Geological knowledge graph driven mineral exploration decision-making method and system

Abstract

The present application relates to the technical field of intelligent decision-making, in particular to a mineral exploration decision-making method and system driven by a geological knowledge graph, comprising the following steps: collecting stratum characteristic quantities of depth space and absolute age values to establish stratum characteristics, extracting attribute elements and comparing them with a cutoff threshold to obtain sequence graph node relationships, screening heterogeneous edge lithology ratios to obtain fluid migration connectivity graphs, extracting intersection fault surface modulus values and comparing them with reference values to obtain drilling attitude offset sets, and generating exploration driving instructions based on decay offset values to adjust actions. In the present application, stratum characteristic quantities are collected to construct sequence graph node relationships and capture underlying geological evolution rules, fluid migration connectivity graphs are obtained based on the graphs to realize spatial three-dimensional feature reconstruction, intersection fault topology modulus is extracted to obtain attitude offset sets to correct and deduce solidification deviations, drilling attitudes are dynamically adjusted to generate driving decisions, subjective experience weight setting mechanisms are broken, and deposit positioning accuracy is comprehensively enhanced.

Description

Technical Field

[0001] This invention relates to the field of intelligent decision-making technology, and in particular to a mineral exploration decision-making method and system driven by geological knowledge graphs. Background Technology

[0002] The field of intelligent decision-making technology encompasses core matters involving the integration of data computing and logical reasoning to replace humans in business path selection and resource allocation. Overall, it involves collecting multi-dimensional business environment data and relying on computational mechanisms such as statistical inference, heuristic search, and expert reasoning to output execution strategies in complex business environments. This broadly involves data mining, rule construction, and multi-objective optimization. Traditional mineral exploration decision-making methods refer to technical matters related to the spatial location of ore deposits, the delineation of prospecting target areas, and the layout of exploration projects. Geologists manually extract geological features such as stratigraphic lithology, fault trends, and geochemical element anomaly concentrations by consulting regional geological records, mineral prospecting reports, and geophysical and geochemical anomaly maps. These extracted ore-forming elements are input into the weighted evidence method or logistic regression model. Weight coefficients for each element are set based on subjective experience, and the posterior probability of ore formation within the predicted area is calculated. Based on this, the potential ore body boundaries and the planar coordinates and dip angles of exploration boreholes are delineated on a two-dimensional topographic map.

[0003] Traditional mineral exploration decision-making methods rely excessively on geologists to consult massive amounts of geological records and survey reports and manually extract geological features. This operational mode, which sets the weights of mineralization elements based on subjective experience and calculates posterior probabilities, is prone to subjective cognitive biases and insufficient data utilization. Relying on two-dimensional topographic maps to delineate the boundaries of ore bodies and borehole coordinates often results in a lack of spatial information and a rigidity in the dimensions of spatial feature extraction. This leads to low accuracy in delineating prospecting target areas and causes the planning of exploration borehole postures to deviate from the actual spatial distribution patterns of ore bodies. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a geological knowledge graph-driven mineral exploration decision-making method, comprising the following steps:

[0005] S1: Collect the spatial variation rate of elevation depth and absolute age of the stratigraphic floor and compare them with the zero value to establish stratigraphic sequence evolution characteristic quantities;

[0006] S2: Extract the spatiotemporal attribute elements within the stratigraphic sequence evolution characteristic quantities, collect the non-negative truncation threshold values, compare the spatiotemporal attribute elements with the non-negative truncation threshold values, and select the corresponding compliant elements to obtain the node relationships of the geological sequence map.

[0007] S3: Based on the node relationship of the geological sequence map, extract the lithological permeability ratio value in the heterogeneous edge data of the center node of the unconformity surface, compare the lithological permeability ratio value with the barrier constraint benchmark value and remove the heterogeneous edge data that crosses the boundary to obtain the fluid migration connectivity map.

[0008] S4: For the fluid transport connectivity map, extract the topological nodes of the fracture surface that intersect with the borehole trajectory line, obtain the rock mass shear modulus value and fracture development density value within the fracture surface topological node, compare the rock mass shear modulus value with the standard bedrock modulus benchmark value, and obtain the drilling trajectory attitude offset set.

[0009] S5: Call the drilling trajectory attitude offset set, extract the dip angle natural decay offset value, detect the drill bit target dip angle adjustment action value issued by the drilling equipment, perform dynamic offset adjustment based on the drill bit target dip angle adjustment action value and the dip angle natural decay offset value, and generate exploration drill bit drive decision command.

[0010] As a further aspect of the present invention, the stratigraphic sequence evolution characteristics include sedimentary facies change rate, erosion discontinuity attributes, and stratigraphic stacking patterns; the geological sequence map node relationships include topological adjacency matrix, connections between homologous strata, and temporally directed edges; the fluid migration connectivity map includes flow channels, water-blocking boundaries, and fluid migration dominance surfaces; the drilling trajectory attitude offset set includes wellbore dip drift, azimuth abrupt change, and build-up rate variation; and the exploration drill bit drive decision commands include directional motor control parameters, drilling pressure ratio weights, and top drive speed adjustment signals.

[0011] As a further aspect of the present invention, the specific steps of S1 are as follows:

[0012] S101: Collect the spatial change rate of elevation and depth of the stratigraphic floor and the absolute age change value. Extract the floating-point bit width and positive / negative sign bit identifier from the data dimension for the spatial change rate of elevation and depth of the stratigraphic floor and the absolute age change value respectively. Calculate the memory addressing offset corresponding to the floating-point bit width and positive / negative sign bit identifier to obtain the addressing index value of the evolution parameter.

[0013] S102: Based on the evolution parameter addressing index value, retrieve the numerical vector content in the corresponding location in the memory space, perform element-wise subtraction operation on the numerical vector content and the fixed zero reference vector, extract the sign bit of the operation difference to determine the positive or negative deviation from the zero reference vector, and separate the operation difference data components that are equivalent to the zero reference vector to obtain the parameter zero value comparison situation matrix.

[0014] S103: For the zero-value comparison situation matrix of the parameters, calculate the cumulative gradient summation parameter corresponding to the positive and negative deviation situation components on the spatial axis, screen the spatial distribution clusters of the zero extreme value node coordinates mapped by the deviation situation components, extract the relative Euclidean distance parameter between the geometric center of the distribution cluster and the distribution boundary line of the cumulative gradient summation parameter, and arrange the serialized data to establish the stratigraphic sequence evolution characteristic quantity.

[0015] As a further aspect of the present invention, the specific steps of S2 are as follows:

[0016] S201: Extract the spatiotemporal attribute elements within the stratigraphic sequence evolution characteristic quantities, separate the time series items and spatial location items within the spatiotemporal attribute elements, eliminate the dimensional differences by introducing spatiotemporal conversion coefficients, perform vector product operations on the time series items and spatial location items to generate a multidimensional array, allocate memory address pointers to each element in the multidimensional array according to the row and column index parameters in the multidimensional array, and establish spatiotemporal feature matrix components.

[0017] S202: For the spatiotemporal feature matrix components, collect non-negative truncation threshold values, perform difference calculation between the element values ​​in the spatiotemporal feature matrix components and the non-negative truncation threshold values, extract the polarity identifier of the sign bit of the difference calculation output data, assign compliance labels to the elements corresponding to the positive polarity identifiers, and filter the spatiotemporal attribute elements with the compliance labels to obtain the non-negative threshold truncation compliance set.

[0018] S203: Based on the non-negative threshold truncation set, analyze the spatial connectivity vector and temporal span parameter corresponding to each element, introduce dimension conversion coefficients to the spatial connectivity vector and temporal span parameter, calculate the Euclidean distance value in the mapped coordinate system, and create a directed connection topological edge data structure between each element based on the Euclidean distance value to obtain the node relationship of the geological sequence map.

[0019] As a further aspect of the present invention, the specific steps of S3 are as follows:

[0020] S301: Based on the node relationship of the geological sequence map, retrieve the topological node coordinate parameters of the geometric center of the unconformity surface, extract the heterogeneous edge connection sequence corresponding to the topological node coordinate parameters, separate the porosity component and permeability component of the heterogeneous edge connection sequence, and perform division quotient calculation on the porosity component and permeability component to obtain the heterogeneous edge lithology permeability ratio array.

[0021] S302: Call the heterogeneous edge lithological permeability ratio array, collect the barrier constraint benchmark value, perform a term-by-term difference operation between the element values ​​in the heterogeneous edge lithological permeability ratio array and the barrier constraint benchmark value, extract the polarity sign bit of the difference operation output value, filter the heterogeneous edge memory address pointer where the polarity sign bit is mapped and perform address blocking release processing to generate a set of heterogeneous edges that have been screened out due to permeability overrun.

[0022] S303: For the set of heterogeneous edges that have been screened out due to permeation, read the addressing index parameters of the retained nodes and the Boolean value variables of the connectivity between nodes, draw a directed topological connection path in a three-dimensional Cartesian coordinate system for the addressing index parameters, assign weight coefficient labels to the Boolean value variables of the connectivity along the directed topological connection path, and encapsulate and rearrange the directed graph data structure according to the weight coefficient labels to obtain the fluid transport connectivity graph.

[0023] As a further aspect of the present invention, the highest bit of the polarity sign bit is identified. When the highest bit is equal to zero, the associated heterogeneous edge memory address pointer is extracted, and an all-zero overwrite operation is performed on the memory segment pointed to by the heterogeneous edge memory address pointer to change it to a null state.

[0024] As a further aspect of the present invention, the specific steps of S4 are as follows:

[0025] S401: For the fluid transport connectivity map, collect the borehole trajectory line coordinate equation, perform an intersection operation between the borehole trajectory line coordinate equation and the node position matrix in the fluid transport connectivity map, extract the operation output coordinate parameters to match the fracture surface topology node address, separate the attached grid cell index variable, and establish the intersection fracture topology addressing array.

[0026] S402: Based on the index position of the intersection fracture topology addressing array, retrieve the node database, extract the rock mass shear modulus value and fracture development density value in the database, separate the tensor scalar of the rock mass shear modulus value, splice the polarization vector of the fracture development density value, encapsulate the tensor scalar and polarization vector into the same cache space, and obtain the rock mass mechanical property sequence.

[0027] S403: For the rock mass mechanical property sequence, collect the standard bedrock modulus benchmark value, perform differential calculation between the rock mass shear modulus value in the rock mass mechanical property sequence and the standard bedrock modulus benchmark value, extract the polarity position identifier of the calculation output difference, map the coordinate parameter calculation space deflection angle variable of the node where the polarity position is located, and obtain the drilling trajectory attitude offset set.

[0028] As a further aspect of the present invention, the node position matrix in the fluid transport connectivity map is converted into a coordinate array, the coordinate array is substituted into the coordinate equation of the borehole trajectory line to find the root, and the points corresponding to the root calculation result being equal to zero are extracted as the output coordinate parameters of the operation.

[0029] As a further aspect of the present invention, the specific steps of S5 are as follows:

[0030] S501: Extract the natural attenuation offset values ​​of the dip angle inside the drilling trajectory attitude offset set, separate the natural attenuation offset values ​​of the dip angle with the spatial attenuation rate parameter and the underlying gravity polarization scalar, perform matrix dot multiplication on the spatial attenuation rate parameter and the underlying gravity polarization scalar to generate the natural attenuation offset matrix of the dip angle.

[0031] S502: Detect the drill bit target inclination adjustment action value issued by the working register of the drilling equipment, extract the spatial elevation angle change parameter and the lateral azimuth displacement variable inside the drill bit target inclination adjustment action value, perform a series splicing operation on the spatial elevation angle change parameter and the lateral azimuth displacement variable to combine them into an attitude increment sequence, and establish the target inclination adjustment action vector.

[0032] S503: Based on the dip angle natural attenuation offset matrix and the target dip angle adjustment action vector, perform vector summation and differentiation calculations by multiplying the data elements in the target dip angle adjustment action vector and the corresponding numerical elements in the dip angle natural attenuation offset matrix by the corresponding dimension conversion coefficients. Based on the calculation output parameters, compile the timing sequence for writing the equipment's underlying control hardware and generate exploration drill bit drive decision instructions.

[0033] A geological knowledge graph-driven mineral exploration decision-making system includes:

[0034] The stratigraphic feature extraction module collects the spatial variation rate of elevation and depth of the stratigraphic floor and the absolute age variation, and compares them with zero values ​​to establish stratigraphic sequence evolution characteristic quantities.

[0035] The geological sequence map node construction module extracts the spatiotemporal attribute elements within the stratigraphic sequence evolution characteristic quantities, collects non-negative truncation threshold values, compares the spatiotemporal attribute elements with the non-negative truncation threshold values, selects the corresponding compliant elements, and obtains the geological sequence map node relationships.

[0036] The connectivity graph generation module extracts the lithological permeability ratio values ​​within the heterogeneous edge data of the center node of the unconformity surface based on the node relationship of the geological sequence map. It then compares the lithological permeability ratio values ​​with the barrier constraint benchmark values ​​and removes the heterogeneous edge data that crosses the boundary to obtain the fluid transport connectivity graph.

[0037] The offset assessment module extracts the topological nodes of the fracture surface corresponding to the intersection with the borehole trajectory line for the fluid transport connectivity map, obtains the rock mass shear modulus value and fracture development density value within the fracture surface topological node, and compares the rock mass shear modulus value with the standard bedrock modulus benchmark value to obtain the drilling trajectory attitude offset set.

[0038] The decision command generation module calls the drilling trajectory attitude offset set, extracts the dip angle natural decay offset value, detects the drill bit target dip angle adjustment action value issued by the drilling equipment, performs dynamic offset adjustment based on the drill bit target dip angle adjustment action value and the dip angle natural decay offset value, and generates exploration drill bit drive decision command.

[0039] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0040] In this invention, by collecting the characteristic quantities of the stratum floor to extract spatiotemporal attributes, a geological sequence map node relationship is constructed to capture the underlying geological evolution law. Based on this map hierarchy, the permeability ratio is screened to obtain a fluid transport connectivity map, realizing the complete reconstruction of spatially endowed three-dimensional features. Relying on this connectivity map, the topological modulus of the confluence fracture surface is extracted to obtain the drilling attitude offset set, effectively correcting the solidified deviation of traditional subjective inference. Based on the attenuation offset value and the target action of the equipment, the drilling attitude is dynamically adjusted to generate driving decision commands to complete the precise control of the prospecting target area and exploration operation. This breaks the traditional manual experience weight setting mechanism and the inherent drawbacks of two-dimensional planar planning, and comprehensively enhances the positioning accuracy of deep ore bodies and the rationality of engineering layout. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a schematic diagram of the steps of the present invention;

[0043] Figure 2 This is a detailed schematic diagram of S1 of the present invention;

[0044] Figure 3 This is a detailed schematic diagram of S2 of the present invention;

[0045] Figure 4 This is a detailed schematic diagram of S3 of the present invention;

[0046] Figure 5 This is a detailed schematic diagram of S4 of the present invention;

[0047] Figure 6 This is a detailed schematic diagram of S5 of the present invention;

[0048] Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0049] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0050] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0051] Please see Figure 1 This invention provides a geological knowledge graph-driven mineral exploration decision-making method, comprising the following steps:

[0052] S1: Collect the spatial variation rate of elevation depth of the stratigraphic floor and the absolute age variation value, compare the spatial variation rate of elevation depth of the stratigraphic floor with the zero value, and combine the comparison of the absolute age variation value with the zero value to establish the stratigraphic sequence evolution characteristic quantity.

[0053] S2: Extract the spatiotemporal attribute elements within the stratigraphic sequence evolution characteristic quantities, collect the non-negative truncation threshold values, compare the spatiotemporal attribute elements with the non-negative truncation threshold values, select the corresponding compliant elements, and obtain the node relationships of the geological sequence map.

[0054] S3: Based on the node relationship of the geological sequence map, extract the lithological permeability ratio value in the heterogeneous edge data of the center node of the unconformity surface, compare the lithological permeability ratio value with the barrier constraint benchmark value and remove the heterogeneous edge data that crosses the boundary to obtain the fluid migration connectivity map.

[0055] S4: For the fluid transport connectivity map, extract the topological nodes of the fracture surface corresponding to the intersection with the borehole trajectory line, obtain the rock mass shear modulus value and fracture development density value within the fracture surface topological node, compare the rock mass shear modulus value with the standard bedrock modulus benchmark value to obtain the lithological weakening ratio value, and perform attribute association calculation on the lithological weakening ratio value and fracture development density value to obtain the drilling trajectory attitude offset set;

[0056] S5: Call the drilling trajectory attitude offset set, extract the dip angle natural decay offset value, detect the drill bit target dip angle adjustment action value issued by the drilling equipment, perform dynamic offset adjustment based on the drill bit target dip angle adjustment action value and the dip angle natural decay offset value, and generate exploration drill bit drive decision command.

[0057] The stratigraphic sequence evolution characteristics include sedimentary facies change rate, erosion discontinuity properties, and stratigraphic stacking patterns. The geological sequence map node relationships include topological adjacency matrix, connections between homologous strata, and temporally directed edges. The fluid migration connectivity map includes flow channels, water-blocking boundaries, and fluid migration dominance surfaces. The drilling trajectory attitude offset set includes wellbore dip drift, azimuth abrupt change, and build-up rate variation. The exploration drill bit drive decision commands include directional motor control parameters, drill pressure ratio weighting, and top drive speed adjustment signals.

[0058] Please see Figure 2The specific steps of S1 are as follows:

[0059] S101: Collect the spatial change rate of elevation and depth of the stratigraphic floor and the absolute age change value. Extract the floating-point bit width and positive / negative sign bit identifier from the data dimension for the spatial change rate of elevation and depth of the stratigraphic floor and the absolute age change value respectively. Calculate the memory addressing offset corresponding to the floating-point bit width and positive / negative sign bit identifier to obtain the addressing index value of the evolution parameter.

[0060] Geological data was continuously collected for 30 days using a multi-channel grating strain sensor array deployed at depths of 500 to 1200 meters underground. The elevation depth data of the stratigraphic floor was obtained. Analog-to-digital conversion was performed on the analog electrical signals from the acquisition channels every 15 minutes to extract the raw elevation fluctuation sequence, which was then considered as basic environmental entity data in the underlying ontology of the knowledge graph. For this elevation fluctuation sequence, a sliding window averaging filter was performed. The sliding window length was set to 5 sampling points, and the arithmetic mean of the values ​​within the window was calculated as the cleaned depth elevation data, eliminating discrete anomalies caused by environmental thermal noise. Combined with previously conducted radiocarbon isotope determination data from core samples, an absolute age baseline parameter was extracted for the corresponding depth. The current real-time monitored and extrapolated stratigraphic age value was subtracted from this absolute age baseline parameter to calculate the absolute age change value. Based on the actual monitoring data, the spatial change rate of the current stratigraphic floor elevation depth was determined to be 0.08 per meter, and the absolute age change value was 3500 years. For the obtained values ​​of the spatial variation rate of formation floor elevation depth and the absolute age variation, the binary storage structure of the data is read from the bottom-level data register, and the floating-point bit width and positive / negative sign bits are extracted from the data dimensions. Boolean quantization is performed on the positive / negative sign bits; a positive sign bit is assigned a quantization result of 1, and a negative sign bit bit is assigned a quantization result of 0. Based on the actual data parsing, the floating-point bit width of the spatial variation rate of formation floor elevation depth is 64 bits, and the positive sign bit corresponds to a quantization result of 1. The memory addressing offset corresponding to the floating-point bit width and the positive / negative sign bits is calculated. The specific calculation logic is as follows: the basic memory displacement step size parameter is extracted, and the floating-point bit width and the basic memory displacement step size parameter are multiplied to obtain the median value of the bit width span. Then, the median value of the bit width span is summed with the quantization result of the positive / negative sign bits to obtain the memory addressing offset. The specific process for obtaining the basic memory displacement step size parameter involves conducting multiple memory hit rate tests under different step size configurations. The number of cache misses was counted when the step size was set to 2 bytes, 4 bytes, and 8 bytes. The experiments showed that the number of cache misses was lowest when the step size was set to 4 bytes; therefore, the basic memory displacement step size parameter was selected as 4 bytes. A multiplication operation was performed between the 64-bit floating-point width and the 4-byte basic memory displacement step size parameter, resulting in a 256-byte midpoint width span. This 256-byte midpoint width span was then added to the quantization result of the positive / negative sign bit, resulting in a 257-byte memory addressing offset. This memory addressing offset was directly mapped as the evolution parameter addressing index value, achieving efficient collaboration between underlying hardware addressing and knowledge graph memory mapping.

[0061] S102: Based on the addressing index value of the evolution parameter, retrieve the numerical vector content in the corresponding location in the memory space, perform element-wise subtraction operation on the numerical vector content and the fixed zero reference vector, extract the sign bit of the operation difference to determine the positive or negative deviation from the zero reference vector, and separate the data components of the operation difference that are equivalent to the zero reference vector to obtain the parameter zero value comparison situation matrix.

[0062] The corresponding carrying numerical vector content is retrieved, which is represented as a one-dimensional floating-point sequence containing 128 geological attribute feature elements. A fixed zero reference vector, pre-written into a solid-state storage chip, is extracted. This fixed zero reference vector also consists of 128 all-zero elements, representing the initial theoretical state of the strata in absolute stillness without evolution. An element-wise subtraction operation is performed between the retrieved carrying numerical vector content and the fixed zero reference vector. Since all elements of the fixed zero reference vector are 0, the subtraction operation essentially extracts the actual evolutionary deviation of each element in the carrying numerical vector content. The sign bit of each data item in the operational difference sequence output by the above element-wise subtraction operation is extracted. By determining whether the sign bit is 0 or 1, the positive or negative deviation trend relative to the zero reference vector is determined, where 0 corresponds to a positive deviation trend and 1 corresponds to a negative deviation trend. The extracted operational difference sequence is iteratively traversed and detected to separate elements whose operational difference is equivalent to the zero reference vector data component, that is, to separate the static component whose absolute value is equal to 0. By removing the isolated static components, the remaining data items with positive and negative deviations are rearranged according to their initial spatial coordinate indices to obtain the parameter zero-value comparison situation matrix. For example, in the retrieved carrying value vector, the first element is 1.5 mm / s, the second element is -0.8 mm / s, and the third element is 0.0 mm / s. After subtracting from the fixed zero reference vector, the first element produces a positive deviation, the second element produces a negative deviation, and the third element is separated and removed as an equivalent zero reference vector data component. Then, the first and second elements are reconstructed into a 2-row, 1-column parameter zero-value comparison situation matrix.

[0063] Table 1: Statistical Table of Parameter Deviation Trends

[0064] Phase number Total number of nodes Number of nodes with positive deviation Number of negative deviation nodes Number of zero-value nodes 1 128 85 31 12 2 128 92 28 8 3 128 76 41 11

[0065] Table 1 shows the specific node distribution data after subtraction and classification separation processing of the carrying numerical vector content in three consecutive detection stages. It intuitively shows the temporal changes in the distribution of entity attribute states in the knowledge graph.

[0066] S103: For the zero-value comparison situation matrix of parameters, calculate the cumulative gradient summation parameter corresponding to the positive and negative deviation situation components on the spatial axis, screen the spatial distribution clusters of zero extreme value node coordinates mapped by the deviation situation components, extract the relative Euclidean distance parameter between the geometric center of the distribution cluster and the distribution boundary line of the cumulative gradient summation parameter, and arrange the serialized data to establish stratigraphic sequence evolution characteristic quantities.

[0067] The cumulative gradient summation parameter for positive and negative deviations from the situation is calculated along the three-dimensional spatial axes. Specifically, the calculation logic involves extracting the numerical differences between adjacent nodes of each deviation component along the longitude, latitude, and depth axes. The absolute values ​​of these differences are extracted, and the absolute values ​​along the three spatial axes are summed to obtain the local spatial deviation gradient. Then, the zero value of this parameter is compared to the local spatial deviation gradients of all elements in the situation matrix, and the summation is performed to obtain the cumulative gradient summation parameter. For example, if the absolute value of the adjacent numerical difference for the first element along the longitude axis is 2.1, along with 1.4 along the latitude axis and 3.2 along the depth axis, the summation yields a local spatial deviation gradient of 6.7. If the matrix contains 116 valid elements, the cumulative gradient summation of the local deviation gradients of all elements results in a final cumulative gradient summation parameter of 785.4. The distribution clusters of zero-extreme node coordinates mapped by the deviation from the situational components are selected. The coordinate sequences of all nodes within each cluster are extracted, and the mean coordinates are calculated by summing the coordinates of all nodes along each spatial axis and dividing by the total number of nodes to obtain the geometric center coordinates of the distribution cluster. The relative Euclidean distance parameter between the geometric center coordinates of the distribution cluster and the boundary line of the cumulative gradient summation parameter distribution is extracted. The coordinates of the projected control point closest to the geometric center on the distribution boundary line are obtained. The coordinate differences between the geometric center coordinates of the distribution cluster and the coordinates of this projected control point in each corresponding dimension are squared. All squared results are summed to obtain the sum of squares. Finally, the square root of the sum of squares is performed to obtain the relative Euclidean distance parameter. For example, if the geometric center coordinates of the distribution cluster are 150 on the horizontal axis, 200 on the vertical axis, and 50 on the depth axis, and the coordinates of the projected control points are 155 on the horizontal axis, 195 on the vertical axis, and 50 on the depth axis, the difference between the horizontal axis coordinates is calculated to be -5, the difference between the vertical axis coordinates is 5, and the difference between the depth axis coordinates is 0. Squaring these values ​​yields 25, 25, and 0. Summing these values ​​gives a sum of squares of 50. Taking the square root of 50 yields a relative Euclidean distance parameter of 7.07 meters. These multiple relative Euclidean distance parameters are then sequenced according to their chronological order of occurrence to establish stratigraphic sequence evolution characteristics.

[0068] Please see Figure 3 The specific steps of S2 are as follows:

[0069] S201: Extract spatiotemporal attribute elements from stratigraphic sequence evolution features, separate time series terms and spatial location terms within spatiotemporal attribute elements, eliminate dimensional differences by introducing spatiotemporal conversion coefficients, perform vector product operations on time series terms and spatial location terms to generate a multidimensional array, allocate memory address pointers to each element in the multidimensional array according to the row and column index parameters in the multidimensional array, and establish spatiotemporal feature matrix components.

[0070] Based on data timestamps and 3D coordinate labels, the time series items and spatial location items within the spatiotemporal attribute elements are separated. The time series items contain the absolute second timestamp data of each evolution event, while the spatial location items contain the latitude, longitude, and depth 3D coordinate matrix data of the stratigraphic deformation nodes. Vector product operations are performed on the separated time series items and spatial location items to generate a multidimensional array. Specifically, the operation logic is as follows: extract the time series items to form a one-dimensional time row vector, extract the spatial location items to form a set of three-dimensional spatial column vectors, and perform a product operation on each time element in the one-dimensional time row vector with each coordinate element in the three-dimensional spatial column vector set one by one. This produces cross-product data elements containing composite time and spatial dimensional information. These cross-product data elements are then filled into a four-dimensional data structure according to the original time and spatial dimensional correspondences to generate a multidimensional array. For example, extracting time series data includes the elements at 10 seconds and 20 seconds; extracting spatial location data includes the first coordinate element being 30 meters and the second coordinate element being 40 meters. Multiplying 10 seconds with 30 meters and 40 meters respectively yields 300 m / s and 400 m / s; multiplying 20 seconds with 30 meters and 40 meters respectively yields 600 m / s and 800 m / s. These cross-product data elements are arranged to generate a 2x2 multidimensional array. Memory address pointers are allocated to each element in the multidimensional array based on the row and column index parameters. The row and column index values ​​of each element in the multidimensional array are obtained, and the baseline memory starting address parameter is extracted. The row index value is multiplied by the total number of columns in the matrix to obtain the offset column number. Then, the offset column number is added to the column index value to obtain the relative element offset. Finally, the relative element offset is multiplied by the number of bytes occupied by each element's memory, and then added to the baseline memory starting address parameter to establish the spatiotemporal feature matrix components.

[0071] S202: For the spatiotemporal feature matrix components, collect the non-negative truncation threshold values, perform difference calculation between the element values ​​in the spatiotemporal feature matrix components and the non-negative truncation threshold values, extract the polarity identifier of the sign bit of the difference calculation output data, assign compliance labels to the elements corresponding to the positive polarity identifiers, and filter the spatiotemporal attribute elements with the compliance labels to obtain the non-negative threshold truncation compliance set.

[0072] The process of obtaining the non-negative truncation threshold involves collecting a stable background thermal noise intensity fluctuation sequence for the geological region over the past six consecutive months. A set of positive fluctuation extreme values ​​is extracted from this sequence. The arithmetic mean of all data points within this set is calculated, and the resulting average is multiplied by a conservative redundancy coefficient to obtain the non-negative truncation threshold. The conservative redundancy coefficient is obtained by analyzing the signal-to-noise ratio attenuation curve of the hardware sensor. When the sensor operates for more than 3000 hours, the background noise amplitude approximately doubles. To cover this noise disturbance boundary, a conservative redundancy coefficient of 1.5 is set. The arithmetic mean of the stable-state positive fluctuation extreme values ​​is 0.4. This average value of 0.4 is multiplied by the conservative redundancy coefficient of 1.5 to obtain a non-negative truncation threshold of 0.6, which is used as the saliency constraint boundary in knowledge graph relation reasoning. The values ​​of each element in the spatiotemporal feature matrix component are compared with the non-negative truncation threshold value. This involves subtracting the non-negative truncation threshold value from the element value and extracting the sign bit polarity identifier of the difference data at the binary level. The logical state corresponding to this polarity identifier is then detected. When the logical state is positive, it indicates that the element value is greater than or equal to the non-negative truncation threshold value. A compliance label is then assigned to the spatiotemporal feature matrix element corresponding to this positive logical state. After performing a comparison and judgment logic on all elements in the spatiotemporal feature matrix component, all spatiotemporal attribute elements with accompanying compliance labels are selected. These elements with compliance labels are aggregated into an independent dataset, resulting in the non-negative threshold truncation compliance set.

[0073] S203: Based on non-negative threshold truncation of the set, the spatial connectivity vector and temporal span parameter corresponding to each element are analyzed. After introducing the dimension conversion coefficient to the spatial connectivity vector and temporal span parameter, the Euclidean distance value in the mapped coordinate system is calculated. Based on the Euclidean distance value, a directed connection topology edge data structure between each element is created to obtain the node relationship of the geological sequence map.

[0074] The spatial connectivity vector represents the three-dimensional direction and distance attributes between the physical node corresponding to the element and the reference center coordinates. The temporal span parameter represents the time difference between the element's acquisition timestamp and the initial timestamp of the reference evolution. The combined Euclidean distance between the spatial connectivity vector and the temporal span parameter in the multidimensional mapping coordinate system is calculated to construct an entity relationship mapping network for the geological knowledge map. The specific operational logic involves extracting the coordinate displacement components of the spatial connectivity vector in the horizontal, vertical, and longitudinal directions, extracting the temporal span parameter, obtaining the spatiotemporal transformation scaling coefficient, multiplying the temporal span parameter and the spatiotemporal transformation scaling coefficient to convert them into equivalent spatial distance components, then squaring the horizontal, vertical, and equivalent spatial distance components respectively, summing the four squared results to obtain a four-dimensional displacement sum of squares, and finally performing a square root operation on this four-dimensional displacement sum to obtain the Euclidean distance value. The spatiotemporal transformation scaling coefficient is obtained by calibrating it based on the average propagation rate of historical rock mass fracture waves in the region. Experimentally, the average propagation rate of fracture waves is determined to be 3.5 meters per second; therefore, the spatiotemporal transformation scaling coefficient is set to 3.5 meters per second. For example, the extracted lateral displacement component is 4 meters, the longitudinal displacement component is 3 meters, the vertical displacement component is 12 meters, and the temporal span parameter is 2 seconds. Multiplying the temporal span parameter 2 seconds with the spatiotemporal transformation scaling coefficient 3.5 meters per second yields an equivalent spatial distance component of 7 meters. Squaring 4 meters, 3 meters, 12 meters, and 7 meters respectively yields 16, 9, 144, and 49. Adding 16, 9, 144, and 49 together yields a four-dimensional displacement sum of squares, 218. Taking the square root of 218 yields an Euclidean distance of approximately 14.76 meters. Based on the obtained Euclidean distance values ​​between elements, a directed connection topology edge data structure is created between the elements. A connectivity limit parameter is set, which is 25 meters based on the maximum connectivity radius of rock mass fractures. The Euclidean distance between any two elements is compared to the connectivity limit parameter of 25 meters. If the Euclidean distance is less than 25 meters, a directional topological connection is created between the corresponding nodes, with the direction pointing from the node with the earlier timestamp to the node with the later timestamp. If the Euclidean distance is greater than or equal to 25 meters, no connection is created. This calculation and judgment operation is performed on every pair of nodes within the set to generate a network topology containing all connected edges and directional information, thus obtaining the node relationships of the geological sequence map.

[0075] Table 2: Parameter Measurement Table for Node Connectivity Relationships

[0076] No. 1 and No. 2 7 meters 218 14.76 meters Connected No. 1 and No. 5 14 meters 645 25.39 meters Not connected No. 2 and No. 8 10.5 meters 450 21.21 meters Connected

[0077] Table 2 shows the core parameters and the final topology network connectivity distribution after pairing representative nodes within the compliance set to perform four-dimensional distance calculation.

[0078] Please see Figure 4 The specific steps of S3 are as follows:

[0079] S301: Based on the node relationship of the geological sequence map, retrieve the topological node coordinate parameters of the geometric center of the unconformity surface, extract the heterogeneous edge connection sequence corresponding to the topological node coordinate parameters, separate the porosity component and permeability component of the heterogeneous edge connection sequence, and perform division quotient calculation on the porosity component and permeability component to obtain the heterogeneous edge lithology permeability ratio array.

[0080] All geological nodes marked as unconformity contact zones are extracted. The spatial latitude, longitude, and elevation coordinates of these nodes are read, and an arithmetic mean is calculated for each dimension to obtain the topological node coordinates of the geometric center of the unconformity. For example, if three nodes with unconformity characteristics are detected, with depth coordinates of 500m, 510m, and 520m respectively, the arithmetic mean calculation yields a depth dimension topological node coordinate of 510m. The topological node coordinates of the geometric center of the unconformity are located, and all connecting lines emanating from or converging from these coordinates are retrieved and extracted. It is determined whether the nodes at both ends of the connecting line belong to different lithological classification labels. If the lithological classification labels of the two ends are different, the connecting line is included in the heterogeneous edge connection sequence, corresponding to the heterogeneous attribute association paths connecting different categories of conceptual entities in the knowledge graph. Each connecting structure in this heterogeneous edge connection sequence is traversed, and the porosity and permeability components attached to the memory storage block of the connecting structure are extracted. Both porosity and permeability components were derived from laboratory mercury intrusion porosimetry (MIP) tests conducted on rock samples drilled at corresponding strata depths in the early stages. A division calculation was performed on the extracted porosity and permeability components. Specifically, the permeability component value was extracted as the dividend variable, and the porosity component value was extracted as the divisor variable. The division calculation yielded a lithological fluid permeability evaluation index. This index was then stored in a new array according to the index number of the corresponding heterogeneous edge, resulting in a heterogeneous edge lithological permeability ratio array.

[0081] S302: Call the heterogeneous edge lithological permeability ratio array, collect the barrier constraint benchmark value, perform a term-by-term difference operation between the element values ​​in the heterogeneous edge lithological permeability ratio array and the barrier constraint benchmark value, extract the polarity sign bit of the output value of the difference operation, filter the heterogeneous edge memory address pointer where the polarity sign bit is mapped and perform address blocking release processing, and generate a set of heterogeneous edges that have been screened out due to permeability overrun.

[0082] The logic for obtaining the barrier constraint benchmark value is as follows: Historical permeability ratio data of all fault zones in the region that experienced fluid stagnation or closure over the past 10 years are compiled. The minimum effective flow-blocking value is extracted from these historical data and used as the lower limit of the basic barrier. Considering that the deep strata are under high stress and compaction, the pores will be compressed. An additional stress adjustment constant is added to the above lower limit value of the basic barrier to obtain the final barrier constraint benchmark value. A step-by-step difference operation is performed between the values ​​of each element in the heterogeneous edge lithological permeability ratio array and the barrier constraint benchmark value. The specific operation logic is to sequentially extract the values ​​of the elements in the array and subtract the barrier constraint benchmark value from each element value. For the output value sequence of the difference operation, the polarity sign bit of each value is extracted in the hardware storage unit. The status of the polarity sign bit is detected. If the polarity sign bit is negative, it indicates that the permeability ratio of the heterogeneous edge lithology is lower than the barrier constraint benchmark value, meaning that the flow resistance of the rock interface represented by this edge is extremely high, and the fluid cannot effectively cross the boundary. For the differential results that present negative polarity status indicators, the underlying memory address pointers of the corresponding heterogeneous edges where the polarity sign bit mapping is located are filtered. The system memory management function is called to perform address blocking and release processing on the extracted memory address pointers. That is, the connection relationship data in the memory address segment is overwritten and cleared and the memory occupied by the segment is reclaimed. In this way, the edge connections that cannot undergo fluid movement are disconnected in the topology map, generating a set of heterogeneous edges that have been filtered out due to penetration.

[0083] S303: For the heterogeneous edge set that is screened out due to infiltration and cross-boundary, read the addressing index parameter of the retained node and the Boolean value variable of the connectivity between nodes, draw the directed topological connection path in the three-dimensional rectangular coordinate system for the addressing index parameter, assign weight coefficient labels to the connectivity Boolean value variable along the directed topological connection path, and encapsulate and rearrange the directed graph data structure according to the weight coefficient labels to obtain the fluid transport connectivity graph.

[0084] The system reads the reserved node addressing index parameters of the two ends of each edge in the memory pool, and simultaneously reads the Boolean value variable indicating whether there is an actual conduction channel between these two ends. Since invalid edges have been cleared in the early stage, the remaining Boolean value variables of the reserved nodes are all truth values ​​representing the connectivity status. For the obtained reserved node addressing index parameters, the corresponding three-dimensional spatial coordinate data is extracted. In the three-dimensional Cartesian coordinate system drawing engine, a directed topological connection path with a direction is drawn according to the order of the start node and the end node. Along the above-drawn directed topological connection path, composite feature data including the permeability ratio and edge distance are retrieved, and weight quantization is performed to generate weight coefficient labels. The specific quantization logic is as follows: extract the lithological permeability ratio value of the current edge, extract the spatial straight-line distance value of the current edge, perform a division operation on the lithological permeability ratio value and the spatial straight-line distance value to obtain the permeability gradient decay parameter, and assign the permeability gradient decay parameter as the weight coefficient label to the corresponding weight attribute field. The directed graph data structure to be processed is encapsulated and rearranged according to the allocated weight coefficient label size. The specific rearrangement logic is to adjust the arrangement order of each directed edge data block in the contiguous memory space according to the descending order of the weight coefficient label values. Edge data with larger weight values ​​are arranged in the front position closer to the address base address, thereby completing the data structure encapsulation and rearrangement, and finally obtaining the fluid transport connectivity graph.

[0085] Please see Figure 5 The specific steps of S4 are as follows:

[0086] S401: For the fluid transport connectivity map, collect the borehole trajectory line coordinate equation, perform an intersection operation between the borehole trajectory line coordinate equation and the node position matrix in the fluid transport connectivity map, extract the operation output coordinate parameters to match the fracture surface topology node address, separate the attached grid cell index variable, and establish the intersection fracture topology addressing array.

[0087] The borehole trajectory coordinate equation is composed of a sequence of three-dimensional spatial curve equations generated by the drilling engineering design platform using a cubic spline interpolation algorithm based on the target point coordinates and the ground drilling point coordinates. The collected borehole trajectory coordinate equations are then intersected with the node position matrices of each node within the fluid transport connectivity graph. Specifically, the intersection operation logic involves extracting a continuous discrete set of points in three-dimensional space from the borehole trajectory coordinate equations, setting a judgment envelope radius parameter, traversing the three-dimensional coordinates of all nodes within the node position matrix of the fluid transport connectivity graph, calculating the three-dimensional Euclidean distance between each point in the trajectory discrete set and each graph node, and comparing the calculated Euclidean distance value with the judgment envelope radius parameter. If an intersection occurs, an intersection fracture topology addressing array is established, completing the feature alignment of trajectory entities and fracture entities in the knowledge graph.

[0088] S402: Retrieve the node database based on the index position of the intersection fracture topology addressing array, extract the rock mass shear modulus value and fracture development density value in the database, separate the tensor scalar of the rock mass shear modulus value, splice the polarization vector of the fracture development density value, encapsulate the tensor scalar and polarization vector into the same cache space, and obtain the rock mass mechanical property sequence.

[0089] A corresponding search command is initiated in the rock physics and mechanics property node database mounted within the local area network. The rock mass shear modulus and fracture density values ​​at the depth of the corresponding intersection node are extracted from the database. The rock mass shear modulus is calculated by extracting shear wave velocity data obtained from dipole sonic logging at the same stratum in adjacent wells, combining it with formation bulk density parameters, and performing a product operation of density and the square of the shear wave velocity. The fracture density value is calculated by extracting bilateral lateral resistivity logging data, combining it with array microresistivity scanning imaging data, and performing image pixel thresholding to extract the area ratio of black spots and fractures. For example, the rock mass shear modulus value of a certain intersection node is obtained as 15 gigapascals through the product operation of the square of the shear wave velocity and the density parameter; the fracture density value is calculated as 45 fractures per cubic meter through resistivity imaging. During data analysis, tensor scalar attribute data accompanying the rock mass shear modulus values ​​are extracted. This tensor scalar reflects the anisotropic strength differences of the shear modulus in different directions. Simultaneously, the polarization vector of the fracture development density values ​​is spliced, containing the three-dimensional dip angle and strike distribution information of the main fractures. The extracted tensor scalar and the spliced ​​polarization vector are then packaged in memory. A contiguous high-speed cache memory space is allocated, and the tensor scalar and polarization vector are encapsulated together within the same cache address segment, thus forming a set of structured data units. The same iterative reading and data encapsulation operation is performed on all node index positions within the intersection fracture topology addressing array. All generated structured data units are combined and arranged to obtain the rock mass mechanical property sequence.

[0090] S403: For the rock mass mechanical property sequence, collect the standard bedrock modulus benchmark value, perform differential calculation between the rock mass shear modulus value in the rock mass mechanical property sequence and the standard bedrock modulus benchmark value, extract the polarity position identifier of the calculation output difference, map the coordinate parameter calculation space deflection angle variable of the node where the polarity position is located, and obtain the drilling trajectory attitude offset set.

[0091] The logic for obtaining the standard bedrock modulus benchmark value is as follows: Historical rock shear modulus data collected from 15 stable vertical wells in the target layer within the oil and gas field block over the past three years are extracted. After filtering out extremely large or small erroneous data due to equipment failure, the median is extracted from the remaining valid historical data, and the median is selected as the standard bedrock modulus benchmark value. After median extraction, the standard bedrock modulus benchmark value for this depth layer in this block is 25 gigapascals. The rock shear modulus values ​​contained in each intersection node within the rock mass mechanical property sequence are then differentially calculated with the aforementioned standard bedrock modulus benchmark value. Specifically, the standard bedrock modulus benchmark value is subtracted from the rock shear modulus value at the intersection node to generate the calculated difference data. The polarity identifier of this calculated difference data is extracted, and its state is analyzed. For example, the rock mass shear modulus of a certain intersection node in the sequence is extracted to be 15 GPa. A difference calculation is performed between this and the standard bedrock modulus benchmark of 25 GPa, yielding a difference of -10 GPa. The polarity identifier is extracted as negative polarity, indicating that the shear strength of the rock mass at this location is significantly lower than that of standard bedrock due to the development of fractures. This negative polarity identifier is mapped to the three-dimensional coordinate parameters of the intersection node. Based on the dip and strike values ​​in the polarization vector, the spatial deflection angle variable that the drill bit will experience is calculated. Specifically, the absolute value of the difference is extracted as a scaling factor, the basic deflection anomaly number is extracted, and the absolute value of the difference is multiplied by the basic deflection anomaly number to obtain the basic deflection anomaly amplitude. Then, the basic deflection anomaly amplitude is multiplied by the dip value in the polarization vector after a sine function mapping to obtain the spatial deflection angle variable at this node due to rock mechanical heterogeneity. The logic for setting the basic deflection anomaly number is based on fitting data from previous indoor build-up rate experiments for this type of drill bit. Experiments determined the basic deflection anomaly number to be 0.15 degrees per gigapascal. The absolute value of the difference was extracted as 10 gigapascals. Multiplying this absolute value by the basic deflection anomaly number of 0.15 yielded a basic deflection anomaly amplitude of 1.5 degrees. The polarization vector dip angle of this node was extracted as 30 degrees, and the sine function value of 30 degrees was calculated to be 0.5. Multiplying the basic deflection anomaly amplitude of 1.5 degrees by 0.5 yielded a spatial deflection angle variable of 0.75 degrees. This operation was performed on all nodes within the sequence, and the resulting set of all calculated spatial deflection angle variables was arranged to obtain the drilling trajectory attitude offset set.

[0092] Table 3: Experimental Verification Table for Drilling Trajectory Attitude Deviation Prediction

[0093] A1 15 Gipascals -10 gigascals 0.50 0.75 degrees 0.81 degrees A2 18 Gipascal -7 Gipascals 0.86 0.90 degrees 0.95 degrees A3 22 Gipascal -3 Gipascals 0.70 0.31 degrees 0.28 degrees

[0094] Table 3 shows the results of a comparison between the spatial deflection angle variables calculated from some key intersection nodes after applying the above-mentioned mechanical property difference and dip angle combined deduction logic and the downhole measured data.

[0095] Please see Figure 6 The specific steps of S5 are as follows:

[0096] S501: Extract the natural attenuation offset values ​​of the dip angle inside the drilling trajectory attitude offset set, separate the natural attenuation offset values ​​of the dip angle with the spatial attenuation rate parameter and the underlying gravity polarization scalar, perform matrix dot multiplication on the spatial attenuation rate parameter and the underlying gravity polarization scalar to generate the natural attenuation offset matrix of the dip angle.

[0097] The dip angle natural attenuation offset reflects the objective sagging deflection of the drill string assembly under the combined effects of its own gravity and formation heterogeneity. The data packet format of this dip angle natural attenuation offset is analyzed to separate the internal spatial attenuation rate parameter and the bottom gravity polarization scalar. The spatial attenuation rate parameter is obtained statistically from the historical natural change angle of the trajectory dip angle per meter of drilling depth, while the bottom gravity polarization scalar reflects the projection correlation strength between the equivalent center of gravity of the current drill string assembly and the vertical gravity vector. Matrix multiplication is then performed on the separated spatial attenuation rate parameter and bottom gravity polarization scalar. The specific operational logic is as follows: extract the spatial decay rate parameter sequence calibrated in segments due to different well depths to form a one-dimensional horizontal vector; extract the components of the bottom gravity polarization scalar in different spatial axis directions to form a one-dimensional vertical vector; perform dot product calculation on the horizontal vector and the vertical vector, that is, multiply the corresponding elements and then sum all the product results to generate the comprehensive deviation value corresponding to each well depth; arrange the comprehensive deviation values ​​of each segment according to the well depth sequence to form a new two-dimensional grid structure, and generate the dip angle natural decay offset matrix.

[0098] S502: Detect the drill bit target inclination adjustment action value issued by the working register of the drilling equipment, extract the spatial elevation angle change parameter and the lateral azimuth displacement variable inside the drill bit target inclination adjustment action value, perform a series splicing operation on the spatial elevation angle change parameter and the lateral azimuth displacement variable to combine them into an attitude increment sequence, and establish the target inclination adjustment action vector.

[0099] The system monitors the drilling equipment's working registers installed at the wellhead control station, reading real-time drill bit target inclination adjustment values ​​generated by operators or upper-level closed-loop control algorithms in response to current geological challenges. This value encapsulates the expected three-dimensional spatial increment command for the drill bit's subsequent active deflection. The system extracts the spatial elevation angle variation parameter and lateral azimuth displacement variable encapsulated within this drill bit target inclination adjustment value, and uses a knowledge graph-based real-time command semantic conversion interface to map these two degrees of freedom requirements into entity actions at the execution layer. The spatial elevation angle variation parameter represents the expected pitch angle adjustment in the vertical plane, while the lateral azimuth displacement variable represents the expected left-right angle adjustment in the horizontal projection. For example, reading the register command reveals a spatial elevation angle variation parameter of 1.5 degrees and a lateral azimuth displacement variable of 2.0 degrees right turn. The extracted spatial elevation angle variation parameters and lateral azimuth angle displacement variables are extracted into the memory buffer and then concatenated. The data segment of the spatial elevation angle variation parameter is placed in the high address and the data segment of the lateral azimuth angle displacement variable is placed in the low address. Bit concatenation and combination operations are then performed to establish the target tilt angle adjustment action vector.

[0100] S503: Based on the natural attenuation offset matrix of the dip angle and the target dip angle adjustment action vector, the vector summation and differentiation calculation is performed after multiplying the data elements in the target dip angle adjustment action vector and the corresponding numerical elements in the natural attenuation offset matrix of the dip angle by the corresponding dimension conversion coefficient. Based on the calculation output parameters, the timing sequence of the equipment's underlying control hardware is compiled to generate the exploration drill bit drive decision command.

[0101] Extract all active adjustment data elements stored within the target dip angle adjustment action vector, and simultaneously locate the objective attenuation values ​​corresponding to the current well depth in the dip angle natural attenuation offset matrix. Perform vector summation on the data elements within the target dip angle adjustment action vector and their corresponding values ​​in the dip angle natural attenuation offset matrix. Since dip angle natural attenuation can hinder or promote adjustment, performing summation is equivalent to calculating the actual absolute compensation that the drill string must overcome to achieve the desired attitude. Specifically, the calculation logic involves extracting the spatial elevation angle variation parameter extracted from the target dip angle adjustment action vector and performing a summation operation on the corresponding comprehensive deviation value in the dip angle natural attenuation offset matrix. For example, if the target spatial elevation angle variation parameter is a 1.5-degree rise, and the corresponding comprehensive deviation value at the corresponding depth is a 0.48-degree droop, then sum and add the 1.5-degree deviation to the inverse offset of the 0.48-degree droop (negative 0.48 degrees), obtaining the actual absolute compensation required as 1.98 degrees. Perform a time-dimension derivative calculation on this actual absolute compensation. The hardware response time constant of the underlying servo motor is extracted, and the actual absolute compensation amount is divided with the hardware response time constant to calculate the drill bit angle adjustment rate command. The adjustment rate command and the absolute compensation amount parameter are converted into the corresponding level pulse width modulation duty cycle value. These duty cycle control words are pushed into the underlying hardware first-in-first-out queue in sequence according to the clock cycle to generate the exploration drill bit drive decision command.

[0102] Please see Figure 7 A geological knowledge graph-driven mineral exploration decision-making system, including:

[0103] The stratigraphic feature extraction module collects the spatial variation rate of elevation and depth of the stratigraphic floor and the absolute age variation, and compares them with zero values ​​to establish stratigraphic sequence evolution characteristic quantities.

[0104] The geological sequence map node construction module extracts spatiotemporal attribute elements within the stratigraphic sequence evolution characteristic quantities, collects non-negative truncation threshold values, compares the spatiotemporal attribute elements with the non-negative truncation threshold values, selects the corresponding compliant elements, and obtains the geological sequence map node relationships.

[0105] The connectivity graph generation module extracts the lithological permeability ratio values ​​within the heterogeneous edge data of the center node of the unconformity surface based on the node relationship of the geological sequence map. It then compares the lithological permeability ratio values ​​with the barrier constraint benchmark values ​​and removes the heterogeneous edge data that crosses the boundary to obtain the fluid migration connectivity graph.

[0106] The offset assessment module extracts the topological nodes of the fracture surface corresponding to the intersection with the borehole trajectory line based on the fluid transport connectivity map, obtains the rock mass shear modulus value and fracture development density value within the fracture surface topological node, and compares the rock mass shear modulus value with the standard bedrock modulus benchmark value to obtain the drilling trajectory attitude offset set.

[0107] The decision command generation module calls the drilling trajectory attitude offset set, extracts the dip angle natural decay offset value, detects the drill bit target dip angle adjustment action value issued by the drilling equipment, performs dynamic offset adjustment based on the drill bit target dip angle adjustment action value and the dip angle natural decay offset value, and generates exploration drill bit driven decision command.

[0108] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the described technical solutions.

Claims

1. A geological knowledge graph-driven mineral exploration decision-making method, characterized in that, Includes the following steps: S1: Collect the spatial variation rate of elevation depth and absolute age of the stratigraphic floor and compare them with the zero value to establish stratigraphic sequence evolution characteristic quantities; S2: Extract the spatiotemporal attribute elements within the stratigraphic sequence evolution characteristic quantities, collect the non-negative truncation threshold values, compare the spatiotemporal attribute elements with the non-negative truncation threshold values, and select the corresponding compliant elements to obtain the node relationships of the geological sequence map. S3: Based on the node relationship of the geological sequence map, extract the lithological permeability ratio value in the heterogeneous edge data of the center node of the unconformity surface, compare the lithological permeability ratio value with the barrier constraint benchmark value and remove the heterogeneous edge data that crosses the boundary to obtain the fluid migration connectivity map. S4: For the fluid transport connectivity map, extract the topological nodes of the fracture surface that intersect with the borehole trajectory line, obtain the rock mass shear modulus value and fracture development density value within the fracture surface topological node, compare the rock mass shear modulus value with the standard bedrock modulus benchmark value, and obtain the drilling trajectory attitude offset set. S5: Call the drilling trajectory attitude offset set, extract the dip angle natural decay offset value, detect the drill bit target dip angle adjustment action value issued by the drilling equipment, perform dynamic offset adjustment based on the drill bit target dip angle adjustment action value and the dip angle natural decay offset value, and generate exploration drill bit drive decision command.

2. The geological knowledge graph-driven mineral exploration decision-making method according to claim 1, characterized in that, The stratigraphic sequence evolution characteristics include sedimentary facies change rate, erosion discontinuity attributes, and stratigraphic stacking patterns. The geological sequence map node relationships include topological adjacency matrix, connections between homologous strata, and temporally directed edges. The fluid migration connectivity map includes flow channels, water-blocking boundaries, and fluid migration dominance surfaces. The drilling trajectory attitude offset set includes wellbore dip drift, azimuth abrupt change, and build-up rate variation. The exploration drill bit drive decision commands include directional motor control parameters, drill pressure ratio weighting, and top drive speed adjustment signals.

3. The geological knowledge graph-driven mineral exploration decision-making method according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Collect the spatial change rate of elevation and depth of the stratigraphic floor and the absolute age change value. Extract the floating-point bit width and positive / negative sign bit identifier from the data dimension for the spatial change rate of elevation and depth of the stratigraphic floor and the absolute age change value respectively. Calculate the memory addressing offset corresponding to the floating-point bit width and positive / negative sign bit identifier to obtain the addressing index value of the evolution parameter. S102: Based on the evolution parameter addressing index value, retrieve the numerical vector content in the corresponding location in the memory space, perform element-wise subtraction operation on the numerical vector content and the fixed zero reference vector, extract the sign bit of the operation difference to determine the positive or negative deviation from the zero reference vector, and separate the operation difference data components that are equivalent to the zero reference vector to obtain the parameter zero value comparison situation matrix. S103: For the zero-value comparison situation matrix of the parameters, calculate the cumulative gradient summation parameter corresponding to the positive and negative deviation situation components on the spatial axis, screen the spatial distribution clusters of the zero extreme value node coordinates mapped by the deviation situation components, extract the relative Euclidean distance parameter between the geometric center of the distribution cluster and the distribution boundary line of the cumulative gradient summation parameter, and arrange the serialized data to establish the stratigraphic sequence evolution characteristic quantity.

4. The geological knowledge graph-driven mineral exploration decision-making method according to claim 3, characterized in that, The specific steps of S2 are as follows: S201: Extract the spatiotemporal attribute elements within the stratigraphic sequence evolution characteristic quantities, separate the time series items and spatial location items within the spatiotemporal attribute elements, eliminate the dimensional differences by introducing spatiotemporal conversion coefficients, perform vector product operations on the time series items and spatial location items to generate a multidimensional array, allocate memory address pointers to each element in the multidimensional array according to the row and column index parameters in the multidimensional array, and establish spatiotemporal feature matrix components. S202: For the spatiotemporal feature matrix components, collect non-negative truncation threshold values, perform difference calculation between the element values ​​in the spatiotemporal feature matrix components and the non-negative truncation threshold values, extract the polarity identifier of the sign bit of the difference calculation output data, assign compliance labels to the elements corresponding to the positive polarity identifiers, and filter the spatiotemporal attribute elements with the compliance labels to obtain the non-negative threshold truncation compliance set. S203: Based on the non-negative threshold truncation set, analyze the spatial connectivity vector and temporal span parameter corresponding to each element, introduce dimension conversion coefficients to the spatial connectivity vector and temporal span parameter, calculate the Euclidean distance value in the mapped coordinate system, and create a directed connection topological edge data structure between each element based on the Euclidean distance value to obtain the node relationship of the geological sequence map.

5. The geological knowledge graph-driven mineral exploration decision-making method according to claim 4, characterized in that, The specific steps for S3 are as follows: S301: Based on the node relationship of the geological sequence map, retrieve the topological node coordinate parameters of the geometric center of the unconformity surface, extract the heterogeneous edge connection sequence corresponding to the topological node coordinate parameters, separate the porosity component and permeability component of the heterogeneous edge connection sequence, and perform division quotient calculation on the porosity component and permeability component to obtain the heterogeneous edge lithology permeability ratio array. S302: Call the heterogeneous edge lithological permeability ratio array, collect the barrier constraint benchmark value, perform a term-by-term difference operation between the element values ​​in the heterogeneous edge lithological permeability ratio array and the barrier constraint benchmark value, extract the polarity sign bit of the difference operation output value, filter the heterogeneous edge memory address pointer where the polarity sign bit is mapped and perform address blocking release processing to generate a set of heterogeneous edges that have been screened out due to permeability overrun. S303: For the set of heterogeneous edges that have been screened out due to permeation, read the addressing index parameters of the retained nodes and the Boolean value variables of the connectivity between nodes, draw a directed topological connection path in a three-dimensional Cartesian coordinate system for the addressing index parameters, assign weight coefficient labels to the Boolean value variables of the connectivity along the directed topological connection path, and encapsulate and rearrange the directed graph data structure according to the weight coefficient labels to obtain the fluid transport connectivity graph.

6. The geological knowledge graph-driven mineral exploration decision-making method according to claim 5, characterized in that, Identify the highest bit of the polarity sign bit. When the highest bit is equal to zero, extract the associated heterogeneous edge memory address pointer and perform an all-zero overwrite operation on the memory segment pointed to by the heterogeneous edge memory address pointer to change it to a null state.

7. The geological knowledge graph-driven mineral exploration decision-making method according to claim 5, characterized in that, The specific steps of S4 are as follows: S401: For the fluid transport connectivity map, collect the borehole trajectory line coordinate equation, perform an intersection operation between the borehole trajectory line coordinate equation and the node position matrix in the fluid transport connectivity map, extract the operation output coordinate parameters to match the fracture surface topology node address, separate the attached grid cell index variable, and establish the intersection fracture topology addressing array. S402: Based on the index position of the intersection fracture topology addressing array, retrieve the node database, extract the rock mass shear modulus value and fracture development density value in the database, separate the tensor scalar of the rock mass shear modulus value, splice the polarization vector of the fracture development density value, encapsulate the tensor scalar and polarization vector into the same cache space, and obtain the rock mass mechanical property sequence. S403: For the rock mass mechanical property sequence, collect the standard bedrock modulus benchmark value, perform differential calculation between the rock mass shear modulus value in the rock mass mechanical property sequence and the standard bedrock modulus benchmark value, extract the polarity position identifier of the calculation output difference, map the coordinate parameter calculation space deflection angle variable of the node where the polarity position is located, and obtain the drilling trajectory attitude offset set.

8. The geological knowledge graph-driven mineral exploration decision-making method according to claim 7, characterized in that, The node position matrix within the fluid transport connectivity map is converted into a coordinate array. The coordinate array is then substituted into the coordinate equation of the borehole trajectory line to find the roots. The points corresponding to the root calculation result being equal to zero are extracted as the output coordinate parameters.

9. The geological knowledge graph-driven mineral exploration decision-making method according to claim 7, characterized in that, The specific steps of S5 are as follows: S501: Extract the natural attenuation offset values ​​of the dip angle inside the drilling trajectory attitude offset set, separate the natural attenuation offset values ​​of the dip angle with the spatial attenuation rate parameter and the underlying gravity polarization scalar, perform matrix dot multiplication on the spatial attenuation rate parameter and the underlying gravity polarization scalar to generate the natural attenuation offset matrix of the dip angle. S502: Detect the drill bit target inclination adjustment action value issued by the working register of the drilling equipment, extract the spatial elevation angle change parameter and the lateral azimuth displacement variable inside the drill bit target inclination adjustment action value, perform a series splicing operation on the spatial elevation angle change parameter and the lateral azimuth displacement variable to combine them into an attitude increment sequence, and establish the target inclination adjustment action vector. S503: Based on the dip angle natural attenuation offset matrix and the target dip angle adjustment action vector, perform vector summation and differentiation calculations by multiplying the data elements in the target dip angle adjustment action vector and the corresponding numerical elements in the dip angle natural attenuation offset matrix by the corresponding dimension conversion coefficients. Based on the calculation output parameters, compile the timing sequence for writing the equipment's underlying control hardware and generate exploration drill bit drive decision instructions.

10. A geological knowledge graph-driven mineral exploration decision-making system, characterized in that, The system is used to implement the geological knowledge graph-driven mineral exploration decision-making method according to any one of claims 1-9, and the system includes: The stratigraphic feature extraction module collects the spatial variation rate of elevation and depth of the stratigraphic floor and the absolute age variation, and compares them with zero values ​​to establish stratigraphic sequence evolution characteristic quantities. The geological sequence map node construction module extracts the spatiotemporal attribute elements within the stratigraphic sequence evolution characteristic quantities, collects non-negative truncation threshold values, compares the spatiotemporal attribute elements with the non-negative truncation threshold values, selects the corresponding compliant elements, and obtains the geological sequence map node relationships. The connectivity graph generation module extracts the lithological permeability ratio values ​​within the heterogeneous edge data of the center node of the unconformity surface based on the node relationship of the geological sequence map. It then compares the lithological permeability ratio values ​​with the barrier constraint benchmark values ​​and removes the heterogeneous edge data that crosses the boundary to obtain the fluid transport connectivity graph. The offset assessment module extracts the topological nodes of the fracture surface corresponding to the intersection with the borehole trajectory line for the fluid transport connectivity map, obtains the rock mass shear modulus value and fracture development density value within the fracture surface topological node, and compares the rock mass shear modulus value with the standard bedrock modulus benchmark value to obtain the drilling trajectory attitude offset set. The decision command generation module calls the drilling trajectory attitude offset set, extracts the dip angle natural decay offset value, detects the drill bit target dip angle adjustment action value issued by the drilling equipment, performs dynamic offset adjustment based on the drill bit target dip angle adjustment action value and the dip angle natural decay offset value, and generates exploration drill bit drive decision command.

Images