. Through the statistical analysis of the area of the LERI at the grid scale of 1–6 km, the optimal grid scale was determined to be 5 km. There was little change in land use patterns, with the most significant increases in artificial surfaces at 3.29% and 3.58%, respectively. Cultivated land was the only land use type to decrease by 184.3 km2. The LERI drops with the reduced cultivated land area; the landscape ecological medium risk area and cultivated land keep the same spatial distribution. Due to the limitation of the topography, cultivated land is generally distributed below 2500 m altitude, so 2500 m becomes the turning point in the spatial distribution of the LERI. The medium risk below 2500 m dominates the LERI type. Reduced cultivated land was the leading cause of reduced ecological risk according to an overlay analysis. The study of LERI evaluations provides a theoretical basis for sustainable and ecological environmental protection in the BLRB.

Algebraic multiscale grid coarsening using unsupervised machine learning for subsurface flow simulation

Subsurface flow simulation is vital for many geoscience applications, including geoenergy extraction and gas (energy) storage. Reservoirs are often highly heterogeneous and naturally fractured. Therefore, scalable simulation strategies are crucial to enable efficient and reliable operational strategies. One of these scalable methods, which has also been recently deployed in commercial reservoir simulators, is algebraic multiscale (AMS) solvers. AMS, like all multilevel schemes, is found to be highly sensitive to the types (geometries and size) of coarse grids and local basis functions. Commercial simulators benefit from a graph-based partitioner; e.g., METIS to generate the multiscale coarse grids. METIS minimizes the amount of interfaces between coarse partitions, while keeping them of similar size which may not be the requirement to create a coarse grid. In this work, we employ a novel approach to generate the multiscale coarse grids, using unsupervised learning methods which is based on optimizing different parameter. We specifically use the Louvain algorithm and Multi-level Markov clustering. The Louvain algorithm optimizes modularity, a measure of the strength of network division while Markov clustering simulates random walks between the cells to find clusters. It is found that the AMS performance is improved when compared with the existing METIS-based partitioner on several field-scale test cases. This development has the potential to enable reservoir engineers to run ensembles of thousands of detailed models at a much faster rate.

A Post Processing Technique to Automatically Remove Floater Artifacts in Neural Radiance Fields

AbstractNeural Radiance Fields have revolutionized Novel View Synthesis by providing impressive levels of realism. However, in most in‐the‐wild scenes they suffer from floater artifacts that occur due to sparse input images or strong view‐dependent effects. We propose an approach that uses neighborhood based clustering and a consistency metric on NeRF models trained on different scene scales to identify regions that contain floater artifacts based on Instant‐NGPs multiscale occupancy grids. These occupancy grids contain the position of relevant optical densities in the scene. By pruning the regions that we identified as containing floater artifacts, they are omitted during the rendering process, leading to higher quality resulting images. Our approach has no negative runtime implications for the rendering process and does not require retraining of the underlying Multi Layer Perceptron. We show on a qualitative base, that our approach is suited to remove floater artifacts while preserving most of the scenes relevant geometry. Furthermore, we conduct a comparison to state‐of‐the‐art techniques on the Nerfbusters dataset, that was created with measuring the implications of floater artifacts in mind. This comparison shows, that our method outperforms currently available techniques. Our approach does not require additional user input, but can be be used in an interactive manner. In general, the presented approach is applicable to every architecture that uses an explicit representation of a scene's occupancy distribution to accelerate the rendering process.

A parallel geometric multigrid method for adaptive topology optimization

This work presents an efficient parallel geometric multigrid (GMG) implementation for preconditioning Krylov subspace methods solving differential equations using non-conforming meshes for discretization. The approach does not constrain such meshes to the typical multiscale grids used by Cartesian hierarchical grid methods, such as octree-based approaches. It calculates the restriction and interpolation operators for grid transferring between the non-conforming hierarchical meshes of the cycle scheme. Using non-Cartesian grids in topology optimization, we reduce the mesh size discretizing only the design domain and keeping the geometry of boundaries in the final design. We validate the GMG method operating on non-conforming meshes using an adaptive density-based topology optimization method, which coarsens the finite elements dynamically following a weak material estimation criterion. The GMG method requires the generation of the hierarchical non-conforming meshes dynamically from the one used by the adaptive topology optimization to analyze to the one coarsening all the mesh elements until the coarsest level of the mesh hierarchy. We evaluate the performance of the adaptive topology optimization using the GMG preconditioner operating on non-conforming meshes using topology optimization on a fine-conforming mesh as the reference. We also test the strong and weak scaling of the parallel GMG preconditioner with two three-dimensional topology optimization problems using adaptivity, showing the computational advantages of the proposed method.

Parallel Scheme for Multi-Layer Refinement Non-Uniform Grid Lattice Boltzmann Method Based on Load Balancing

The large-scale numerical simulation of complex flows has been an important research area in scientific and engineering computing. The lattice Boltzmann method (LBM) as a mesoscopic method for solving flow field problems has become a relatively new research direction in computational fluid dynamics. The multi-layer grid-refinement strategy deals with different-level of computing complexity through multi-scale grids, which can be used to solve the complex flow field of the non-uniform grid LBM without destroying the parallelism of the standard LBM. It also avoids the inefficiencies and waste of computational resources associated with standard LBMs using uniform and homogeneous Cartesian grids. This paper proposed a multi-layer grid-refinement strategy for LBM and implemented the corresponding parallel algorithm with load balancing. Taking a parallel scheme for two-dimensional non-uniform meshes as an example, this method presented the implementation details of the proposed parallel algorithm, including a partitioning scheme for evaluating the load in a one-dimensional direction and an interpolation scheme based on buffer optimization. Simply by expanding the necessary data transfer of distribution functions and macroscopic quantities for non-uniform grids in different parallel domains, our method could be used to conduct numerical simulations of the flow field problems with complex geometry and achieved good load-balancing results. Among them, the weak scalability performance could be as high as 88.90% in a 16-threaded environment, while the numerical simulation with a specific grid structure still had a parallel efficiency of 77.4% when the parallel domain was expanded to 16 threads.

Numerical visualisation on applying a five-stage multi-scaled square turbulence-generating grid to generating turbulence

This study presents the numerical visualisation of turbulence generated by a five-stage square-type multi-scale turbulence grid. The present study applies a turbulence grid with an increased number of stages to five, in contrast to previous studies using multi-scale turbulence grids with four stages. For the present numerical visualisation, direct numerical analysis is used in this analysis. This analysis uses the high-order central difference schemes and the third-order Runge-Kutta scheme. In addition to the multi-scale turbulence grid with five stages, the present study also covers two turbulence grids with four stages for reference. By using the external force term in the governing equations, the present multi-scale turbulence grids were formed numerically in the computational domain. The spatial homogeneity of the downstream turbulence field is approached in this work. The turbulence generated by the turbulence grids with four stages is characterised by a flow structure with a doughnut-like instantaneous velocity figure in the downstream region. By examining contour diagrams of the streamwise instantaneous velocity, the present study can see that the spatial homogeneity of the turbulence field downstream is improved by increasing the number of stages for the turbulence grid.

A prediction method for spatially decaying freestream turbulence

A simple set of equations, capable of quantifying and predicting the spatial decay of freestream turbulence (FST) is derived in the current study. The prediction equations are based on the inviscid estimate of the turbulent kinetic energy (TKE) dissipation rate. The new set of model equations includes the integral length scale and the turbulent kinetic energy as variables and is superior to the previous set of decay equations because, unlike those, they are not dependent on any physical grid parameters (b or M). This new set of equations, when compared and validated against 17 sets (2 active grids, 2 multi-scale grids, 9 square-cross-sectioned grids and 4 circular cross-sectioned grids) of previous, well-accepted, experimental data, including those relating to grid-generated turbulence and covering a wide range of turbulent Reynolds number (ReLu 0) (7.5 × 101 to 6.9 × 104), where Lu 0 is the initial integral length scale, showed very good agreement (within ±15%). This set of correlation equations can be used to estimate the local and/or initial turbulent kinetic energy and integral length scale (Lu ) in an FST flow and to locate the region within a flow domain where nearly-constant turbulence conditions are expected to prevail.

Data-Driven Selection of Land Product Validation Station Based on Machine Learning

Validation is a crucial technique used to strengthen the application capabilities of earthobservation satellite data and solve the quality problems of remote-sensing products. Observing land-surface parameters in the field is one of the key steps of validation. Therefore, the demand for long-term stable validation stations has gradually increased. However, the current location-selection procedure of validation stations lacks a systematic and objective evaluation system. In this research, a data-driven selection of a land product validation station (DSS-LPV) based on Machine Learning is proposed. Firstly, we construct an evaluation indicator system in which all factors affecting the location of validation stations are divided into surface characteristics, atmospheric conditions and the social environment. Then, multi-scale evaluation grids are constructed and indicators are allocated for spatial evaluation. Finally, four Machine Learning (ML) methods are used to learn the established reliable stations, and different data-driven scoring models are constructed to explore the intrinsic relationship between evaluation indicators and station locations. In this article, the reliability of DSS-LPV is effectively validated by the example of China using the national-level land product validation station that has been established. After a comparison between the four ML models, the random forest (RF) with the highest accuracy was selected as the modeling method of DSS-LPV. The correlation between the regression value of test stations and the target value is 0.9133. The average score of test stations is 0.8304. The test stations are generally located within the calculated hot-spot area of the score density map, which means that it is highly consistent with the location of the built stations. Research results indicate that DSS-LPV is an effective method that can provide a reasonable geographical distribution of the stations. The location-selection results can provide scientific decision-making support for the construction of land product validation stations.

Grid-stretching capability for the GEOS-Chem 13.0.0 atmospheric chemistry model

Abstract. Modeling atmospheric chemistry at fine resolution globally is computationally expensive; the capability to focus on specific geographic regions using a multiscale grid is desirable. Here, we develop, validate, and demonstrate stretched grids in the GEOS-Chem atmospheric chemistry model in its high-performance implementation (GCHP). These multiscale grids are specified at runtime by four parameters that offer users nimble control of the region that is refined and the resolution of the refinement. We validate the stretched-grid simulation versus global cubed-sphere simulations. We demonstrate the operation and flexibility of stretched-grid simulations with two case studies that compare simulated tropospheric NO2 column densities from stretched-grid and cubed-sphere simulations to retrieved column densities from the TROPOspheric Monitoring Instrument (TROPOMI). The first case study uses a stretched grid with a broad refinement covering the contiguous US to produce simulated columns that perform similarly to a C180 (∼ 50 km) cubed-sphere simulation at less than one-ninth the computational expense. The second case study experiments with a large stretch factor for a global stretched-grid simulation with a highly localized refinement with ∼10 km resolution for California. We find that the refinement improves spatial agreement with TROPOMI columns compared to a C90 cubed-sphere simulation of comparable computational demands. Overall, we find that stretched grids in GEOS-Chem are a practical tool for fine-resolution regional- or continental-scale simulations of atmospheric chemistry. Stretched grids are available in GEOS-Chem version 13.0.0.

Direct inversion of surface wave dispersion data with multiple-grid parametrizations and its application to a dense array in Chao Lake, eastern China

SUMMARY The direct surface wave tomography has become an efficient tool in imaging 3-D shallow Earth structure. However, some fundamental problems still exist in selecting the grids to parametrize the model space. This study proposes to implement a model parametrization approach with multiple grids to the direct surface wave tomography. These multiple grids represent several overlapping collocated grids with the same or different grid spacings, such as staggered grids, multiscale grids and multiscale-staggered grids. At each iteration, direct inversion is applied to each individual set of collocated grids to invert for the shear wave velocity (Vs) model; the models are then projected onto a set of predefined base grids (usually the finest grids) using 3-D B-spline interpolation. At the end of each iteration, we average the Vs models of all sets of collocated grids to obtain the average 3-D Vs model, which is then used as the initial model for the next iteration. The properties of this approach are explored by applying it to a newly deployed dense array in Chao Lake (CL), eastern China. Synthetic and field data tests demonstrate that the method using multiple grids recovers anomaly patterns better than that using the individual set of collocated grids, though it does not necessarily achieve the smallest traveltime residual. We then obtain a high-resolution 3-D shallow crustal Vs model beneath the CL. The 3-D Vs model reveals two prominent features: (1) a stripe-like structural pattern of velocity variations, where the Hefei basin and eastern CL display low-velocity anomalies while the Tanlu fault zone (TFZ), Zhangbaling uplift and Yinping mountain present high-velocity anomalies and (2) north-shifted low-velocity anomalies beneath the eastern CL as depths go shallow. The shallow Vs features are consistent well with the local geological units and topography. We suggest that the two main features could be associated with the multistage tectonic activities of the Tanlu fault. The multiple-grid scheme proposed in this study could be conveniently extended to other 3-D direct inversion approaches in the near future.

Surface seismics with DAS: An emerging alternative to modern point-sensor acquisition

Land seismic acquisition is moving toward “light and dense” geometries, with point receiver systems believed to be an ultimate configuration of choice. Cableless land nodal systems enable more flexible spatial sampling at the price of eliminating even small arrays. For large surveys in a desert environment, such spacing remains insufficient to address the complex near surface, while recordings with single sensors exhibit a significant reduction in data quality. At the same time, exploration problems increasingly demand smaller uncertainty in all seismic products. While 1 m geophone sampling could have addressed these problems, it remains out of economic reach as point sensor cost plateaus. We examine an emerging alternative technology of distributed acoustic sensing (DAS) that revolutionized borehole geophysics but is still mostly unknown in the seismic world. Fully broadband DAS sensors promise massive channel count and uncompromised inline sampling down to 0.25 m. Their distributed nature offers the unique capability to conduct a continuous recording with multiscale grids of “shallow,” “deep,” and “full-waveform inversion” receivers, all implemented with a single set of fixed cables and only one round of shooting. These distinct features allow us to simultaneously pursue near-surface characterization, imaging of deeper targets, and velocity model evaluation. Specifically, in a desert environment, distributed sensors may offer superior data quality compared to point sensors, whereas DAS capability of “seismic zoom” in the near surface becomes instrumental for near-surface characterization. Finally, simultaneous acquisition of surface seismic and vertical arrays that can be achieved easily with DAS can effectively address the exploration of subtler targets such as low-relief structures. We support these findings with a field case study from a desert environment and synthetic examples. With many distinct advantages, surface seismic with DAS emerges as a compelling alternative to modern point-sensor acquisitions.

Rapid computation of set boundaries of multi-scale grids and its application in coverage analysis of remote sensing images

With the rapid development of remote sensing technology, the amount of remote sensing data is increasing, service objects are increasingly extensive, and requests from users to query the coverage of remote sensing images under specific conditions have also increased. These conditions have increased efficiency and precision requirements for concurrent access and response. Remote sensing images provide multiple coverages of the same area, and the nested and overlapping relationships among data are complex. Therefore, calculating the range of multiple overlaps of images becomes increasingly difficult as the number of images increases. To query the range of a coverage area of multiple images is to calculate its boundary, which is an important part of spatial overlay analysis. Traditional vector methods are inefficient in processing many spatial objects or complex shapes. The overlay of traditional spatial grids is mostly addressed via single-scale methods, which have low computational efficiency and low boundary fitting precision. Combined with the current gridding management method for multi-source remote sensing images, we proposed a new algorithm for rapid computation of the set boundaries of multi-scale grids and applied it to coverage analysis of remote sensing images. The algorithm can effectively trace the boundaries of multi-scale grids formed in the regions covered by remote sensing images, and it can solve all types of complex boundaries, such as convex and concave boundaries, holes and islands. The experiments in this study show that the new algorithm greatly improves computational efficiency and boundary fitting precision compared with the single-scale grid methods. Compared with the vector algorithms of ArcGIS and other commercial software, this study's algorithm can greatly improve calculation efficiency while ensuring a precision above 99%. The new algorithm is suitable for rapid calculations of large areas and widespread coverage of remote sensing images.

Adaptive dynamic multilevel simulation of fractured geothermal reservoirs

An algebraic dynamic multilevel (ADM) method for fully-coupled simulation of flow and heat transport in heterogeneous fractured geothermal reservoirs is presented. Fractures are modeled explicitly using the projection-based embedded discrete method (pEDFM), which accurately represents fractures with generic conductivity values, from barriers to highly-conductive manifolds. A fully implicit scheme is used to obtain the coupled discrete system including mass and energy balance equations with two main unknowns (i.e., pressure and temperature) at fine-scale level. The ADM method is then developed to map the fine-scale discrete system to a dynamic multilevel coarse grid, independently for matrix and fractures. To obtain the ADM map, multilevel multiscale coarse grids are constructed for matrix as well as for each fracture at all coarsening levels. On this hierarchical nested grids, multilevel multiscale basis functions (for flow and heat) are solved locally at the beginning of the simulation. They are used during the ADM simulation to allow for accurate multilevel systems in presence of parameter heterogeneity. The resolution of ADM simulations is defined dynamically based on the solution gradient (i.e. front tracking technique) using a user-defined threshold. The ADM mapping occurs algebraically using the so-called ADM prolongation and restriction operators, for all unknowns. A variety of 2D and 3D fractured test cases with homogeneous and heterogeneous permeability maps are studied. It is shown that ADM is able to model the coupled mass-heat transport accurately by employing only a fraction of fine-scale grid cells. Therefore, it promises an efficient approach for simulation of large and real-field scale fractured geothermal reservoirs. All software developments of this paper is publicly available at https://gitlab.com/DARSim2simulator.

Conversion of the geological model (fine-mesh) to a dynamic (coarse-mesh) hydrocarbon model with the nature approach in the simulation of thermal recovery in a fractured reservoir

The operation and proper management of reservoirs requires the prediction of reservoir performance. This prediction is generally done by computer simulation. Since simulation software generates static models in the number of millions and even billions, dynamic simulation on these models is difficult and sometimes impossible; therefore, multi-scale block generation is necessary. In this study, we introduce a methodology that was inspired by Earth’s pattern to generate multi-scale grids on a multi-phase, heterogeneous reservoir, where the enhanced oil recovery process is steam injection. From a reservoir engineer’s point of view, Earth is equivalent to a multi-scale grid model that could be a pattern for multi-scale grid generation in a hydrocarbon reservoir to minimize central processing unit time for dynamic simulation. The principle of multi-scale grid generation in hydrocarbon reservoirs is as follows: regions with a high Darcy’s velocity should be fine-scale and other segments with low intensity of heterogeneity regions could resize into the up-scale. The intersection of latitude and longitude lines means that Earth’s model could be a practical pattern for dynamic simulation. This intersection creates segments with fine-size blocks, like the South and North Poles, that are equal to the injection and production wells in the reservoir model and other segments that have different intensities of coarse-size blocks. Earth’s magnetic field, which enters from South and exits from the North Pole, leads us to could consider Earth as a hydrocarbon model. The results of the multi-scale grid generation method, which was inspired by Earth’s pattern, were compared with the fine-mesh (geological) model; the results show that the proposed method predicts with high accuracy and less run-time compared with the fine-mesh model.

A novel lesion detection algorithm based on multi-scale input convolutional neural network model for diabetic retinopathy

Objective To propose a multi-scale convolutional neural network (CNN) based lesions detection method of fundus image, and evaluate its application in diabetic retinopathy (DR) assisted diagnosis. Methods A multi-scale CNN based on lesions detection method of fundus image was proposed.Compared with the existing detection methods, the problem of poor robustness based on threshold segmentation and morphological segmentation was overcome.The idea of multi-scale grids detection without relying on manual pixel-by-pixel labeling was adopted in this algorithm, and the detection performance of small lesions was significantly improved.In addition, multiple DR lesions with high accuracy could be detected by the proposed loss function under the condition of weak labels and small data sets. Results At the level of lesions, the sensitivity and specificity of hard exudation lesions detection were 92.17% and 97.17%, respectively.Compared with single-scale method, the sensitivity and accuracy of multi-scale method proposed in this paper increased by 7.41% and 5.02%, respectively, and compared with other algorithm using the same public dataset IDRiD, the specificity of this algorithm increased by 55.82%.This method could effectively detect the lesions in fundus images, and could give the basic range of the lesions.The average detection time of fundus images with a large number of lesions was 1.59 seconds. Conclusions The DR lesions in the fundus image can be quickly and reliably identified, the location information of the lesions can be marked, and the influence of subjective factors can be reduced by using this algorithm, and it can be used to assist the clinician to conduct more effectively. Key words: Artificial intelligence; Diabetic retinopathy/diagnosis; Fundus color photograph; Multi-scale convolutional neural network