Fine urban geology three-dimensional model construction method and system and medium

Abstract

The present application relates to the technical field of image processing, and proposes a refined urban geology three-dimensional model construction method, system and medium, comprising: collecting data of a large number of sampling points in the city through multi-source data, and recording the detection data of multiple attributes at different depths of each sampling point; stratigraphic division is carried out based on depth to obtain the confidence weight of each stratum; the data correlation of two attributes in the same attribute of each sampling point is obtained; the framework correlation of each attribute is obtained; the framework sampling point of each attribute is obtained; the framework attribute is screened to generate a large-scale grid; the related attribute and its framework sampling point of the framework attribute are obtained to generate a medium-scale grid; the key nodes corresponding to the buildings in the city planning data in the multi-source data are analyzed to generate a small-scale grid; the interpolation grid is fused to construct a refined urban geology three-dimensional model. The present application aims to solve the problem of affecting interpolation accuracy caused by stratigraphic change in the process of grid interpolation construction of urban geology three-dimensional model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a fine urban geology three-dimensional model construction method, system and medium. BACKGROUND

[0002] Urban geology three-dimensional model construction is a complex process integrating geology, surveying and mapping, computer science and engineering technology, aiming to provide accurate underground space information support for urban planning, construction, management and disaster prevention and mitigation. With the development of urbanization, the development of underground space (such as subway, pipe gallery, underground commercial body) needs geology three-dimensional model to provide scientific basis for planning decision, so it is necessary to construct fine urban geology three-dimensional model.

[0003] When constructing fine geology three-dimensional model, due to the limited way of obtaining geological data, and the high cost of underground drilling, drilling sampling points are often arranged in the area range of the model to be constructed, and then the range data is obtained by spatial interpolation according to the data of the arranged drilling sampling points, and then the geology three-dimensional model is constructed according to the range data. However, in the traditional spatial interpolation process, due to the diversity of geological layers, the relationship between the layers is complex, and faults, folds and other phenomena may occur. When constructing three-dimensional model, the geological profile needs to be gridded, and the gridding in different ways will also have a great influence on the interpolation result, resulting in low interpolation accuracy, and at the same time, the stratigraphic interface of the geology three-dimensional model is not smooth, which will affect the accuracy of subsequent geology three-dimensional model construction. SUMMARY

[0004] The present application provides a fine urban geology three-dimensional model construction method, system and medium to solve the problem of low interpolation accuracy caused by stratigraphic change in the process of constructing urban geology three-dimensional model by grid interpolation. The technical solution adopted is as follows:

[0005] The present application provides a fine urban geology three-dimensional model construction method, system and medium to solve the problem of low interpolation accuracy caused by stratigraphic change in the process of constructing urban geology three-dimensional model by grid interpolation. The technical solution adopted is as follows:

[0006] A large number of sampling points in the city are collected by multi-source data, and the detection data of different depths of each sampling point are recorded;

[0007] Based on the stratigraphic division of depth in the geological exploration data of multi-source data, the confidence weight of each stratum is obtained by combining the number of detection data of each data attribute in each stratum. Based on the sampling relationship between each attribute, the multi-class attribute is divided, the correlation relationship of the detection data of the same class attribute with the same depth and the same sampling point is analyzed, and the data correlation of two attributes in the same class attribute under each sampling point is obtained. The framework correlation of each attribute is obtained by combining the corresponding relationship between each sampling point and the attribute;

[0008] Based on the data correlation difference of two attributes in the same attribute at different sampling points, the framework sampling points of each attribute are obtained; based on the framework correlation of the attribute, the framework attribute is screened, and the large-scale grid is generated based on the framework sampling points of the framework attribute; the related attributes and their framework sampling points of the framework attribute are obtained by combining the data correlation between the framework attribute and the same attribute at each sampling point, and then the medium-scale grid is generated;

[0009] The key nodes corresponding to the buildings in the urban planning data in the multi-source data are analyzed, and the small-scale grid is generated; the interpolation grid is obtained by combining the medium-scale grid, and the refined urban geological three-dimensional model is constructed by interpolating the interpolation grid.

[0010] Optionally, the confidence weight of each stratum is obtained by the following method:

[0011] In the geological exploration data of multi-source data, each depth corresponds to actual stratum information, and a plurality of strata are divided based on the depth;

[0012] For a plurality of detection data of any attribute, the ratio of the number of detection data of the attribute in any stratum to the total number of detection data of the attribute is obtained, and the ratio is taken as the data amount proportion of the attribute in the stratum.

[0013] The average of the data amount proportions of all attributes in the stratum is obtained, and is taken as the data confidence factor of the stratum; the data confidence factors of all strata are weight normalized, and the obtained result is taken as the confidence weight of each stratum.

[0014] Optionally, the data correlation of two attributes in the same attribute at each sampling point includes the following specific method:

[0015] A plurality of attributes collected at the same time are taken as a type of attribute; for two attributes in any type of attribute at any sampling point, a coordinate system is constructed with depth as horizontal coordinate and detection data as vertical coordinate, and the detection change curve of the two attributes at the sampling point is obtained based on the detection data of the two attributes changing with depth at the sampling point.

[0016] For any stratum, two segments of curves corresponding to the two detection change curves in the stratum are obtained, the Pearson correlation coefficient of the two segments of curves is calculated, and is taken as the correlation coefficient of the two attributes in the stratum at the sampling point.

[0017] Based on the confidence weight of each stratum, the correlation coefficients of the two attributes in the stratum at the sampling point are weighted and summed, and the obtained result is taken as the data correlation of the two attributes in the same attribute at the sampling point.

[0018] Optionally, the framework correlation of each attribute includes the following specific method:

[0019] For two properties in any one type of properties, the average of the data correlation of the two properties at all sampling points of the detection data of the two properties is collected as the change correlation of the two properties; the number of sampling points corresponding to each property in the type of properties is obtained, and the ratio of the number of sampling points corresponding to any property to the maximum value of the number of sampling points corresponding to each property in the type of properties is taken as the sampling weight of the property;

[0020] For any one property in the type of properties, the change correlation of the property and other properties in the type of properties except the property is weighted and averaged according to the sampling weights of the other properties, and the result is taken as the framework correlation of the property.

[0021] Optionally, the framework sampling points of the properties are obtained by the following method:

[0022] For any one sampling point of all sampling points of the detection data of two properties in any one type of properties, the difference between the data correlation of the two properties corresponding to the sampling point and the maximum value of the data correlation of the two properties corresponding to all sampling points is taken as the correlation deviation coefficient of the two properties of the sampling point;

[0023] If the correlation deviation coefficient is greater than a deviation threshold, the sampling point is taken as a framework sampling point of the two properties; all sampling points of the detection data of the two properties are judged to obtain a plurality of framework sampling points;

[0024] For any one property of the two properties, the framework sampling points of the property are obtained from other properties in the type of properties to obtain a plurality of framework sampling points of the property.

[0025] Optionally, the method for generating a large-scale grid comprises the following steps:

[0026] A property with a framework correlation greater than a framework threshold is taken as a framework property; for any framework property, a large-scale grid of the framework property is generated based on the framework sampling points of the framework property and the depth positions corresponding to the detection data of the framework property collected by each framework sampling point.

[0027] Optionally, the method for generating a medium-scale grid comprises the following steps:

[0028] For any one property and any one other property in the same type of properties as the property, if the change correlation of the framework property and the property is greater than or equal to a correlation threshold, the property is taken as a correlation property of the framework property.

[0029] All relevant attributes of the framework attribute and a plurality of framework sampling points of each relevant attribute are acquired, and in the large-scale grid of the framework attribute, a middle-scale grid of the framework attribute is generated in combination with the framework sampling points of each relevant attribute of the framework attribute and each depth position corresponding to the detection data.

[0030] Optionally, the method for generating the small-scale grid comprises the following specific method:

[0031] Based on the urban planning data in the multi-source data, the key positions corresponding to important buildings in the multi-source data are extracted by training a neural network, and key nodes are output; and the small-scale grid is generated based on the key nodes.

[0032] The application further provides a fine urban geological three-dimensional model construction system, which comprises:

[0033] The urban data acquisition module is configured to acquire data of a large number of sampling points in the city by using multi-source data, and record detection data of different depths and different attributes of each sampling point.

[0034] The urban data processing module is configured to divide strata based on the geological exploration data in the multi-source data, acquire confidence weights of each stratum in combination with the number of detection data of each data attribute in each stratum, divide multi-class attributes based on the sampling relationship between each attribute, analyze the correlation between the detection data of the same class of attributes at the same sampling point and with the depth, and obtain the data correlation between two attributes in the same class of attributes at each sampling point; and obtain the framework correlation of each attribute in combination with the corresponding relationship between each sampling point and the attribute.

[0035] The framework sampling points of each attribute are acquired based on the difference in the data correlation between two attributes in the same class of attributes at different sampling points; the framework attributes are screened based on the framework correlation of the attributes; the large-scale grid is generated based on the framework sampling points of the framework attributes; the relevant attributes of the framework attribute and the framework sampling points thereof are acquired in combination with the data correlation between the same class of attributes to which the framework attribute belongs at each sampling point, and the middle-scale grid is further generated.

[0036] The urban model construction module is configured to analyze the key nodes corresponding to buildings in the urban planning data in the multi-source data, and generate a small-scale grid; the interpolation grid is obtained by fusing the middle-scale grid; and the fine urban geological three-dimensional model is constructed by interpolating the interpolation grid.

[0037] The application further provides a fine urban geological three-dimensional model construction medium, which comprises a memory, a processor and a computer program stored in the memory and running on the processor, and the processor implements the steps of the above method when executing the computer program.

[0038] The beneficial effects of the present application are: the interpolation method provided by the present application expects to consider the multi-dimensional representation of the stratum information between the multi-source data, and comprehensively represents the relationship between the strata by using the multi-source data information; referring to the idea of multiple profile multi-pillars in the three-dimensional modeling process, the correlation between two attributes in the same attribute is quantified by detecting the change correlation between the detection data changing with depth under the same sampling point, and then the framework attribute and the corresponding framework sampling point are obtained, so as to expect that the framework attribute can play a prediction role in the interpolation process of other same attributes, and a large-scale grid is generated, and the correlation between the same attributes is further combined to generate a medium-scale grid; meanwhile, the key nodes of the positions of important building facilities in the city planning data are considered, so as to generate a small-scale grid, and the small-scale grid is fused with the medium-scale grid to construct an interpolation grid; interpolation under a reasonable interpolation framework can ensure the accuracy of the interpolation result, and the result of interpolation is affected by gridding, if gridding analysis is performed under a reasonable interpolation framework, the grid will be arranged according to the interpolation framework, and the difference between different grid forms can also be used as the basis for interpolation, so that accurate data interpolation results are obtained by fusing grids of different scales, the accuracy of the city geological three-dimensional model is ensured, and a scientific basis is provided for subsequent city planning decision-making. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0040] Figure 1 The flow chart of the fine city geological three-dimensional model construction method provided by an embodiment of the present application is shown in the figure.

[0041] Figure 2 The system structure block diagram of the fine city geological three-dimensional model construction system provided by another embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0042] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0043] Please refer to Figure 1Fig. 1 shows a flow chart of a method for constructing a refined urban geological three-dimensional model according to an embodiment of the present application, which comprises the following steps:

[0044] In step S001, data of a large number of sampling points in the city are collected through multi-source data, and detection data of different depths and multi-attributes of each sampling point are recorded.

[0045] The purpose of this embodiment is to construct a stratum profile grid based on multi-source data and to achieve refined urban geological three-dimensional model construction through interpolation. Therefore, multi-source data need to be collected to extract detection data of a large number of sampling points and their multi-attributes in the city.

[0046] Specifically, the collected multi-source data include geological exploration data (such as drilling data, geophysical exploration data (seismic wave, resistivity), etc.), remote sensing and topographic data (such as LiDAR point cloud, high-resolution satellite image, DEM / DSM, etc.), engineering investigation data (such as real-time geological records in foundation pit, tunnel, pile foundation construction), geological profile data, and other data such as underground water level, ground subsidence InSAR monitoring, urban planning data, etc.; all the data are uniformly coded, and all the data are in a unified coordinate system, all the data are cleaned and denoised, and data of the same attribute are unified in dimension; three-dimensional geological profiles are generated according to the obtained two-dimensional geological profile vector data and remote sensing and topographic data DEM, and then a large number of sampling points in the three-dimensional geological profiles and the corresponding depth positions of each sampling point in the geological profiles are obtained, and the data of each sampling point corresponding to several attributes are obtained as detection data of each attribute, which correspond to the depth positions of the corresponding sampling points.

[0047] It should be noted that, considering the multi-dimensional representation of stratum information among multi-source data, the relationship between strata is comprehensively represented by using multi-source data information; meanwhile, during the interpolation process, the grid is first processed according to the arranged drilling sampling points, but the gridization in different ways also has a great influence on the interpolation result; therefore, referring to the idea of multiple profiles and multiple pillars in the three-dimensional modeling process, an interpolation architecture is constructed during the interpolation process, and interpolation under a reasonable interpolation architecture can ensure the accuracy of the interpolation result, and the result of interpolation is affected by gridization, and if gridization analysis is performed under a reasonable interpolation architecture, the grid will be arranged according to the interpolation architecture, and the difference between different grid forms can also be used as the basis for interpolation.

[0048] Step S002, stratigraphic division is performed on the depth based on the geological exploration data in the multi-source data, confidence weights of each stratum are obtained in combination with the number of detection data of each data attribute in each stratum, multi-class attributes are divided based on the sampling relationship between each attribute, the correlation relationship of detection data of the same class attribute at the same sampling point in the same stratum with the depth is analyzed, and the data correlation of two attributes in the same class attribute at each sampling point is obtained. The framework correlation of each attribute is obtained in combination with the corresponding relationship between each sampling point and the attribute.

[0049] It should be noted that the stratum relationship exists in the geological model itself, each stratum contains different geological information, for example, different strata have different lithofacies and different hydrological information, therefore, the change trend of the attribute in the depth of different strata is different, and before the model framework is constructed, stratified calculation needs to be performed to obtain the framework in each layer. Meanwhile, due to the geological stratification characteristics, shallow data is easy to collect, but the shallow data is greatly affected by human factors, the middle layer often has a fault zone and anisotropy, and deep data is less but the stratum property is relatively stable.

[0050] Preferably, in an embodiment of the present application, stratigraphic division is performed on the depth based on the geological exploration data in the multi-source data, confidence weights of each stratum are obtained in combination with the number of detection data of each data attribute in each stratum, and the specific method comprises:

[0051] In the geological exploration data of the multi-source data, each depth corresponds to actual stratum information, a plurality of strata are divided based on the depth, for example, the Quaternary loose layer in the collected urban geological data is 10-50 m; for a plurality of detection data of any attribute, the ratio of the number of detection data of the attribute in any stratum to the number of all detection data of the attribute (for all sampling points) is obtained, and the ratio is taken as the data amount proportion of the attribute in the stratum; the average of the data amount proportions of all attributes in the stratum is obtained and taken as the data confidence factor of the stratum; the data confidence factors of all strata are weight-normalized, and the obtained result is taken as the confidence weight of each stratum.

[0052] It should be noted that the more the data quantity in the stratum is, the greater the data amount proportion is, the greater the confidence of the data of the stratum for constructing the interpolation framework is, and the greater the confidence weight is.

[0053] It should be further noted that the expected framework position is at the position where the change occurs, that is, the position where the fault may occur, and in the process of setting the framework, the attribute problem of the framework needs to be considered, and the expected attribute needs to have a certain predictability, that is, the data attribute of other strata can be predicted by interpolation according to the data attribute, and the interpolation is provided with a basis.

[0054] Preferably, in one embodiment of the present application, the multi-class attributes are divided based on the sampling relationship between each attribute, the correlation relationship between the detection data of the same class attributes at the same sampling point and the same stratum with depth is analyzed, the data correlation of two attributes in the same class attribute at each sampling point is obtained, and the specific method includes:

[0055] The simultaneously collected multiple attributes are taken as a class of attributes, that is, multiple attributes of sampling time and sampling time interval, and it needs to be noted that the same class of attributes does not necessarily need to be collected at the same sampling point, for example, the sampling relationship of soil moisture content and soil conductivity is the same, but there is a sampling point that only collects soil moisture content or soil conductivity, and then multiple class attributes are obtained.

[0056] Further, for any two attributes in any class of attributes at any sampling point, a coordinate system is constructed with depth as the horizontal coordinate and detection data as the vertical coordinate, the detection change curve of the two attributes at the sampling point is obtained based on the detection data of the two attributes changing with depth at the sampling point; for any stratum, two corresponding curves in the two detection change curves are obtained, the Pearson correlation coefficient of the two curves is calculated as the correlation coefficient of the two attributes at the sampling point in the stratum; based on the confidence weight of each stratum, the correlation coefficients of the two attributes at the sampling point in each stratum are weighted and summed, and the result is taken as the data correlation of the two attributes in the class of attributes at the sampling point.

[0057] It needs to be noted that based on the change correlation relationship between the detection data of the same class of attributes changing with depth at the same sampling point, the data correlation is preliminarily quantified to provide a basis for subsequent comprehensive acquisition of the correlation and to provide a basis for subsequent acquisition of the attribute and its related attributes.

[0058] Preferably, in one embodiment of the present application, the correlation of the framework of each attribute is obtained based on the corresponding relationship between each sampling point and the attribute, and the specific method includes:

[0059] For any two attributes in a class of attributes, the average of the data correlation of the two attributes at all sampling points where the detection data of the two attributes are collected is taken as the change correlation of the two attributes; the number of sampling points corresponding to each attribute in the class of attributes, that is, the number of sampling points where the detection data of any attribute is collected, is obtained, and the ratio of the number of sampling points corresponding to any attribute to the maximum value of the number of sampling points corresponding to the attribute in the class of attributes is taken as the sampling weight of the attribute.

[0060] Further, for any attribute in the class of attributes, the sampling weights of the other attributes in the class of attributes except the attribute are used to weight and average the change correlation of the attribute and the other attributes in the class of attributes except the attribute, and the result is taken as the framework correlation of the attribute.

[0061] It is to be noted that the more the corresponding sampling points, the greater the corresponding detection data, and the more the quantification of the correlation of the same type of attributes, thereby reducing the influence of the contingency on the analysis of the correlation of the same type of attributes.

[0062] Thus, the data correlation of two attributes in the same type of attributes at each sampling point and the framework correlation of each attribute are obtained.

[0063] In step S003, the framework sampling points of each attribute are obtained based on the difference in the data correlation of two attributes in the same type of attributes at different sampling points, the framework attribute is screened based on the framework correlation of the attribute, the large-scale grid is generated based on the framework sampling points of the framework attribute, the related attribute of the framework attribute and the framework sampling points thereof are obtained by combining the data correlation between the framework attribute and the same type of attributes at each sampling point, and then the medium-scale grid is generated.

[0064] Preferably, in an embodiment of the present application, the framework sampling points of each attribute are obtained based on the difference in the data correlation of two attributes in the same type of attributes at different sampling points, and the specific method comprises the following steps:

[0065] For two attributes in any type of attribute and any sampling point of all sampling points of the detection data of the two attributes, the data correlation of the two attributes corresponding to the sampling point is obtained, the difference absolute value between the data correlation of the two attributes corresponding to all sampling points and the maximum value of the data correlation of the two attributes is obtained, the difference absolute value is taken as the correlation deviation coefficient of the two attributes of the sampling point, a preset deviation threshold is set, the deviation threshold is 0.4 in this embodiment, if the correlation deviation coefficient is greater than the deviation threshold, the sampling point is taken as the framework sampling point of the two attributes, the all sampling points of the detection data of the two attributes are judged according to the above method, and a plurality of framework sampling points are obtained, any one attribute in the two attributes is judged according to the above method, and the framework sampling points of the other attributes in the same type of attribute are obtained, and a plurality of framework sampling points of the attribute are obtained.

[0066] It is to be noted that the framework sampling points are obtained based on two attributes, and the framework sampling points of two attributes are obtained, and the framework sampling points of each attribute include the framework sampling points of the corresponding attribute and all other attributes in the same type of attribute.

[0067] It is to be further noted that the framework sampling point is the sampling point at which the data correlation of the same type of attribute changes, which is more likely to correspond to the change position of the stratum such as fault and fold, and therefore the framework process needs to be considered first, and the correlation between the corresponding attribute and other attributes is reflected by the framework correlation, and the greater the framework correlation, the greater the accuracy of the interpolation of the corresponding attribute for predicting the interpolation of other attributes, so as to obtain the framework attribute.

[0068] Preferably, in one embodiment of the present application, the framework correlation based on the attribute filters the framework attribute, generates the large-scale grid based on the framework attribute sampling point of the framework attribute, and the specific method comprises:

[0069] The preset framework threshold is 0.65 in this embodiment, and the attribute with a framework correlation greater than the framework threshold is regarded as a framework attribute. For any framework attribute, the large-scale grid of the framework attribute is generated based on the framework attribute sampling point of the framework attribute and the depth position corresponding to the detection data of the framework attribute collected by each framework sampling point, that is, the corresponding large-scale grid is generated based on the framework attribute sampling point and the corresponding sampling depth.

[0070] It should be further explained that the large-scale grid is generated based on the framework attribute sampling point, and the construction of the medium-scale grid further needs to combine the related attributes of the framework attribute. Therefore, the medium-scale grid is further generated based on the framework sampling point of the related attribute.

[0071] Preferably, in one embodiment of the present application, the related attribute and the framework sampling point of the framework attribute are obtained by combining the data correlation of the framework attribute between the same attributes at each sampling point, and then the medium-scale grid is generated, and the specific method comprises:

[0072] The preset related threshold is 0.6 in this embodiment. For any framework attribute and any other attribute in the same attribute, if the change correlation between the framework attribute and the attribute is greater than or equal to the related threshold, the attribute is regarded as the related attribute of the framework attribute. All related attributes (obtained based on other attributes in the same attribute) of the framework attribute and the framework sampling points of each related attribute are obtained, and the medium-scale grid of the framework attribute is generated in the large-scale grid of the framework attribute by combining the framework sampling points of each related attribute of each related attribute and the depth positions corresponding to the detection data.

[0073] At this point, a plurality of framework attributes and medium-scale grids are obtained.

[0074] Step S004, analyze the key nodes corresponding to the building in the urban planning data in the multi-source data, generate a small-scale grid, fuse the small-scale grid with the medium-scale grid to obtain an interpolation grid, and construct a refined urban geological three-dimensional model by interpolating the interpolation grid.

[0075] It should be noted that the small-scale represents a network that requires precision, and there are some key positions of urban planning in the urban geological three-dimensional model, such as landmark buildings, parks and gardens, large high-rise buildings, and transportation hubs. These key positions need to ensure the accuracy of the geological monitoring results, so the key nodes corresponding to the key positions need to be determined according to the urban planning map.

[0076] Preferably, in one embodiment of the present application, the key nodes corresponding to the buildings in the urban planning data in the multi-source data are analyzed, and a small-scale grid is generated, including the specific method:

[0077] Based on the urban planning data in the multi-source data, various buildings in the urban planning data are classified by experts according to building time, building size, and building type as classification standards of the buildings, and are respectively classified as key nodes and basic nodes. The key nodes are key positions corresponding to important buildings, which are determined by experts. A cross-entropy loss function is used to train a neural network model to classify the basic nodes and the key nodes in the urban planning data, and then the urban planning data is input into the trained neural network model to output a plurality of key nodes, and a small-scale grid is generated based on the key nodes.

[0078] Preferably, in one embodiment of the present application, the interpolation grid is obtained by combining the mesoscale grid, and a refined urban geological three-dimensional model is constructed by interpolating the interpolation grid, including the specific method:

[0079] The mesoscale grid of each framework attribute and the small-scale grid generated based on the key nodes are fused as the interpolation grid. Based on the detection data of each framework attribute, the existing interpolation method is used to interpolate in the interpolation grid, and then all data attributes are interpolated in the interpolation grid. The existing interpolation method is used, and the interpolation result of the interpolation grid is the detection data and interpolation data of each attribute in different stratigraphic sections.

[0080] Further, the detection data and interpolation data of each attribute in different stratigraphic sections are generated as SHP vector profile data. The data is converted into an irregular triangle network using the 3D Analyst Tools to TIN module in the ArcGIS software, and each stratigraphic unit data is generated in the order of stratigraphic sequence using the module. The stratigraphic profile data is the corresponding interpolation result, and the conversion of the two-dimensional profile data into a three-dimensional model is a known method, which will not be repeated in this embodiment. The specific implementation can be determined according to the specific implementation, and the data interpolation method is limited in this embodiment. Based on the interpolation result of the interpolation grid, the construction of the refined urban geological three-dimensional model is completed.

[0081] Thus, the embodiment is completed.

[0082] Please refer to Figure 2 which shows a refined urban geological three-dimensional model construction system provided by another embodiment of the present application. The system includes:

[0083] The urban data acquisition module 101 acquires data of a large number of sampling points in the city through multi-source data, and records the detection data of multiple attributes at different depths of each sampling point.

[0084] The city data processing module 102: based on the geological exploration data in the multi-source data, the stratum is divided according to the depth, the confidence weight of each stratum is obtained by combining the detection data quantity of each data attribute in each stratum; based on the sampling relationship between each attribute, multi-class attributes are divided, the correlation relationship of the detection data of the same attribute in the same stratum and the same sampling point with the depth is analyzed, and the data correlation of two attributes in the same attribute at each sampling point is obtained; the corresponding relationship between each sampling point and the attribute is combined to obtain the framework correlation of each attribute;

[0085] Based on the data correlation difference of two attributes in the same attribute at different sampling points, the framework sampling point of each attribute is obtained; based on the framework correlation of the attribute, the framework attribute is screened, and the large-scale grid is generated based on the framework sampling point of the framework attribute; the related attribute of the framework attribute and the framework sampling point thereof are obtained by combining the data correlation between the same attribute to which the framework attribute belongs at each sampling point, and then the medium-scale grid is generated;

[0086] The city model construction module 103: analyzing the key nodes corresponding to the building in the city planning data in the multi-source data, generating a small-scale grid; combining the medium-scale grid to obtain an interpolation grid, and constructing a refined city geological three-dimensional model by interpolating the interpolation grid.

[0087] The application also has an embodiment to provide a refined city geological three-dimensional model construction medium, which comprises a memory, a processor and a computer program stored in the memory and running on the processor, and the processor executes the computer program to realize the above method steps S001 to S004.

[0088] The above is only the preferred embodiment of the application, and is not used to limit the application, any modification, equivalent replacement, improvement, etc. within the principle of the application, should be included in the protection scope of the application.

Claims

1. A method for constructing a refined urban geological three-dimensional model, characterized in that, The method comprises the following steps: Collecting data of a large number of sampling points in a city through multi-source data, and recording detection data of multiple attributes at different depths of each sampling point; Dividing strata based on depth in geological exploration data in multi-source data, combining the number of detection data of each data attribute in each stratum to obtain the confidence weight of each stratum; dividing multiple attributes based on the sampling relationship between attributes, analyzing the correlation of detection data of the same attribute with depth in the same stratum and the same sampling point to obtain the data correlation of two attributes in the same attribute at each sampling point; combining the corresponding relationship between each sampling point and attribute to obtain the framework correlation of each attribute; Based on the difference in data correlation of two attributes in the same attribute at different sampling points, the framework sampling points of each attribute are obtained; based on the framework correlation of the attribute, the framework attribute is screened, and the large-scale grid is generated based on the framework sampling points of the framework attribute; combining the data correlation between the framework attribute and the same attribute at each sampling point, the related attribute and its framework sampling point of the framework attribute are obtained, and then the medium-scale grid is generated; Analyzing the key nodes corresponding to the building in the city planning data in multi-source data to generate a small-scale grid; combining the medium-scale grid to obtain an interpolation grid, and constructing a refined city geological three-dimensional model by interpolating the interpolation grid; The framework sampling points of each attribute are obtained in the following specific method: The multiple attributes collected at the same time are taken as one attribute; for two attributes in any attribute and any sampling point of all sampling points collecting detection data of the two attributes, the data correlation of the two attributes corresponding to the sampling point is obtained, and the absolute value of the difference between the data correlation and the maximum value of the data correlation of the two attributes corresponding to all sampling points is taken as the correlation deviation coefficient of the two attributes of the sampling point; If the correlation deviation coefficient is greater than the deviation threshold, the sampling point is taken as the framework sampling point of the two attributes; the all sampling points collecting detection data of the two attributes are judged to obtain a plurality of framework sampling points; For any one attribute of the two attributes, the framework sampling points of the other attributes in the same attribute are obtained to obtain a plurality of framework sampling points of the attribute. 2.The method according to claim 1, characterized in that, The confidence weight of each stratum is obtained in the following specific method: In the geological exploration data of multi-source data, each depth corresponds to actual stratum information, and a plurality of strata are divided based on depth; For a plurality of detection data of any attribute, the ratio of the number of detection data of the attribute in any stratum to the total number of detection data of the attribute is obtained, and the ratio is taken as the data amount proportion of the attribute in the stratum; The average of the data amount proportions of all attributes in the stratum is obtained and taken as the data confidence factor of the stratum; the data confidence factors of all strata are weight-normalized to obtain the confidence weight of each stratum. 3.The method of claim 1, wherein, The data correlation of two attributes in the same attribute at each sampling point is obtained in the following specific method: For any two attributes in any type of attribute at any sampling point, a coordinate system is constructed with depth as the abscissa and detection data as the ordinate, and a detection change curve of the two attributes at the sampling point is obtained based on the detection data of the two attributes varying with depth at the sampling point; For any stratum, two segments of curves corresponding to the two detection change curves of the stratum are obtained, and a Pearson correlation coefficient of the two segments of curves is calculated as the correlation coefficient of the two attributes of the stratum at the sampling point; The correlation coefficients of the two attributes of each stratum at the sampling point are weighted and summed based on the confidence weights of the strata, and the result is taken as the data correlation of the two attributes in the type of attribute at the sampling point. 4.The method according to claim 3, characterized in that, The specific method for obtaining the framework correlation of each attribute includes: For any two attributes in any type of attribute, the average of the data correlation of the two attributes at all sampling points where the detection data of the two attributes are collected is taken as the change correlation of the two attributes; The number of sampling points corresponding to each attribute in the type of attribute is obtained, and the ratio of the number of sampling points corresponding to any attribute to the maximum number of sampling points corresponding to the type of attribute is taken as the sampling weight of the attribute; For any attribute in the type of attribute, the change correlation of the attribute and other attributes in the type of attribute except the attribute is weighted and averaged based on the sampling weights of the other attributes, and the result is taken as the framework correlation of the attribute. 5.The method according to claim 3, characterized in that, The specific method for generating a large-scale grid includes: Attributes with a framework correlation greater than a framework threshold are taken as framework attributes; For any framework attribute, a large-scale grid of the framework attribute is generated based on the framework sampling points of the framework attribute and the depth positions corresponding to the detection data of the framework attribute collected at each framework sampling point. 6.The method according to claim 3, characterized in that, The specific method for generating a medium-scale grid includes: For any framework attribute and any other attribute in the same type of attribute as the framework attribute, if the change correlation of the framework attribute and the attribute is greater than or equal to a correlation threshold, the attribute is taken as a related attribute of the framework attribute; All related attributes of the framework attribute and a number of framework sampling points of each related attribute are obtained, and a medium-scale grid of the framework attribute is generated in the large-scale grid of the framework attribute in combination with the framework sampling points and the depth positions corresponding to the detection data of each related attribute. 7.The method according to claim 1, characterized in that, The specific method for generating a small-scale grid includes: Based on the urban planning data in the multi-source data, the key positions corresponding to important buildings are extracted by training a neural network, and key nodes are output; and a small-scale grid is generated based on the key nodes.

8. The system for constructing a refined urban geological three-dimensional model, characterized in that, The system includes: A city data acquisition module for acquiring data of a large number of sampling points in a city through multi-source data, and recording detection data of multiple attributes at different depths of each sampling point; The urban data processing module is configured to divide strata based on the depth of geological exploration data in the multi-source data, obtain confidence weights of the strata based on the number of detection data of each data attribute in each stratum, divide multi-class attributes based on the sampling relationship between the attributes, analyze the correlation between the detection data of the same-class attributes at the same sampling point and the same stratum with respect to the depth, and obtain the data correlation between two attributes in the same-class attributes at each sampling point; and obtain the framework correlation of each attribute based on the corresponding relationship between each sampling point and the attribute. Based on the differences in the data correlation between two attributes in the same-class attributes at different sampling points, the framework sampling points of each attribute are obtained; based on the framework correlation of the attributes, the framework attributes are screened, and the large-scale grid is generated based on the framework sampling points of the framework attributes; based on the data correlation between the framework attributes and the related attributes thereof at each sampling point, the related attributes and the framework sampling points thereof are obtained, and the medium-scale grid is generated. The framework sampling points of each attribute are obtained in the following manner: a plurality of attributes collected at the same time are taken as one class of attributes; for two attributes in any class of attributes and any sampling point of all sampling points at which the detection data of the two attributes are collected, the data correlation of the two attributes corresponding to the sampling point is obtained, and the absolute value of the difference between the data correlation of the two attributes corresponding to the sampling point and the maximum value of the data correlation of the two attributes corresponding to all sampling points is taken as the correlation deviation coefficient of the two attributes at the sampling point; if the correlation deviation coefficient is greater than a deviation threshold, the sampling point is taken as the framework sampling point of the two attributes; the all sampling points at which the detection data of the two attributes are collected are judged to obtain a plurality of framework sampling points; for any one of the two attributes and other attributes in the class of attributes, the framework sampling points are obtained to obtain a plurality of framework sampling points of the attribute; The urban model construction module is configured to analyze key nodes corresponding to buildings in the urban planning data in the multi-source data, generate a small-scale grid, fuse the small-scale grid and the medium-scale grid to obtain an interpolation grid, and construct a refined urban geological three-dimensional model by interpolating the interpolation grid. 9.A medium for constructing a refined urban geology three-dimensional model, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, and characterized in that, The processor executes the computer program to implement the steps of the method for constructing the refined urban geological three-dimensional model according to any one of claims 1-7.

Images