Geological exploration data trend prediction system and method based on multi-task learning

Abstract

This invention discloses a geological exploration data trend prediction system and method based on multi-task learning, belonging to the field of geological exploration data trend prediction technology. The invention includes: S10: obtaining the monitoring areas corresponding to various geological monitoring objects in the marked and re-rendered initial 3D geological exploration model; S20: predicting the real-time anomaly degree of each monitoring area; S30: constructing an anomaly trend relationship model for each monitoring area; S40: analyzing the geological early warning time of the target area. Based on the marking and re-rendering results, this invention can not only analyze new geological problems but also analyze secondary changes in existing geological changes, and determine the degree of secondary changes based on the re-rendering degree. This improves the system's analysis effect on geological exploration data. Furthermore, when analyzing geological change trends, it focuses on point and line analysis, narrowing the analysis scope and thus improving the system's analysis efficiency.

Description

Technical Field

[0001] This invention relates to the field of geological exploration data trend prediction technology, specifically a geological exploration data trend prediction system and method based on multi-task learning. Background Technology

[0002] Geological exploration is an important means of understanding the Earth's surface and underground structure, and it is of great significance for resource development, environmental protection, engineering construction and other fields.

[0003] Existing technologies focus on analyzing newly emerging geological changes, but tend to overlook secondary changes that have already occurred. Furthermore, existing technologies mainly analyze geological change trends by processing geological change parameters during geological monitoring. This process focuses on surface and volume analysis, which has a large scope and cannot intuitively perceive the actual geological changes, thus leading to a decrease in analytical accuracy. Summary of the Invention

[0004] The purpose of this invention is to provide a geological exploration data trend prediction system and method based on multi-task learning, so as to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for predicting trends in geological exploration data based on multi-task learning, the method comprising:

[0006] S10: Collect real-time geological survey data at various locations in the target area, mark and re-render the initial geological survey 3D model of the target area, and then obtain the monitoring area corresponding to each geological monitoring object in the marked and re-rendered initial geological survey 3D model.

[0007] S20: Based on the number of aberration lines in each monitoring area at adjacent acquisition times, the line length corresponding to each aberration line, and the degree of re-rendering corresponding to each re-rendering area, the real-time aberration degree of each monitoring area is predicted.

[0008] S30: Construct a model of the anomaly trend relationship in each monitoring area;

[0009] S40: Analyze the geological warning time for the target area and issue a warning signal before the geological warning time.

[0010] Furthermore, S10 includes:

[0011] S101: At time T, collect geological exploration data, geological data and soil characteristics at various locations in the target area. The geological exploration data includes topographic elevation, topographic slope, topographic aspect, groundwater level and surface deformation data. Based on the collected information, construct and render the initial three-dimensional geological exploration model of the target area.

[0012] At time T+d, geological exploration data, geological data, and soil characteristics are collected from various locations in the target area. At the same time, a location is selected. If the geological exploration data collected at the selected location differs from the geological exploration data corresponding to the selected location in the initial geological exploration 3D model, the selected location is marked in the initial geological exploration 3D model. The initial geological exploration 3D model is then re-rendered based on the geological data and soil characteristics collected at various locations in the target area at time T+d.

[0013] S102: The initial geological exploration 3D model marked and re-rendered within the time period [T+d, T+2×d] is designated as the first geological exploration 3D model. The monitoring areas corresponding to each geological monitoring object are obtained from the first geological exploration 3D model.

[0014] In the first geological exploration 3D model, the vertex coordinates of each monitoring area and the rendering degree of each re-rendered area within each monitoring area are obtained. The rendering degree value is calculated as follows: (rendered color grayscale value of the re-rendered area - initial rendered color grayscale value of the re-rendered area) / initial rendered color grayscale value of the re-rendered area.

[0015] S103: At time T+2×d, the initial geological exploration three-dimensional model is updated for the first time based on the first geological exploration three-dimensional model;

[0016] Repeat the above operations to obtain the vertex coordinates of each monitoring area and the re-rendering degree of each re-rendered area in the second geological exploration 3D model, where d represents the data acquisition interval of the geological exploration data.

[0017] Furthermore, S20 includes:

[0018] S201: Number each monitoring area. The numbering result is: c=1,2,…,C; C represents the total number of monitoring areas. Select any monitoring area and denote its number as s, s=1,2,…,C;

[0019] The coordinates of the intersection vertices between the selected monitoring area s in the p-th geological exploration 3D model and the selected monitoring area s in the (p-1)-th updated initial geological exploration 3D model are obtained. Then, any one of the obtained intersection vertex coordinates is used to define the intersection vertex coordinates and the vertex coordinates A in the selected monitoring area s of the p-th geological exploration 3D model. ip Connect the segments to obtain line segment L_A. ip If line segment L_A ip and vertex coordinates A ip If all values ​​are within the initial 3D geological exploration model updated in the (p-1)th time, then the vertex coordinates A will be...ip If line segment L_A is placed in the mutated vertex coordinate set Mps, then... ip Not within the selected monitoring area s in the initial geological exploration 3D model updated in the p-1th time, and the line segment L_A ip If the vertex coordinates A do not completely coincide with any boundary of the selected monitoring area s in the initial geological exploration 3D model updated in the (p-1)th time, then the vertex coordinates A will be... ip The coordinates are stored in the aberration vertex coordinate set Nps. The differences between the number of vertex coordinates stored in Nps, the number of vertex coordinates stored in Mps, and the value 1 are calculated to obtain the number F of the first aberration lines existing in the selected monitoring area s in the p-th geological exploration 3D model. Aps1 The number of second variant lines F Aps2 Where i=1,2,…,m represents the number corresponding to the coordinates of each vertex in the selected monitoring area s, m represents the total number of vertex coordinates in the selected monitoring area s, p=1,2,…,q represents the number corresponding to the number of times the initial geological exploration model is re-rendered, and q represents the total number of numbers;

[0020] Existing methods analyze the degree of variation in each monitoring area by comparing the area range of each monitoring area before and after time intervals. However, they cannot analyze the variation within the monitoring area. This scheme analyzes the variation within the selected monitoring area s by using the vertex coordinates stored in the variation vertex coordinate set Mp. It also analyzes the variation between the selected monitoring area s in the p-th geological exploration 3D model and the selected monitoring area s in the p-1-th updated initial geological exploration 3D model by using the vertex coordinates stored in the variation vertex coordinate set Np. By combining the analysis of internal and external variations, the accuracy of the analysis of the variation in the monitoring area is improved.

[0021] S202: For F A(p+1)s1 With F Aps1 The difference f between A(p+1)s1→Aps1 Perform the calculation, if f A(p+1)s1→Aps1 If the line segment is equal to 0, then the line segment formed by the coordinates of the selected intersection vertices and the coordinates of each vertex in the set of variant vertices Nps is matched one-to-one with the line segment formed by the coordinates of the selected intersection vertices and the coordinates of each vertex in the set of variant vertices N(p+1)s. The absolute value of the difference in line segment length between the two successfully matched line segments is obtained, and the maximum value of the absolute value of the difference in line segment length is denoted as G. A(p+1)s1→Aps1 If f A(p+1)s1→Aps1 If the value is greater than 0, then the number of line segments h left over from the matching is... A(p+1)s1 And the maximum length H of the remaining line segment. A(p+1)s1 To obtain;

[0022] Similarly, we can obtain the maximum value g of the absolute value of the difference in line segment lengths. A(p+1)s2→Aps2The number of remaining line segments h to match A(p+1)s2 And the maximum length H of the remaining line segment. A(p+1)s2 ;

[0023] S203: In the second geological exploration three-dimensional model, G was respectively... A(p+1)s1→Aps1 g A(p+1)s2→Aps2 The re-rendering level R corresponding to the re-rendering area where the matching line segment is located. A(p+1)s1→Aps1 r A(p+1)s2→Aps2 and H A(p+1)s1 H A(p+1)s2 The re-rendering level R corresponding to the re-rendering area where the matching legacy line segment is located. A(p+1)s1 r A(p+1)s2 To obtain;

[0024] When f A(p+1)s1→Aps1 When =0, according to W A(p+1)s1 =[1-exp(R A(p+1)s1→Aps1 )]×[1-exp(-G A(p+1)s1→Aps1 The degree of first anomaly is predicted for the selected monitoring area s in the p-th geological exploration three-dimensional model;

[0025] When f A(p+1)s1→Aps1 When >0, according to W A(p+1)s1 =[1-exp(R A(p+1)s1 )]×[1-exp(-H A(p+1)s1 ×h A(p+1)s1 The degree of first anomaly is predicted for the selected monitoring area s in the p-th geological exploration three-dimensional model;

[0026] When f A(p+1)s2→Aps2 When =0, according to W A(p+1)s2 =[1-exp(r A(p+1)s2→Aps2 )]×[1-exp(-g A(p+1)s2→Aps2 The degree of second variation in the selected monitoring area s in the p-th geological exploration three-dimensional model is predicted.

[0027] When f A(p+1)s2→Aps2 When >0, according to W A(p+1)s2 =[1-exp(r A(p+1)s2 )]×[1-exp(-H A(p+1)s2 ×h A(p+1)s2 The degree of second variation in the selected monitoring area s in the p-th geological exploration three-dimensional model is predicted.

[0028] According to W A(p+1)s =W A(p+1)s1 +W A(p+1)s2 The degree of anomaly in the selected monitoring area s in the p-th geological exploration three-dimensional model is predicted.

[0029] Furthermore, S30 includes:

[0030] Using time p×d and the degree of mutation W A(p+1)s The linear model Y=k×X+b was trained using the training data to obtain the variation trend model W between the degree of variation in the selected monitoring area s and time. A(p+1)s =k s ×p×d+b s ;

[0031] By traversing all monitoring areas, a model of the relationship between the degree of variation and the trend of variation over time is obtained for each monitoring area.

[0032] Where, k s b represents the weights in the trend relationship model. s denoted by , k represents the weight of the linear model, b represents the bias of the linear model, X represents the independent variable of the linear model, and Y represents the dependent variable of the linear model.

[0033] Furthermore, S40 includes:

[0034] Based on the anomaly trend model between the degree of anomaly and time in each monitoring area, the weights of each anomaly trend model are summed, and the ratio between the summation result and the value 3 is calculated to obtain the anomaly index k´ of the target area. Based on each anomaly trend model, the summation value Q of the anomaly degree between each monitoring area in the second geological exploration 3D model is calculated. A2 The degree of change Q of the target region at time T+2×d is calculated. A2 / 3;

[0035] For k´ and Q A2 The product of / 3 is k´×(Q A2 / 3) Calculate the geological early warning threshold V for the target area and k´×(Q) A2 The geological early warning time t for the target area is obtained by calculating the ratio between / 3).

[0036] A warning signal is issued before the geological warning time t.

[0037] A geological exploration data trend prediction system based on multi-task learning, the system includes a monitoring area identification and acquisition module, an anomaly degree prediction module, an anomaly trend relationship model construction module, and a geological early warning module;

[0038] The monitoring area identification and acquisition module is used to acquire the monitoring areas corresponding to each geological monitoring object in the marked and re-rendered initial geological exploration 3D model;

[0039] The anomaly prediction module is used to predict the real-time anomaly level of each monitoring area.

[0040] The mutation trend relationship model construction module is used to construct mutation trend relationship models for each monitoring area;

[0041] The geological early warning module is used to analyze the geological early warning time of the target area and issue an early warning signal before the geological early warning time.

[0042] Furthermore, the monitoring area identification module includes an initial geological exploration 3D model construction unit, a marking and re-rendering unit, a monitoring area identification unit, and an information acquisition unit;

[0043] The initial geological exploration 3D model construction unit constructs and renders the initial geological exploration 3D model of the target area based on the geological exploration data, geological data and soil characteristics of each location point in the target area collected in the first collection.

[0044] The marking and re-rendering unit selectively marks and re-renders each location point in the target area based on the geological exploration data, geological data, and soil characteristics of each location point in the target area collected at the next moment.

[0045] The monitoring area identification unit identifies the monitoring area corresponding to each monitoring object in the marked and re-rendered initial geological exploration 3D model based on the marking and re-rendering results of the initial geological exploration 3D model.

[0046] The information acquisition unit updates the initial geological exploration 3D model based on the marking and re-rendering results of the initial geological exploration 3D model. Based on the update results, it determines the vertex coordinates of each identified monitoring area and the re-rendering degree of each re-rendered area within each identified monitoring area.

[0047] Furthermore, the anomaly degree prediction module includes an analysis unit, an anomaly line quantity calculation unit, and an anomaly degree prediction unit;

[0048] The analysis unit analyzes whether the line segments obtained are within the corresponding monitoring area in the initial geological exploration 3D model, and the positional relationship between the endpoints of the obtained line segments and the corresponding monitoring area in the initial geological exploration 3D model, and classifies the endpoints of the obtained line segments.

[0049] The variable line quantity calculation unit obtains the first variable line quantity and the second variable line quantity in each monitoring area of ​​the marked and re-rendered geological exploration 3D model based on the number of endpoints of various line segments.

[0050] The anomaly prediction unit analyzes the number of first and second anomaly lines, and based on the analysis results, predicts the anomaly degree of each monitoring area in the marked and re-rendered three-dimensional geological exploration model.

[0051] Furthermore, the mutation trend relationship model construction module uses time and mutation degree as training data to train the linear model, thereby obtaining the mutation trend relationship model between the mutation degree and time in each monitoring area.

[0052] Furthermore, the geological early warning module obtains the anomaly index of the target area based on the anomaly trend relationship model between the degree of anomaly in each monitoring area and time, analyzes the geological early warning time of the target area in combination with the geological early warning threshold of the target area, and issues an early warning signal before the geological early warning time.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] 1. This invention uses geological exploration data, geological data and soil characteristics collected from various locations in the target area to mark and re-render the initial three-dimensional geological exploration model. Based on the marking and re-rendering results, it can not only analyze new geological problems, but also analyze the secondary changes of existing geological changes, and determine the degree of secondary changes of geological changes according to the degree of re-rendering, thereby improving the system's analysis effect on geological exploration data.

[0055] 2. This invention analyzes the anomaly lines in each monitoring area by analyzing the vertex coordinates and re-rendering level of each monitoring area, thereby enabling the analysis of geological change trends. It focuses on the analysis of points and lines, narrowing the analysis scope and thus improving the system's analysis efficiency.

[0056] 3. This invention conducts geological monitoring of the target area through multiple monitoring objects, and addresses the diverse trends and complex environments of geological changes by constructing a model of the anomaly trends of multiple monitoring objects. At the same time, it can predict the geological early warning time of the target area based on the anomaly trend relationship model, which is conducive to ensuring the construction safety of the target area. Attached Figure Description

[0057] Figure 1 This is a schematic diagram illustrating the workflow of the geological exploration data trend prediction method based on multi-task learning according to the present invention. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] like Figure 1As shown, this invention provides a technical solution for a geological exploration data trend prediction system and method based on multi-task learning. The method for predicting geological exploration data trends based on multi-task learning includes:

[0060] S10: Collect real-time geological survey data at various locations in the target area, mark and re-render the initial geological survey 3D model of the target area, and then obtain the monitoring area corresponding to each geological monitoring object in the marked and re-rendered initial geological survey 3D model.

[0061] S10 includes:

[0062] S101: At time T, collect geological exploration data, geological data and soil characteristics at various locations in the target area. The geological exploration data includes topographic elevation, topographic slope, topographic aspect, groundwater level and surface deformation data. Based on the collected information, construct and render the initial three-dimensional geological exploration model of the target area.

[0063] At time T+d, geological exploration data, geological data, and soil characteristics are collected from various locations in the target area. Simultaneously, a location is randomly selected. If the geological exploration data collected at the selected location differs from the geological exploration data corresponding to the selected location in the initial geological exploration 3D model, the selected location is marked in the initial geological exploration 3D model. Otherwise, no marking is required. The initial geological exploration 3D model is then re-rendered based on the geological data and soil characteristics collected at various locations in the target area at time T+d. Locations with the same geological data / soil characteristics within the target area will have the same rendering effect. Geological data refers to the stability of the rock strata, and soil characteristics refer to the bearing capacity of the soil. The stronger the stability of the rock strata or the stronger the bearing capacity of the soil, the larger the corresponding gray value of the rendering color.

[0064] S102: The initial geological exploration 3D model marked and re-rendered within the time period [T+d, T+2×d] is designated as the first geological exploration 3D model. The monitoring areas corresponding to various geological monitoring objects are obtained in the first geological exploration 3D model. The geological monitoring objects include groundwater level, surface deformation, and geological environment. The monitoring areas corresponding to the geological monitoring objects refer to the groundwater level variation areas, surface deformation areas, and geological environment variation areas existing in the first geological exploration 3D model. The groundwater level variation areas refer to the marked and re-rendered areas in the groundwater level area of ​​the first geological exploration 3D model. The surface deformation areas refer to the marked and re-rendered areas in the surface area of ​​the first geological exploration 3D model. The geological environment variation areas refer to the marked and re-rendered areas in the geological environment area of ​​the first geological exploration 3D model.

[0065] In the first geological exploration 3D model, the vertex coordinates of each monitoring area and the rendering degree of each re-rendered area within each monitoring area are obtained. The rendering degree value is calculated as follows: (rendered color grayscale value of the re-rendered area - initial rendered color grayscale value of the re-rendered area) / initial rendered color grayscale value of the re-rendered area. The initial rendered color refers to the rendering color of the re-rendered area in the initial geological exploration 3D model.

[0066] S103: At time T+2×d, the initial geological exploration three-dimensional model is updated for the first time based on the first geological exploration three-dimensional model;

[0067] Repeat the above operation to obtain the vertex coordinates of each monitoring area and the re-rendering degree of each re-rendered area in the second geological exploration 3D model. Here, d represents the data collection interval of the geological exploration data, and the second geological exploration 3D model refers to the initial geological exploration 3D model marked and re-rendered within the time period [T+2×d, T+3×d].

[0068] S20: Based on the number of aberration lines in each monitoring area at adjacent acquisition times, the line length corresponding to each aberration line, and the degree of re-rendering corresponding to each re-rendering area, the real-time aberration degree of each monitoring area is predicted.

[0069] S20 includes:

[0070] S201: Number each monitoring area. The numbering result is: c=1,2,…,C; C represents the total number of monitoring areas. Select any monitoring area and denote its number as s, s=1,2,…,C;

[0071] The coordinates of the intersection vertices between the selected monitoring area s in the p-th geological exploration 3D model and the selected monitoring area s in the (p-1)-th updated initial geological exploration 3D model are obtained. When p=1, the 0th updated initial geological exploration 3D model refers to the initial geological exploration 3D model constructed and rendered at time T. Any one of the obtained intersection vertex coordinates is selected, and a line segment is used to represent the coordinates A of the selected intersection vertex and the vertex coordinates A in the p-th geological exploration 3D model's selected monitoring area s. ip Connect the segments to obtain line segment L_A. ip If line segment L_A ip and vertex coordinates A ip If all values ​​are within the initial 3D geological exploration model updated in the (p-1)th time, then the vertex coordinates A will be... ip If line segment L_A is placed in the mutated vertex coordinate set Mps, then... ip Not within the selected monitoring area s in the initial geological exploration 3D model updated in the p-1th time, and the line segment L_A ipIf the vertex coordinates A do not completely coincide with any boundary of the selected monitoring area s in the initial geological exploration 3D model updated in the (p-1)th time, then the vertex coordinates A will be... ip The coordinates are stored in the aberration vertex coordinate set Nps. The differences between the number of vertex coordinates stored in Nps, the number of vertex coordinates stored in Mps, and the value 1 are calculated to obtain the number F of the first aberration lines existing in the selected monitoring area s in the p-th geological exploration 3D model. Aps1 The number of second variant lines F Aps2 Where i=1,2,…,m represents the number corresponding to the coordinates of each vertex in the selected monitoring area s, m represents the total number of vertex coordinates in the selected monitoring area s, p=1,2,…,q represents the number corresponding to the number of times the initial geological exploration model is re-rendered, and q represents the total number of numbers;

[0072] S202: For F A(p+1)s1 With F Aps1 The difference f between A(p+1)s1→Aps1 Perform the calculation, if f A(p+1)s1→Aps1 If the coordinates of the selected intersection vertices and the coordinates of each vertex in the variant vertex set Nps are equal to 0, then a one-to-one matching is performed between the line segments formed by the selected intersection vertices and the coordinates of each vertex in the variant vertex set N(p+1)s and the line segments formed by the selected intersection vertices and the coordinates of each vertex in the variant vertex set N(p+1)s. If the pointing directions of the two line segments are the same, the matching is successful; if the pointing directions of the two line segments are different, the matching is unsuccessful. The absolute value of the difference in line segment length between the two successfully matched line segments is obtained, and the maximum value of the absolute value of the difference in line segment length is denoted as G. A(p+1)s1→Aps1 If f A(p+1)s1→Aps1 If the value is greater than 0, then the number of line segments h left over from the matching is... A(p+1)s1 And the maximum length H of the remaining line segment. A(p+1)s1 To obtain;

[0073] For F A(p+1)s2 With F Aps2 The difference f between A(p+1)s2→Aps2 Perform the calculation, if f A(p+1)s2→Aps2 If the coordinates of the selected intersection vertices and the coordinates of each vertex in the variant vertex set Mps are equal to 0, then a one-to-one matching is performed between the line segments formed by the selected intersection vertices and the coordinates of each vertex in the variant vertex set M(p+1)s and the line segments formed by the selected intersection vertices and the coordinates of each vertex in the variant vertex set M(p+1)s. If the pointing directions of the two line segments are consistent, the matching is successful; if the pointing directions of the two line segments are inconsistent, the matching is unsuccessful. The absolute value of the difference in line segment length between the two successfully matched line segments is obtained, and the maximum value of the obtained absolute value of the difference in line segment length is denoted as g. A(p+1)s2→Aps2 If f A(p+1)s2→Aps2 If the value is greater than 0, then the number of line segments h left over from the matching is... A(p+1)s2 And the maximum length H of the remaining line segment. A(p+1)s2 To obtain;

[0074] S203: In the second geological exploration three-dimensional model, G was respectively... A(p+1)s1→Aps1 g A(p+1)s2→Aps2 The re-rendering level R corresponding to the re-rendering area where the matching line segment is located. A(p+1)s1→Aps1 r A(p+1)s2→Aps2 and H A(p+1)s1 H A(p+1)s2 The re-rendering level R corresponding to the re-rendering area where the matching legacy line segment is located. A(p+1)s1 r A(p+1)s2 To obtain;

[0075] When f A(p+1)s1→Aps1 When =0, according to W A(p+1)s1 =[1-exp(R A(p+1)s1→Aps1 )]×[1-exp(-G A(p+1)s1→Aps1 The degree of first anomaly is predicted for the selected monitoring area s in the p-th geological exploration three-dimensional model;

[0076] When f A(p+1)s1→Aps1 When >0, according to W A(p+1)s1 =[1-exp(R A(p+1)s1 )]×[1-exp(-H A(p+1)s1 ×h A(p+1)s1 The degree of first anomaly is predicted for the selected monitoring area s in the p-th geological exploration three-dimensional model;

[0077] When f A(p+1)s2→Aps2 When =0, according to W A(p+1)s2 =[1-exp(r A(p+1)s2→Aps2 )]×[1-exp(-g A(p+1)s2→Aps2 The degree of second variation in the selected monitoring area s in the p-th geological exploration three-dimensional model is predicted.

[0078] When f A(p+1)s2→Aps2 When >0, according to W A(p+1)s2 =[1-exp(r A(p+1)s2 )]×[1-exp(-H A(p+1)s2 ×h A(p+1)s2 The degree of second variation in the selected monitoring area s in the p-th geological exploration three-dimensional model is predicted.

[0079] According to W A(p+1)s =W A(p+1)s1 +W A(p+1)s2 The degree of anomaly in the selected monitoring area s in the p-th geological exploration three-dimensional model is predicted;

[0080] S30: Construct a model of the anomaly trend relationship in each monitoring area;

[0081] S30 includes:

[0082] Using time p×d and the degree of mutation W A(p+1)s The linear model Y=k×X+b was trained using the training data to obtain the variation trend model W between the degree of variation in the selected monitoring area s and time. A(p+1)s =k s ×p×d+b s ;

[0083] By traversing all monitoring areas, a model of the relationship between the degree of variation and the trend of variation over time is obtained for each monitoring area.

[0084] Where, k s b represents the weights in the trend relationship model. s denoted by , k represents the weight of the linear model, b represents the bias of the linear model, X represents the independent variable of the linear model, and Y represents the dependent variable of the linear model.

[0085] S40: Analyze the geological warning time of the target area and issue a warning signal before the geological warning time;

[0086] S40 includes:

[0087] Based on the anomaly trend model between the degree of anomaly and time in each monitoring area, the weights of each anomaly trend model are summed, and the ratio between the summation result and the value 3 is calculated to obtain the anomaly index k´ of the target area. Based on each anomaly trend model, the summation value Q of the anomaly degree between each monitoring area in the second geological exploration 3D model is calculated. A2 The degree of change Q of the target region at time T+2×d is calculated. A2 / 3;

[0088] For k´ and Q A2 The product of / 3 is k´×(Q A2 / 3) Calculate the geological early warning threshold V for the target area and k´×(Q) A2 The geological early warning time t for the target area is obtained by calculating the ratio between / 3).

[0089] A warning signal is issued before the geological warning time t.

[0090] A geological exploration data trend prediction system based on multi-task learning includes a monitoring area identification and acquisition module, an anomaly degree prediction module, an anomaly trend relationship model construction module, and a geological early warning module.

[0091] The monitoring area identification and acquisition module is used to acquire the monitoring areas corresponding to various geological monitoring objects in the marked and re-rendered initial geological exploration 3D model;

[0092] The monitoring area identification module includes an initial geological exploration 3D model construction unit, a marking and re-rendering unit, a monitoring area identification unit, and an information acquisition unit;

[0093] The initial geological exploration 3D model construction unit constructs and renders the initial geological exploration 3D model of the target area based on the geological exploration data, geological data and soil characteristics of each location point in the target area collected in the first collection.

[0094] The marking and re-rendering unit selectively marks and re-renders each location point in the target area based on the geological exploration data, geological data, and soil characteristics collected at the next time step.

[0095] The monitoring area identification unit identifies the monitoring area corresponding to each monitoring object in the marked and re-rendered initial geological exploration 3D model based on the marking and re-rendering results of the initial geological exploration 3D model;

[0096] The information acquisition unit updates the initial geological exploration 3D model based on the marking and re-rendering results of the initial geological exploration 3D model. Based on the update results, it determines the vertex coordinates of each identified monitoring area and the re-rendering degree of each re-rendered area within each identified monitoring area.

[0097] The anomaly prediction module is used to predict the real-time anomaly level of each monitoring area;

[0098] The anomaly degree prediction module includes an analysis unit, an anomaly line quantity calculation unit, and an anomaly degree prediction unit;

[0099] The analysis unit determines whether the line segments obtained are within the corresponding monitoring area in the initial geological exploration 3D model, and classifies the endpoints of the obtained line segments relative to the corresponding monitoring area in the initial geological exploration 3D model.

[0100] The mutation line quantity calculation unit obtains the number of first and second mutation lines in each monitoring area of ​​the marked and re-rendered geological exploration 3D model based on the number of endpoints of various line segments.

[0101] The anomaly degree prediction unit analyzes the number of first and second anomaly lines, and based on the analysis results, predicts the anomaly degree of each monitoring area in the marked and re-rendered geological exploration 3D model.

[0102] The anomaly trend relationship model building module is used to build an anomaly trend relationship model for each monitoring area;

[0103] The mutation trend relationship model construction module uses time and mutation degree as training data to train the linear model and obtain the mutation trend relationship model between the mutation degree and time in each monitoring area.

[0104] The geological early warning module is used to analyze the geological early warning time of the target area and issue early warning signals before the geological early warning time.

[0105] The geological early warning module obtains the anomaly index of the target area based on the anomaly trend relationship model between the degree of anomaly in each monitoring area and time. Combined with the geological early warning threshold of the target area, it analyzes the geological early warning time of the target area and issues an early warning signal before the geological early warning time.

[0106] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for predicting trends in geological exploration data based on multi-task learning, characterized by: The method includes: S10: Collect real-time geological survey data at various locations in the target area, mark and re-render the initial geological survey 3D model of the target area, and then obtain the monitoring area corresponding to each geological monitoring object in the marked and re-rendered initial geological survey 3D model. S20: Based on the number of aberration lines in each monitoring area at adjacent acquisition times, the line length corresponding to each aberration line, and the degree of re-rendering corresponding to each re-rendering area, the real-time aberration degree of each monitoring area is predicted. S20 includes: S201: Number each monitoring area. The numbering result is: c=1,2,…,C; C represents the total number of monitoring areas. Select any monitoring area and denote its number as s, s=1,2,…,C; The coordinates of the intersection vertices between the selected monitoring area s in the p-th geological exploration 3D model and the selected monitoring area s in the (p-1)-th updated initial geological exploration 3D model are obtained. Then, any one of the obtained intersection vertex coordinates is used to define the intersection vertex coordinates and the vertex coordinates A in the selected monitoring area s of the p-th geological exploration 3D model. ip Connect the segments to obtain line segment L_A. ip If line segment L_A ip and vertex coordinates A ip If all values ​​are within the initial 3D geological exploration model updated in the (p-1)th time, then the vertex coordinates A will be... ip If line segment L_A is placed in the mutated vertex coordinate set Mps, then... ip Not within the selected monitoring area s in the initial geological exploration 3D model updated in the p-1th time, and the line segment L_A ip If the vertex coordinates A do not completely coincide with any boundary of the selected monitoring area s in the initial geological exploration 3D model updated in the (p-1)th time, then the vertex coordinates A will be... ip The coordinates are stored in the aberration vertex coordinate set Nps. The differences between the number of vertex coordinates stored in Nps, the number of vertex coordinates stored in Mps, and the value 1 are calculated to obtain the number F of the first aberration lines existing in the selected monitoring area s in the p-th geological exploration 3D model. Aps1 The number of second variant lines F Aps2 Where i=1,2,…,m represents the number corresponding to the coordinates of each vertex in the selected monitoring area s, m represents the total number of vertex coordinates in the selected monitoring area s, p=1,2,…,q represents the number corresponding to the number of times the initial geological exploration model is re-rendered, and q represents the total number of numbers; S30: Construct a model of the anomaly trend relationship in each monitoring area; S40: Analyze the geological warning time of the target area and issue a warning signal before the geological warning time; S40 includes: Based on the anomaly trend model between the degree of anomaly and time in each monitoring area, the weights of each anomaly trend model are summed, and the ratio between the summation result and the value 3 is calculated to obtain the anomaly index k´ of the target area. Based on each anomaly trend model, the summation value Q of the anomaly degree between each monitoring area in the second geological exploration 3D model is calculated. A2 The degree of change Q of the target region at time T+2×d is calculated. A2 / 3; For k´ and Q A2 The product of / 3 is k´×(Q A2 / 3) Calculate the geological early warning threshold V for the target area and k´×(Q) A2 The geological early warning time t for the target area is obtained by calculating the ratio between / 3). A warning signal is issued before the geological warning time t.

2. The method for predicting geological exploration data trends based on multi-task learning according to claim 1, characterized in that: S10 includes: S101: At time T, collect geological survey data, geological data and soil characteristics at various locations in the target area. The geological survey data includes topographic elevation, topographic slope, topographic aspect, groundwater level and surface deformation data. Based on the collected information, construct and render the initial three-dimensional geological survey model of the target area. At time T+d, geological exploration data, geological data, and soil characteristics are collected from various locations in the target area. At the same time, a location is selected. If the geological exploration data collected at the selected location differs from the geological exploration data corresponding to the selected location in the initial geological exploration 3D model, the selected location is marked in the initial geological exploration 3D model. The initial geological exploration 3D model is then re-rendered based on the geological data and soil characteristics collected at various locations in the target area at time T+d. S102: The initial geological exploration 3D model marked and re-rendered within the time period [T+d, T+2×d] is designated as the first geological exploration 3D model. The monitoring areas corresponding to each geological monitoring object are obtained from the first geological exploration 3D model. In the first geological exploration 3D model, the vertex coordinates of each monitoring area and the rendering degree of each re-rendered area within each monitoring area are obtained. The rendering degree value is calculated as (rendered color grayscale value of the re-rendered area - initial rendered color grayscale value of the re-rendered area) / initial rendered color grayscale value of the re-rendered area. S103: At time T+2×d, the initial geological exploration three-dimensional model is updated for the first time based on the first geological exploration three-dimensional model; Repeat the above operations to obtain the vertex coordinates of each monitoring area and the re-rendering degree of each re-rendered area in the second geological exploration 3D model, where d represents the data acquisition interval of the geological exploration data.

3. The geological exploration data trend prediction method based on multi-task learning according to claim 2, characterized in that: S20 further includes: S202: For F A(p+1)s1 With F Aps1 The difference f between A(p+1)s1→Aps1 Perform the calculation, if f A(p+1)s1→Aps1 If the line segment is equal to 0, then the line segment formed by the coordinates of the selected intersection vertices and the coordinates of each vertex in the set of variant vertices Nps is matched one-to-one with the line segment formed by the coordinates of the selected intersection vertices and the coordinates of each vertex in the set of variant vertices N(p+1)s. The absolute value of the difference in line segment length between the two successfully matched line segments is obtained, and the maximum value of the absolute value of the difference in line segment length is denoted as G. A(p+1)s1→Aps1 If f A(p+1)s1→Aps1 If the value is greater than 0, then the number of line segments h left over from the matching is... A(p+1)s1 And the maximum length H of the remaining line segment. A(p+1)s1 To obtain; Similarly, we can obtain the maximum value g of the absolute value of the difference in line segment lengths. A(p+1)s2→Aps2 The number of remaining line segments h to match A(p+1)s2 And the maximum length H of the remaining line segment. A(p+1)s2 ; S203: In the second geological exploration three-dimensional model, G was respectively... A(p+1)s1→Aps1 g A(p+1)s2→Aps2 The re-rendering level R corresponding to the re-rendering area where the matching line segment is located. A(p+1)s1→Aps1 r A(p+1)s2→Aps2 and H A(p+1)s1 H A(p+1)s2 The re-rendering level R corresponding to the re-rendering area where the matching legacy line segment is located. A(p+1)s1 r A(p+1)s2 To obtain; When f A(p+1)s1→Aps1 When =0, according to W A(p+1)s1 =[1-exp(R A(p+1)s1→Aps1 )]×[1-exp(-G A(p+1)s1→Aps1 The degree of first anomaly is predicted for the selected monitoring area s in the p-th geological exploration three-dimensional model; When f A(p+1)s1→Aps1 When >0, according to W A(p+1)s1 =[1-exp(R A(p+1)s1 )]×[1-exp(-H A(p+1)s1 ×h A(p+1)s1 The degree of first anomaly is predicted for the selected monitoring area s in the p-th geological exploration three-dimensional model; When f A(p+1)s2→Aps2 When =0, according to W A(p+1)s2 =[1-exp(r A(p+1)s2→Aps2 )]×[1-exp(-g A(p+1)s2→Aps2 The degree of second variation in the selected monitoring area s in the p-th geological exploration three-dimensional model is predicted. When f A(p+1)s2→Aps2 When >0, according to W A(p+1)s2 =[1-exp(r A(p+1)s2 )]×[1-exp(-H A(p+1)s2 ×h A(p+1)s2 The degree of second variation in the selected monitoring area s in the p-th geological exploration three-dimensional model is predicted. According to W A(p+1)s =W A(p+1)s1 +W A(p+1)s2 The degree of anomaly in the selected monitoring area s in the p-th geological exploration three-dimensional model is predicted.

4. The geological exploration data trend prediction method based on multi-task learning according to claim 3, characterized in that: S30 includes: Using time p×d and the degree of mutation W A(p+1)s The linear model Y=k×X+b was trained using the training data to obtain the variation trend model W between the degree of variation in the selected monitoring area s and time. A(p+1)s =k s ×p×d+b s ; By traversing all monitoring areas, a model of the relationship between the degree of variation and the trend of variation over time is obtained for each monitoring area. Where, k s b represents the weights in the trend relationship model. s denoted by , k represents the weight of the linear model, b represents the bias of the linear model, X represents the independent variable of the linear model, and Y represents the dependent variable of the linear model.

5. A geological exploration data trend prediction system based on multi-task learning, applied to the geological exploration data trend prediction method based on multi-task learning as described in any one of claims 1-4, characterized in that: The system includes a monitoring area identification and acquisition module, an anomaly degree prediction module, an anomaly trend relationship model construction module, and a geological early warning module. The monitoring area identification and acquisition module is used to acquire the monitoring areas corresponding to each geological monitoring object in the marked and re-rendered initial geological exploration 3D model; The anomaly prediction module is used to predict the real-time anomaly level of each monitoring area. The mutation trend relationship model construction module is used to construct mutation trend relationship models for each monitoring area; The geological early warning module is used to analyze the geological early warning time of the target area and issue an early warning signal before the geological early warning time.

6. The geological exploration data trend prediction system based on multi-task learning according to claim 5, characterized in that: The monitoring area identification module includes an initial geological exploration 3D model construction unit, a marking and re-rendering unit, a monitoring area identification unit, and an information acquisition unit; The initial geological exploration 3D model construction unit constructs and renders the initial geological exploration 3D model of the target area based on the geological exploration data, geological data and soil characteristics of each location point in the target area collected in the first collection. The marking and re-rendering unit selectively marks and re-renders each location point in the target area based on the geological exploration data, geological data, and soil characteristics of each location point in the target area collected at the next moment. The monitoring area identification unit identifies the monitoring area corresponding to each monitoring object in the marked and re-rendered initial geological exploration 3D model based on the marking and re-rendering results of the initial geological exploration 3D model. The information acquisition unit updates the initial geological exploration 3D model based on the marking and re-rendering results of the initial geological exploration 3D model. Based on the update results, it determines the vertex coordinates of each identified monitoring area and the re-rendering degree of each re-rendered area within each identified monitoring area.

7. The geological exploration data trend prediction system based on multi-task learning according to claim 6, characterized in that: The mutation degree prediction module includes an analysis unit, a mutation line quantity calculation unit, and a mutation degree prediction unit. The analysis unit analyzes whether the line segments obtained are within the corresponding monitoring area in the initial geological exploration 3D model, and the positional relationship between the endpoints of the obtained line segments and the corresponding monitoring area in the initial geological exploration 3D model, and classifies the endpoints of the obtained line segments. The variable line quantity calculation unit obtains the first variable line quantity and the second variable line quantity in each monitoring area of ​​the marked and re-rendered geological exploration 3D model based on the number of endpoints of various line segments. The anomaly prediction unit analyzes the number of the first and second anomaly lines, and based on the analysis results, predicts the anomaly degree of each monitoring area in the marked and re-rendered three-dimensional geological exploration model.

8. The geological exploration data trend prediction system based on multi-task learning according to claim 7, characterized in that: The mutation trend relationship model construction module uses time and mutation degree as training data to train the linear model, thereby obtaining the mutation trend relationship model between the mutation degree and time in each monitoring area.

9. The geological exploration data trend prediction system based on multi-task learning according to claim 8, characterized in that: The geological early warning module obtains the anomaly index of the target area based on the anomaly trend relationship model between the degree of anomaly in each monitoring area and time, analyzes the geological early warning time of the target area in combination with the geological early warning threshold of the target area, and issues an early warning signal before the geological early warning time.

Images