Rock high slope excavation anchoring deformation prediction method based on artificial intelligence

Abstract

The invention relates to the technical field of deformation prediction, in particular to a rock high slope excavation anchoring deformation prediction method based on artificial intelligence, and the method comprises the following steps: obtaining multi-source image data, analyzing rock volume and gravity center transfer, extracting a corresponding relation between abnormal stress and displacement, constructing a structure offset path, and screening continuous abnormal features. And extracting a key deformation rate interval to obtain an anchoring deformation trend prediction result. According to the method, the position of an unbalanced section is judged by constructing a gravity center transfer track and a mass difference space sequence and combining a coupling relation between a stress abnormal point and a displacement response direction, a collaborative evolution section of a structure offset path and a stress boundary is extracted, and a continuous abnormal zone is screened according to a crack change and joint fluctuation trend. A key change interval is extracted in combination with the time feature of the shear deformation rate, dynamic judgment of a structural deformation path and an extension direction is completed, and the spatial continuity recognition capability and the time response resolution are enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of deformation prediction, in particular to a rock high slope excavation anchoring deformation prediction method based on artificial intelligence. BACKGROUND

[0002] The technical field of deformation prediction relates to the estimation and determination of the physical form changes such as displacement, crack, settlement of engineering structures or rock-soil bodies under the action of external force or environmental influence. This technical field includes core matters such as data acquisition, deformation identification, mechanical parameter analysis, mathematical model establishment, and prediction result verification. Common methods include trend analysis based on historical monitoring data, analytical method based on geophysical parameters, finite element numerical analysis method based on continuum mechanics theory and its iterative optimization method, etc. Generally, by setting initial boundary conditions and material constitutive relations, displacement or stress field is solved in combination with field monitoring data, and then the deformation state of future specific space-time nodes is derived. Among them, the traditional rock high slope excavation anchoring deformation prediction refers to predicting the tensile, shear slip or displacement variation of rock mass structure caused by stress change, hydrological condition disturbance, etc. during the blasting, excavation and anchoring construction of rock high and steep slope, using methods such as empirical formula fitting, elastoplastic constitutive model calculation, difference method and finite element discrete modeling method. The content usually includes rock mass classification, structure surface trend angle, anchor length and spacing, load boundary condition, blasting vibration velocity measurement in the deformation prediction report, and regression fitting or model iteration are combined with the measured deformation monitoring data to complete the prediction analysis.

[0003] The existing technology focuses on model parameter setting and monitoring quantity regression analysis in the process of structural deformation response identification, lacks deep extraction of spatial trajectory evolution and structure gravity migration process among multi-layer information, and shows response delay and insufficient positioning accuracy when involving stress field non-uniform distribution and structure displacement path identification. It is difficult to construct an abnormal distribution framework with spatial extension at the initial stage of deformation anomaly, especially in the face of unstable continuous profile deformation trend, crack closure and joint angle dramatic fluctuation, etc. The prediction result cannot timely reflect the dynamic evolution characteristics of the key area, causing the accumulation of anchoring response path identification deviation, and it is difficult to support the stability dynamic evaluation in the process of partition regulation and continuous construction. SUMMARY

[0004] In order to solve the technical problems existing in the prior art, the embodiments of the present application provide a rock high slope excavation anchoring deformation prediction method based on artificial intelligence, which comprises the following steps: In order to achieve the above purpose, the present application adopts the following technical scheme: a rock high slope excavation anchoring deformation prediction method based on artificial intelligence, comprising the following steps: S1: Obtain a profile density map, a lithology division map and a construction progress map, analyze rock mass volume changes, extract quality differences, align structure positions according to coordinates, track gravity center shifts, and obtain slope structure migration path change characteristics; S2: Based on the slope structure migration path change characteristics, extract an anchor layout map and a stress monitoring map, analyze anchor rod axial force change trends, search for abnormal points, judge uneven area positions according to displacement response directions, and obtain stress abnormal distribution area position corresponding relationships; S3: Based on the stress abnormal distribution area position corresponding relationships, extract cross-line, judge direction consistency, connect continuous track points, analyze angle and distance changes, and obtain offset path and stress boundary associated paragraphs; S4: Based on the offset path and stress boundary associated paragraphs, read shear deformation, joint inclination and crack characteristics, compare response directions, filter change areas, extract layer positions, and obtain continuous abnormal trend distribution content; S5: Based on the continuous abnormal trend distribution content, extract deformation rate change areas, analyze direction continuity, judge the relationship with construction time, and obtain rock slope anchoring deformation trend prediction results.

[0005] As a further scheme of the present application, the slope structure migration path change characteristics include rock mass volume change trends, rock mass quality spatial differences, and slope gravity center shift trajectories, the stress abnormal distribution area position corresponding relationships include non-uniform section positions, layer number corresponding areas, and stress abnormal points and displacement direction response relationships, the offset path and stress boundary associated paragraphs include trajectory spatial direction continuous consistent points, layer spatial angle change trends, and boundary trend overlapping paragraphs, the continuous abnormal trend distribution content includes shear deformation abnormal profiles, joint inclination dramatic change areas, and fracture opening and closing fluctuation concentrated areas, and the rock slope anchoring deformation trend prediction results include shear deformation rate change concentrated sections, construction time sequence and response change order, and spatial direction extension trend positions.

[0006] As a further scheme of the present application, the gravity center shift refers to the spatial migration of the overall mass center position of the rock mass structure caused by changes in rock mass volume and density during high slope excavation construction; The anchor rod axial force change trend refers to the changes in the axial force data recorded by the differential anchor rods along the time axis during the anchor layout process.

[0007] As a further scheme of the present application, the abnormal point refers to a point that appears different from the surrounding monitoring values during monitoring, which is manifested as a change in displacement direction, a stress mutation, an abnormal rate, and represents early signs of potential risks and structural instability; The uneven location refers to a location in the anchoring area, at which the force of the structure and the displacement response direction change suddenly, indicating that the force distribution of the area is uneven, and the location area has a hidden danger of structural stability.

[0008] As a further scheme of the present application, the specific steps of S1 are: S101: Obtain the profile density map, lithology division map and construction progress map, extract the rock volume and density value of the stage profile coordinate range, configure the quality data frame of the corresponding area, compare the quality distribution change of the stage, extract the regional quality state over time, and obtain the rock mass stage quality evolution sequence; S102: Based on the rock mass stage quality evolution sequence, extract the stage boundary coordinate point, compare the spatial position of the same numbered point, judge the direction change, extract the trajectory data of the position continuous offset, divide the path according to the direction change trend, and obtain the structure position direction migration path set; S103: Call the structure position direction migration path set, retrieve the position of the center of gravity point in the path, compare whether the transfer direction and the path trend are continuous, arrange the transfer segments with consistent direction in the order of coordinates as a continuous path, and obtain the slope structure migration path change characteristics.

[0009] As a further scheme of the present application, the specific steps of S2 are: S201: Based on the slope structure migration path change characteristics, extract the layout position and axial force time sequence change information corresponding to each group of anchor rods in the anchoring layout map and stress monitoring data map, extract the change trend according to the layout number in turn, filter the number section that continuously presents a linear trend turning point, and obtain the axial force change trend annotation sequence; S202: Based on the axial force change trend annotation sequence, extract the coordinate position of the numbered point, compare the displacement response direction with the adjacent area, judge whether the direction is consistent, filter the layout point with direction deviation, extract the spatial distribution information, and obtain the direction response offset coordinate set; S203: Based on the direction response offset coordinate set, locate the numbered index position in the layout map layer, extract the coordinate range of the adjacent numbered area in the same layer, merge the continuous areas in order into the same response area, output the area position corresponding layer coordinates, and obtain the stress abnormal distribution location corresponding relationship.

[0010] As a further scheme of the present application, the specific steps of S3 are: S301: Based on the stress abnormal distribution location corresponding relationship, extract the intersection area of the slope structure trajectory and the stress boundary line segment in the layer, compare the structure offset direction and the boundary extension direction vector at the intersection point, judge whether the direction is consistent, extract the intersection line segment with consistent direction, and obtain the direction consistent intersection paragraph set; S302: Based on the direction consistent cross paragraph set, continuous trajectory points on the coordinate extraction layer are extracted, trajectory nodes continuously consistent in spatial distribution direction are screened, point positions before and after the deflection of the trajectory line segment are compared, line segments without direction deflection are identified, and a spatial direction continuous trajectory set is obtained; S303: The spatial direction continuous trajectory set is called to intercept the spatial angle change and distance distribution range of the line segment in the layer, the segment consistent with the structure offset direction and stress boundary trend continuation is screened, the trajectory number and layer coordinate range are output, and the offset path and stress boundary associated paragraph are obtained.

[0011] As a further scheme of the present application, the specific steps of S4 are: S401: Based on the offset path and stress boundary associated paragraph, the shear deformation sequence of the monitoring point, the joint inclination change range and the crack opening and closing state along the path direction are extracted, the direction scalar of adjacent monitoring points in the parameter sequence is analyzed, the continuous response segment with consistent direction is screened, and a direction unified deformation segment set is obtained. S402: Based on the direction unified deformation segment set, the layer coordinate section where the segment is located is identified, the monitoring curve fluctuation form of the adjacent area is extracted, the response path offset position point at the continuous trend change node is extracted, the spatial connection relationship between the offset points is analyzed, and a continuous response segment positioning index is obtained. S403: Based on the continuous response segment positioning index, the number and coordinate area in the layer are matched, the path response sequence with continuous change form under the same number is screened, the point position not forming a periodic segment is removed, the corresponding index position and response direction information in the path sequence are aggregated, and continuous abnormal trend distribution content is obtained.

[0012] As a further scheme of the present application, the process of screening the path response sequence with continuous change form under the same number is specifically: the layer number and corresponding coordinate area included in the continuous response segment positioning index are extracted, the response direction of the monitoring point in the same number path is compared point by point in sequence, the continuity feature of the response direction change between adjacent points is identified, and whether it constitutes a continuous path segment is judged according to the trend of direction retention and gradual change. The process of removing the point position not forming a periodic segment is specifically: the response path segment is identified, the direction response change of each point position in the path sequence is repeatedly compared, whether there is a regular direction transformation is judged, and the corresponding point position is removed from the sequence when the continuity of the direction between adjacent points in the path is lacking and there is random jump. The process of aggregating the corresponding index position and response direction information in the path sequence is specifically: according to the continuous point position retained in the path, the spatial index and response direction are extracted in sequence according to the number, the change state of the response direction is compared, the response paragraph with consistent direction is analyzed, and the spatial position sequence and direction information are corresponded.

[0013] As a further scheme of the present application, the specific steps of S5 are: S501: Based on the continuous abnormal trend distribution content, extract the section with concentrated change amplitude in the shear deformation rate sequence, call the corresponding index of the monitoring time sequence, analyze the continuity of adjacent data points in the time axis direction, filter the segments with direction maintaining state, and obtain the shear rate continuous section index; S502: Based on the shear rate continuous section index, match the start and end time points of the section with the construction time sequence node position in the layer, analyze the response delay distribution position point, filter the segments with time difference structure existing in the previous section, and obtain the construction response time delay path set; S503: Call the construction response time delay path set, aggregate the path corresponding spatial coordinate point sequence, extract the path segment with consistent direction according to coordinate sorting, exclude the segments with direction jump, and obtain the rock slope anchoring deformation trend prediction result according to the trajectory continuity and position trend maintaining relationship.

[0014] Compared with the prior art, the present application has the advantages and positive effects that: In the present application, by constructing the gravity center transfer trajectory and the mass difference space sequence, combining the coupling relationship between the stress abnormal point and the displacement response direction, determining the position of the unbalanced section, extracting the co-evolution section of the structure offset path and the stress boundary, screening the continuous abnormal zone according to the crack change and joint fluctuation trend, and combining the time characteristics of the shear deformation rate to extract the key change interval, the dynamic judgment of the structure deformation path and the extension direction is completed, and the spatial continuity recognition ability and the time response resolution are enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0016] Figure 1 The step flowchart of the present application is shown in the figure; Figure 2 The S1 refinement diagram of the present application is shown in the figure; Figure 3 The S2 refinement diagram of the present application is shown in the figure; Figure 4 The S3 refinement diagram of the present application is shown in the figure; Figure 5 The S4 refinement diagram of the present application is shown in the figure; Figure 6 The S5 refinement diagram of the present application is shown in the figure. DETAILED DESCRIPTION

[0017] The technical solutions in the present application will be described below with reference to the drawings.

[0018] In the embodiments of the present application, the words such as “example”, “for example” are used to represent an example, illustration or description. Any embodiment or design scheme described as “example” in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word “example” is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by “and / or” can be both, or can be one of the two.

[0019] In the embodiments of the present application, “image” and “picture” can be used interchangeably at times, and it should be pointed out that the meanings expressed are consistent when the distinction is not emphasized. “Of”, “corresponding” and “relevant” can be used interchangeably at times, and it should be pointed out that the meanings expressed are consistent when the distinction is not emphasized.

[0020] In the embodiments of the present application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1, and the meanings expressed are consistent when the distinction is not emphasized.

[0021] To make the technical problems, technical solutions and advantages to be solved by the present application more clear, the following will be described in detail with reference to the drawings and specific embodiments.

[0022] Please refer to Figure 1 The embodiments of the present application provide a rock high slope excavation anchoring deformation prediction method based on artificial intelligence, which comprises the following steps: S1: Obtain the profile density graph, lithology area division graph and stage construction progress distribution graph of the rock slope in the excavation process, judge the rock block volume change trend between profiles according to the construction time sequence, extract the mass difference of the rock layer in the spatial direction, compare the position change according to the coordinate sequence, track the transfer track of the gravity center of the slope with the stripping stage, and obtain the change characteristics of the slope structure migration path; S2: Based on the change characteristics of the slope structure migration path, call the anchoring area layout graph and stress monitoring data graph, screen the linear inclination of the axial force change of the differential position anchor rod, search the abnormal distribution points in turn according to the layout direction, judge whether there is a non-equilibrium section according to the displacement response direction of the corresponding position and the adjacent area, and obtain the corresponding relationship of the stress abnormal distribution area position according to the area position corresponding to the layer number. S3: Based on the stress anomaly distribution area position corresponding relationship, the intersection graph line area is extracted, it is judged whether the slope structure offset direction and the stress change area boundary direction remain consistent in the extension process, the trajectory points of continuous spatial direction are connected, the spatial angle change and distance trend on the layer are analyzed in sections, the trajectory paragraphs continuously coinciding with the offset direction and the boundary trend are extracted, and the offset path and stress boundary associated paragraphs are obtained; S4: Based on the offset path and stress boundary associated paragraphs, the shear deformation change trend, joint dip angle change amplitude and fracture opening and closing change characteristics of the surrounding monitoring points are read along the path direction, whether the fluctuation directions of adjacent regions are consistent is compared, the profile area with continuous abnormal change is screened, and the continuous abnormal trend distribution content is obtained according to the layer index corresponding number and position area; S5: Based on the continuous abnormal trend distribution content, the shear deformation rate change amplitude concentrated section in the monitoring parameter time sequence is extracted, the direction continuation characteristics in the adjacent time window are analyzed, the corresponding construction time sequence node and response change occurrence sequence are analyzed, the time distribution situation is analyzed through the trend line change amplitude, the position content maintaining the extension trend in the spatial direction is extracted, and the rock slope anchoring deformation trend prediction result is obtained.

[0023] The slope structure migration path change characteristics include rock mass volume change trend, rock mass quality spatial difference, and slope gravity center transfer trajectory. The stress anomaly distribution area position corresponding relationship includes non-equilibrium section position, layer number corresponding area, stress anomaly point and displacement direction response relationship. The offset path and stress boundary associated paragraphs include trajectory spatial direction continuous consistent point, spatial angle change trend on the layer, and boundary trend coinciding paragraph. The continuous abnormal trend distribution content includes shear deformation abnormal profile, joint dip angle dramatic change area, and fracture opening and closing fluctuation concentrated area. The rock slope anchoring deformation trend prediction result includes shear deformation rate change concentrated section, construction time sequence and response change sequence, and spatial direction extension trend position.

[0024] Please refer to Figure 2 , and the specific steps of S1 are as follows: S101: Obtain the profile density graph, lithology division graph and construction progress graph, extract the rock mass volume and density value of the stage profile coordinate range, configure the quality data frame corresponding to the area, compare the quality distribution change of the stage, extract the regional quality state with time, and obtain the rock mass stage quality evolution sequence; The layer content in the profile density map, the lithology division map and the construction progress map needs to read the coordinate boundary information, the layer number and the legend mark corresponding to each profile position in the sheet in sequence, convert the sheet frame of each stage profile map in the sheet into a spatial coordinate range, combine the actual coordinate value range calibrated according to the scale in the sheet, extract the graphic element area within the coordinate boundary corresponding to each profile, mark the relationship between the color shown by different regions and the corresponding density value according to the color distribution in the density map, divide the rock block unit density value covered by each coordinate region according to the color distribution in the density map, and then aggregate the pixels in the same coordinate range into a unit region. The area of the unit region is converted into an area value according to the scale of the drawing, and the height is determined according to the profile depth information, and then the three-dimensional volume is derived. The rock block volume is estimated by multiplying the area value of the divided region by the vertical extension height. If the area of a region is 15x10 proportional grid units, the height is 6 grids, then the converted volume is 900 volume units, and the density value of the corresponding region is 2.65 units of mass, then the mass data of the region is 2385 units of mass. The area and matching density of all profile maps need to be read in this way, and the mass value is calculated item by item to form the mass data frame under the profile stage. When processing the construction progress map of each stage, the stage identifier and boundary range marked in the sheet need to be retrieved, the mass data frames of the same coordinate region in each stage are compared, the mass values of the region in different stages are compared horizontally, the change difference and change direction are extracted, and it is judged whether the difference is an increment or a decrement. If the mass of a region is 2500 in stage one and 2150 in stage two, the difference is -350, which represents a decrease in mass in the later stage. According to the sheet annotation order, the change trend of each stage region is sorted in sequence to obtain the mass change state sequence of multiple regions with the progress of the stage. The mass change state set is formed by rearranging the stage sequence, and the mass change state corresponding to all regions is connected in sequence according to the coordinate index, so that the mass change path of the region in the progress of each stage is extracted, and the stage mass evolution sequence of the rock mass is obtained.

[0025] S102: Based on the stage mass evolution sequence of the rock mass, extract the stage boundary coordinate point, compare the spatial positions of the same numbered points, judge the direction change, extract the trajectory data of the continuous position offset, divide the path according to the direction change trend, and obtain the structure position direction migration path set; First, the coordinate points corresponding to the profile boundary positions in each stage are extracted and indexed in a numbered manner. For each numbered point, the coordinate position changes in different stages are read segment by segment. By reading the horizontal and vertical displacement values of the coordinate points, the offset direction of the point is identified and judged. For example, the coordinates of numbered point P1 in three stages are (128, 95), (130, 98), and (132, 102). It is judged that there is a shift in the longitudinal and horizontal coordinates, and it can be concluded that it is shifted to the right and up. Then, the coordinate points of the same number in multiple stages are connected in order to form a trajectory path. Next, the above coordinate comparison and direction judgment operations are performed on all numbered points. Points with continuous shift characteristics are grouped into the same trajectory group. If multiple points move in the same direction during stage advancement, it indicates that they have consistent direction. Record the trajectory set number and spatial position coordinates. These trajectories are divided into different path clusters based on direction changes. If the angle changes in multiple trajectories are within 5 degrees, they can be grouped into a consistent direction path. For example, points numbered P3, P4, and P5 change direction to the east and north in three stages, and the direction angle fluctuation is not more than 5 degrees. They can be grouped into the same trajectory direction path set. In this process, the starting point and end point coordinates of each trajectory need to be identified. The trend of each path in the path set is compared, and the direction of each sub-path in the path set is classified. Finally, the paths are divided by number, and the contents of each path cluster are aggregated to form a unified set. The number, extension direction, and corresponding coordinate point set of each path are output to obtain the structure position direction migration path set.

[0026] S103: Call the structure position direction migration path set, retrieve the center point position in the path, compare the transfer direction and path trend, arrange the direction consistent transfer segment as a continuous path in coordinate order, and obtain the slope structure migration path change characteristics; First, read all the coordinate points of each path in the path set, and extract the centroid position of the area enclosed by the path. The centroid coordinates are taken as the spatial barycenter of the path. The extraction process is performed on each path in turn. For example, the path with path number T01 has coordinate points (118, 92), (121, 94), and (125, 98). The center point is calculated as (121.3, 94.6). After recording the barycenter coordinates, the direction of the path from the starting point to the ending point is annotated. The relative position change direction of the barycenter is confirmed. It is judged whether the change direction is consistent with the original direction of the path. If the directions are consistent, the transfer segment information is retained, and the path number and trend identification are recorded. Further, the transfer segments with consistent directions are sorted according to the order of the barycenter coordinates on the coordinate axis. The continuous sorting results from the starting path to the ending path are obtained. If path numbers T05, T06, and T07 show a monotonic increasing trend in the barycenter coordinates, it indicates that the spatial arrangement has continuity. The connection relationship of the paths in this direction is further expanded. If there is a reverse path direction in the middle, the path sequence expansion is interrupted. This operation is performed on each initial path for multiple rounds of direction consistency comparison and connectivity judgment. The coordinate range covered by each continuous path and the connection order number are annotated in the final output. Finally, the slope structure migration path change characteristics are obtained.

[0027] Please refer to Figure 3 The specific steps of S2 are as follows: S201: Based on the slope structure migration path change characteristics, extract the layout position and axial force time sequence change information corresponding to each group of anchor rods in the anchor layout map and stress monitoring data map. According to the layout number, extract the change trend in turn. Select the number section that continuously presents a linear trend turning point to obtain the axial force change trend annotation sequence. First, locate the layer number corresponding to each path, extract all anchor layout number information covering the layer in the anchor layout map, and record the three-dimensional coordinate layout point for each number. Match the corresponding number position with the number axial force time series curve in the axial force monitoring data map. Segment the axial force change trend sequence under each anchor number, and use the numerical difference sequence in each time period as the basis for judging the change trend. Then, perform directional consistency judgment on each segment of the axial force sequence. If the axial force growth direction changes from positive to negative, or from negative to positive in three or more consecutive monitoring points, record the start and end numbers of the change position and mark them as a trend turning section in the order of the layout numbers. If there are multiple monotonic increasing segments in a certain number interval, such as A15 to A23, that turn into decreasing segments, record it as a linear turning segment. Perform this screening operation on all layout number sequences, extract and label the number sections that meet the turning conditions, and associate the three-dimensional layout position corresponding to the number based on the layout map coordinates. Finally, construct a number and change state comparison sequence table, and distinguish the number segments with different direction turning. Establish a symbol sequence representing the turning direction, such as assigning a negative sign to a section that turns from rising to falling, and vice versa. Get the axial force change trend annotation sequence.

[0028] S202: Based on the axial force change trend annotation sequence, extract the number point coordinate position and compare it with the adjacent area displacement response direction to judge whether the direction is consistent. Screen the layout points with direction deviation, extract the spatial distribution information, and get the direction response deviation coordinate set. Firstly, call each set of anchor rod layout number recorded therein, extract its three-dimensional coordinate information in order of number, in the process of extracting coordinates, the three-dimensional space position of all numbered points in the same layout section is aggregated to form the spatial layout point set corresponding to the layout path, and the displacement response direction information corresponding to the number is searched in the stress monitoring data graph, the direction information is compared with the response direction data of the adjacent two numbers before and after the numbered point, if the response direction change amount continuously remains in single direction change, it is determined that the direction is consistent, if the response direction change sign is different from that of the adjacent point at any numbered point, it is determined that the direction deviates, record the numbered point and its spatial coordinate information, and record its layer index number at the same time, in the specific execution, take a numbered point B12 as an example, its displacement response direction is inclined to the lower left, compared with adjacent numbered points B11 and B13, if the direction change sign changes, extract the three-dimensional coordinates of the point x=42.1, y=108.3, z=16.7, and record its layer number T07, at the same time, the point in the layout path is marked as a direction abnormal path point, after completing the direction consistency judgment of all numbered points, the spatial information aggregation operation is performed on all layout numbered points with direction deviation, the coordinate position is classified according to the layer index, and the numbered points are arranged in ascending order, finally the direction response deviation coordinate set is obtained.

[0029] S203: Based on the direction response deviation coordinate set, the index position in the layout graph layer is located, the coordinate range of the adjacent numbered area in the same layer is extracted, the continuous area is merged in order as a same response area, the area position corresponding layer coordinate is output, and the stress abnormal distribution area position corresponding relationship is obtained. Firstly, read the spatial coordinates and number information of each group of offset points in turn, match each number with the index position of the anchor point in the layout graph layer, locate the specific coordinate area of each offset number in the layer grid in the layout graph, sequentially determine the number value increment or decrement direction between adjacent numbers, if there is a continuous numbering relationship between adjacent numbers, and the coordinate position belongs to the same block or exists boundary contact in the layer grid, it is determined that the point group belongs to the same numbered area, for example, numbered P13, P14, P15 are horizontally distributed in layer T03, and the x-axis coordinate difference value is less than 2 meters, and the y-axis difference value is not more than 1 meter, it is determined that they are continuous numbers in the same area, and the numbered segment is recorded as a group of continuous numbered areas, then according to the coordinate value range of all points in the region, the maximum x, y value and minimum x, y value of the region are extracted, and the coordinate bounding box of the numbered region is established, then the operation is continued in the adjacent layer, the layers are classified according to the layer number, and the region division process of all offset numbered points is completed in turn, finally, multiple numbered points with continuous numbering, close coordinates and consistent layers are aggregated into independent response areas, the layer number and coordinate bounding box range are output, and the stress abnormal distribution area position corresponding relationship is obtained.

[0030] Referring to Figure 4 The specific steps of S3 are as follows: S301: Based on the stress anomaly distribution area position corresponding relationship, the intersection area of the slope structure trajectory and the stress boundary line segment in the layer is extracted, the structure deflection direction and the boundary extension direction vector at the intersection point are compared, it is judged whether the direction consistency is possessed, the intersection line segment with consistent direction is extracted, and the direction consistent intersection paragraph set is obtained; Firstly, the coordinate sequence of each slope structure trajectory line segment is retrieved from the layout layer, and the annotated stress boundary line segment set in the corresponding layer is extracted synchronously. By means of coordinate interval comparison, the intersection area with coordinate overlap or approximate coincidence between the two types of line segments is identified. The intersection point coordinates are obtained in the intersection area, and the strike vector of the trajectory segment at the intersection point and the extension vector of the boundary line segment are located. Further, the angle direction of the two vectors is judged. If the angle between the two vectors is less than the specified threshold angle and the direction cosine value is greater than the direction consistency judgment threshold, it is indicated that the trajectory segment and the boundary segment at the intersection point exist direction consistency, for example, the direction of the trajectory segment at the intersection point is north by east 10 degrees, and the direction of the stress boundary line is north by east 12 degrees, the direction angle is 2 degrees, and the direction consistency judgment threshold is set to 5 degrees. Therefore, the intersection point satisfies the direction consistency condition, and the trajectory segment and the boundary segment corresponding to the intersection point are jointly regarded as the direction consistent intersection line segment, and the number index and the layer number thereof are recorded. After traversing all the intersection areas, the line segment set satisfying the direction consistency judgment is merged and numbered, and is arranged into a unified data frame output, and the direction consistent intersection paragraph set is obtained.

[0031] S302: Based on the direction consistent intersection paragraph set, the continuous trajectory points on the coordinate extraction layer are extracted, the trajectory nodes continuously consistent in spatial distribution direction are screened, the points before and after the direction deflection of the trajectory line segment are compared, the line segment without direction deflection is identified, and the spatial direction continuous trajectory set is obtained; First, the trajectory point sequence associated with each trajectory segment is extracted from the cross paragraph, and the coordinates in the layer are sequentially sorted. The sorted trajectory points are arranged in order according to the number to form an initial trajectory path set. In the set, the directional vector between any two adjacent trajectory points is calculated according to the coordinate difference. Three consecutive points form two line segments. The direction deflection is identified by calculating the included angle of the two line segments. If the included angle of the two directional vectors is less than the spatial deflection threshold, it is considered that the two trajectory segments have not deflected in the direction. The trajectory points here maintain continuity in the spatial distribution direction. For example, if the included angle of the two line segments formed by the trajectory points P1, P2 and P3 is 4 degrees and the deflection threshold is 5 degrees, P2 is retained as a direction continuous point. If the included angle is 6 degrees, the point is excluded. All adjacent points are subjected to the included angle judgment operation. The trajectory points that meet the spatial direction consistency condition are filtered out in each trajectory. Finally, the trajectory segment set with continuous direction characteristics is output according to the trajectory number index, and the spatial direction continuous trajectory set is obtained.

[0032] S303: Call the spatial direction continuous trajectory set, intercept the spatial angle change and distance distribution range of the line segment presented in the layer, filter the segments consistent with the structure offset direction and stress boundary trend continuation, output the trajectory number and layer coordinate range, and obtain the offset path and stress boundary associated paragraph. First, the spatial vector parameter set formed by the start and end coordinate points of each line segment is extracted according to the trajectory number information in the trajectory set. The spatial angle change of each line segment is identified separately. In the identification process, the change of the included angle between the coordinates is used as the angle determination basis. The angle interval between the start point and the end point vector is extracted. The trajectory line segment directional vector is compared with the structure offset directional vector. When the trajectory line segment direction is within the angle offset range threshold and the direction is consistent, it is considered as a consistent segment. For example, in the northeast direction of the structure offset direction, if the angle vector of a certain trajectory line segment is less than 15 degrees with the offset direction and points to the northeast, it is judged that the line segment is a direction consistent segment. The start and end coordinates of the direction consistent line segment in the layer are extracted and the distance between each segment is calculated to construct the distance distribution sequence of each trajectory. Then, the stress boundary trend directional vector is introduced. If the trajectory line segment direction is consistent with the stress boundary trend vector within the included angle range, the trajectory line segment is considered as a line segment that meets the dual consistency conditions of the structure offset direction and the stress boundary trend. Finally, the trajectory number, layer coordinate range and direction attribute of the line segment are output to obtain the offset path and stress boundary associated paragraph.

[0033] Please refer to Figure 5 , the specific steps of S4 are: S401: Based on the offset path and stress boundary association paragraph, the shear deformation sequence of the monitoring point, the joint inclination range and the crack opening and closing state are extracted along the path direction, the direction scalar of the adjacent monitoring point in the parameter sequence is analyzed, the continuous response segment with consistent direction is screened, and the direction unified deformation segment set is obtained; First, index extraction is performed on each monitoring point in the path range, and the shear deformation sequence data, joint inclination range information and crack opening and closing state content of the corresponding monitoring point are called in the layer database. The displacement direction value of the shear deformation sequence is extracted in sequence, the joint inclination is formed into a change interval according to the angle difference between adjacent two points, and the crack state is positively or negatively identified by the gap width change direction between adjacent nodes. In the above three parameter data, the numbering section consistent with the coordinate arrangement order is extracted, and the direction uniformity judgment operation is performed on the direction vector of the continuous monitoring point in each path segment. In the judgment process, the direction vector pair between each adjacent two points is called to perform angle comparison operation. If the included angle is within the set threshold range, for example, less than 20 degrees, and the scalar is positive in the coordinate arrangement direction, it is determined that the directions are consistent. According to this standard, the monitoring point paragraphs in the whole path that meet the direction continuous state are screened, and the shear deformation value, joint inclination value and crack state change trend in the paragraph are recorded respectively. In the example, the shear deformation in a certain numbering segment (such as P101-P108) is southeast displacement, the joint inclination continuously increases from 25 degrees to 32 degrees, and the crack opening increases from 3.5mm to 4.2mm. The directions are consistent. Such segments can be classified into unified response sequence. In all path segments, this screening process is performed in turn, and after the extraction of each segment with consistent direction is completed, the direction unified deformation segment set is finally obtained.

[0034] S402: Based on the direction unified deformation segment set, the layer coordinate section where the segment is located is identified, the monitoring curve fluctuation pattern of the adjacent area is extracted, the response path offset position point is extracted at the continuous trend change node, the spatial connection relationship between the offset points is analyzed, and the continuous response segment positioning index is obtained. First, the coordinate analysis operation is performed on each segment corresponding to the layer number, the minimum rectangular area range covered by the segment in the layer file is extracted, and all monitoring point numbers in the range are listed as a candidate set, then the displacement response curve of the adjacent numbered points in the candidate set is called, the data points are extracted from the curve graph segment by segment, the trend nodes in the fluctuation form are identified by judging the number of positive and negative switching times of adjacent points in the value change direction, wherein if a numbered point has the same curve direction as the previous numbered point and the opposite curve direction as the next numbered point, it is a trend change node, further, the offset position points in the path direction are extracted at these nodes, the offset points should satisfy the condition that the coordinate increment direction changes or the adjacent coordinate increment is negative, then the coordinates between all offset points are combined, the adjacent offset points are connected in ascending order of number, the distance distribution of adjacent two points in horizontal and vertical directions is analyzed, if one of the horizontal direction difference and the vertical direction difference between the two points is zero, or the ratio of the two is an integer multiple of 1, it is considered that the two points have a spatial connection relationship, for example, in the example, if the offset points M105 and M106 are spaced 2.0 meters apart in the X direction and 0.0 meters apart in the Y direction, they are identified as continuous connection points, the connectivity of all offset points is judged in this way, and each offset path sequence with continuous connectivity is recorded in turn, the point position number in the path sequence is output as an index to form the unique positioning sequence of each deformation trend segment in the layer, and finally the continuous response segment positioning index is obtained.

[0035] S403: Based on the continuous response segment positioning index, match the numbered and coordinate area in the layer, filter the path response sequence with continuous change form under the same number, eliminate the points that do not form periodic segments, aggregate the index position and response direction information corresponding to the path sequence, and obtain the continuous abnormal trend distribution content; First, retrieve the layer file one by one according to the order of the index set, extract the coordinate point range corresponding to each number, and determine its spatial position in the monitoring area and the response path number group to which the number belongs. Multiple coordinate segments belonging to the same path number are grouped into a trajectory response sequence. Then, the displacement or shear deformation monitoring value is extracted in the response sequence according to the time sequence. For each sequence, a continuity judgment process is applied, that is, three consecutive points are taken as a judgment unit. If the change trend of the middle point in the unit and the adjacent two points show the same direction of monotonic change, it is recorded as a continuous trend segment. For example, the value corresponding to number P304 is 1.2, 1.5, and 1.8. It can be judged that the trend is increasing and belongs to the continuous change form path. If the trend direction switches repeatedly within three points, it is considered as a non-continuous segment, and the point number in the judgment sequence is removed. Further, in the continuous trend path obtained, it is identified whether it has periodic characteristics between adjacent segments. The judgment logic is: calculate the response amplitude difference between the trend segments and the number of continuous segments. If the amplitude difference fluctuation amplitude between different segments does not exceed the set threshold and the point number error in the trend period does not exceed the allowed range, it is considered that the path has periodic form. Otherwise, the segment number point is also removed. After removing all points without periodic characteristics, the remaining effective points are merged according to the path number. Through the coordinate backtracking of the number position, the spatial position of the layer is obtained, and the response trend direction of each point is counted. If more than 70% of the point trend direction vectors in a path segment are consistent, that is, they are all rising or falling, the segment is assigned the "consistent direction" attribute. Finally, the number index of each path segment and its direction trend value are combined to form a number and response direction pairing sequence, and the continuous abnormal trend distribution content is obtained.

[0036] Please refer to Figure 6 The specific steps of S5 are: S501: Based on the continuous abnormal trend distribution content, extract the section with concentrated change amplitude in the shear deformation rate sequence, call the index corresponding to the monitoring time sequence, analyze the continuity of adjacent data points in the time axis direction, and filter the segments with direction maintaining state to obtain the shear rate continuous segment index. First, relying on the path paragraph and coordinate sequence labeled in the previous stage, the corresponding shear deformation rate sequence is retrieved from each paragraph, and the monitoring point number in each paragraph is corresponded to the rate value point recorded in the rate curve. The value segment with obvious change amplitude is extracted by point-by-point traversal. In the execution process, the monitoring point number is arranged in ascending order, and the amplitude difference operation is performed with a sliding window of five consecutive points. The change value of the current point and the previous period is calculated. If the continuous three difference values are greater than the preset change threshold, such as 0.08 mm / h, it is marked as a region with obvious change. In actual application, the shear rate of numbers P211~P215 is 0.10, 0.21, 0.34, 0.43, and 0.51 respectively, which can be judged as a region with obvious rate increasing trend. Then, the monitoring time axis index of the data segment is located, the index numbers of adjacent points are extracted in chronological order, the time stamps are read in turn, and whether the adjacent time intervals are equal is calculated. The method for judging continuity is: if the time difference remains fixed within the allowable error range, such as the time interval is set to 3 hours, the error is allowed ±0.2 hours, then the current paragraph is continuous in time, otherwise the part of the data segment is excluded from the sequence. After continuity, it is further checked whether the rate change direction of adjacent monitoring points is consistent. The difference between the two adjacent points is used as the basis for judgment. If the three consecutive difference values are all positive or all negative, it is recorded as a direction maintaining state, otherwise it is marked as a direction reversal and not included in the subsequent sequence. For the numbered paragraph that meets the three conditions of amplitude significant jump, time continuity and direction consistency at the same time, the index number set corresponding to it in the original data set is generated as the labeled output, and the shear rate continuous segment index is finally obtained.

[0037] S502: Based on the shear rate continuous segment index, match the start and end time points of the segment with the construction time sequence node position in the layer, analyze the response delay distribution position point, filter the fragments with time difference structure from the previous segment, and obtain the construction response time delay path set; First, the start and end indexes of each numbered paragraph are extracted from the obtained continuation section, and the monitoring time corresponding to the indexes is retrieved. The start and end time points of each continuation segment are read in sequence index. Taking numbered P110 as an example, if the start index is 215 and the end index is 228, the corresponding times are the 45th period and the 51st period, respectively. The duration of the continuation segment is 6 unit periods. Then, the time period is compared with the construction time sequence node list recorded in the layer. The comparison content is segmented and positioned according to the corresponding relationship between the construction stage number and the spatial coordinates of the monitoring point. The coordinate range covered by each construction number in the layer is extracted by looking up the table. The coordinates corresponding to the continuation segment are matched with the coordinate interval of the construction drawing. The matching method is to determine whether the monitoring point belongs to the coordinate falls within the coordinate boundary range corresponding to the specified construction number. If there is a situation where the position falls in but the time does not coincide, it is marked as a response delay area. The start time of the segment and the start time of the construction number to which it belongs are continued to be extracted and the difference is calculated. If the difference exceeds the preset response delay threshold, such as 2 unit periods, the current segment number is recorded as a response delay segment. This operation is performed for each detection point segment to obtain a set of segments with time offset phenomena. Then, all monitoring points marked as delay segments are summarized and intersected with the original path segment to extract the numbered chain that still maintains trajectory continuity as the subsequent analysis object to obtain the construction response time delay path set.

[0038] S503: Call the construction response time delay path set, aggregate the path corresponding spatial coordinate point sequence, extract the path segment with consistent direction according to coordinate sorting, exclude the segment with direction jump, get the rock slope anchoring deformation trend prediction result according to the trajectory continuity and position trend maintaining relationship; First, all path segments on the coordinate point set are extracted from the delay path set according to the number, the spatial position coordinates of the monitoring points in each segment are listed according to the path number, and these point sequence is sorted according to the longitudinal coordinate priority, horizontal coordinate sequence, so that the arrangement order is consistent with the actual space trend. After sorting, the direction extension consistency judgment process is entered, for each two vectors composed of three consecutive points in each path segment, the included angle of the two vectors in two-dimensional space is calculated, if the included angle value is not greater than 15 degrees, it is judged that the direction is consistent, otherwise it is judged that the direction has jumped. Taking path P503 as an example, the point sequence is A (102.1, 208.3), B (103.5, 209.2), C (105.4, 212.7), the angle of AB vector and BC vector is compared, if the angle is 9 degrees, the three points can be regarded as direction continuation consistent, if the angle is 24 degrees, it is regarded as direction jump. The angle threshold of 15 degrees is a preset angle threshold, which is selected based on the conventional amplitude change of displacement change in structure construction process. The selection method refers to the angle change characteristics in the relatively stable section of continuous shear rate change in historical data. The jump section is excluded from the path used for trend fitting. Then the trajectory continuity of each direction continuation path segment retained after screening is judged. The judgment standard is that the distance between any two adjacent points does not exceed the set coordinate interval threshold, for example, set to 3 meters. If there is a point distance in the path that exceeds this value, it is determined that the trajectory is discontinuous, and this section does not participate in trend extraction. Then, the trend extension analysis is performed on all coordinates in the continuous direction maintaining paragraph. Linear fitting is performed on the same direction segment to extract the direction trend of the overall path. Taking number P703 as an example, the fitting result slope is 0.75, indicating that the overall displacement trend develops in an approximate 45-degree direction. Combined with the segment information with trajectory continuity and direction stability, the prediction trend line is constructed, and the rock slope anchoring deformation trend prediction result is obtained.

[0039] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence, characterized in that, Includes the following steps: S1: Obtain profile density map, lithological classification map and construction progress map, analyze the volume change of rock blocks, extract mass differences, compare the structural position by coordinates, track the transfer of gravity center, and obtain the characteristics of slope structure migration path change. S2: Based on the characteristics of the slope structure migration path change, extract the anchorage layout map and stress monitoring map, analyze the trend of anchor axial force change, retrieve abnormal points, determine the location of unbalanced areas according to the displacement response direction, and obtain the corresponding relationship of stress anomaly distribution areas. S3: Based on the locational correspondence of the stress anomaly distribution area, extract the cross lines, determine the consistency of direction, connect the continuous trajectory points, analyze the changes in angle and distance, and obtain the segment associated with the offset path and stress boundary. S4: Based on the offset path and stress boundary associated paragraphs, read shear deformation, joint dip angle and crack characteristics, compare response directions, filter change areas, extract layer positions, and obtain continuous abnormal trend distribution content; S5: Based on the continuous abnormal trend distribution, extract the deformation rate change area, analyze the direction continuity, determine the relationship with construction time, and obtain the prediction result of rock slope anchorage deformation trend.

2. The method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence as described in claim 1, characterized in that, The characteristics of the slope structure migration path include the trend of rock block volume change, spatial differences in rock layer mass, and the trajectory of the slope gravity center transfer. The corresponding relationship of the stress anomaly distribution area includes the location of the non-equilibrium section, the area corresponding to the layer number, and the response relationship between stress anomaly points and displacement direction. The segments associated with the offset path and stress boundary include points with continuous and consistent spatial direction of the trajectory, the trend of spatial angle change on the layer, and segments with overlapping boundary trends. The content of the continuous anomaly trend distribution includes shear deformation anomaly profile, areas with drastic changes in joint dip angle, and areas with concentrated fluctuations in crack opening and closing. The prediction results of the rock slope anchorage deformation trend include concentrated segments of shear deformation rate change, the order of construction sequence and response change, and the location of spatial extension trend.

3. The method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence as described in claim 1, characterized in that, The gravity center shift refers to the spatial migration of the overall mass center of the rock mass structure due to changes in the volume and density of rock blocks during the excavation of high slopes. The trend of axial force variation of the anchor bolts refers to the change in axial force data of the differentiated anchor bolts over time during the anchoring process.

4. The method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence as described in claim 1, characterized in that, The abnormal points refer to the points that differ from the surrounding monitoring values ​​during the monitoring process, which are manifested as changes in displacement direction, sudden stress changes, and abnormal rates, representing early signs of potential risks and structural instability. The unbalanced location refers to a location within the anchorage area where the direction of structural force and displacement response changes abruptly, indicating an uneven distribution of force and potential structural instability.

5. The method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Obtain the profile density map, lithological classification map and construction progress map, extract the volume and density values ​​of rock blocks within the coordinate range of the stage profile, configure the quality data frame in the corresponding area, compare the changes in the quality distribution of the stage, extract the regional quality status over time, and obtain the stage quality evolution sequence of the rock mass. S102: Based on the phased quality evolution sequence of the rock mass, extract the coordinate points of the phase boundary, compare the spatial positions of points with the same number, determine the direction change, extract the trajectory data of continuous position offset, divide the path according to the direction change trend, and obtain the set of structural position direction migration paths. S103: Call the set of migration paths of the structure's position and direction, retrieve the position of the centroid within the path, compare whether the transfer direction and the path direction are continuous, arrange the transfer segments with the same direction in coordinate order as a continuous path, and obtain the change characteristics of the slope structure migration path.

6. The method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the migration path change characteristics of the slope structure, extract the layout position and axial force time sequence change information of each group of anchors in the anchor layout diagram and stress monitoring data diagram, extract the change trend according to the layout number, and filter the numbered segments that continuously show linear trend turning points to obtain the axial force change trend labeling sequence. S202: Based on the axial force change trend labeling sequence, extract the coordinate positions of the numbered points, compare the displacement response direction with the adjacent areas, determine whether the direction is consistent, filter out the layout points with deviation in direction, extract the spatial distribution information, and obtain the direction response offset coordinate set; S203: Based on the directional response offset coordinate set, locate the number index position in the layout map layer, extract the coordinate range of adjacent numbered areas in the same layer, merge continuous areas into the same response area in sequence, output the layer coordinates corresponding to the area position, and obtain the location correspondence of stress anomaly distribution areas.

7. The method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence as described in claim 1, characterized in that, The specific steps for S3 are as follows: S301: Based on the location correspondence of the stress anomaly distribution area, extract the intersection area of ​​the slope structure trajectory and the stress boundary line segment in the layer, compare the structural offset direction and boundary extension direction vector at the intersection point, determine whether there is directional consistency, extract the intersection line segments with consistent direction, and obtain the set of intersection line segments with consistent direction. S302: Based on the set of intersection segments with consistent direction, extract continuous trajectory points on the layer according to coordinates, filter trajectory nodes that are consistent in spatial distribution direction, compare the points of trajectory segments before and after the direction deflection, identify the line segments that have not undergone direction deflection, and obtain a set of continuous trajectories in spatial direction. S303: Call the set of continuous trajectories in the spatial direction, extract the spatial angle changes and distance distribution range of the line segments in the layer, filter the segments that are consistent with the structural offset direction and stress boundary trend, output the trajectory number and layer coordinate range, and obtain the offset path and stress boundary related segments.

8. The method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the offset path and stress boundary associated segment, extract the shear deformation sequence, joint dip angle variation range and crack opening and closing state of the monitoring point along the path direction, analyze the directional scalar of adjacent monitoring points in the parameter sequence, filter out continuous response segments with consistent directions, and obtain a set of deformation segments with unified directions. S402: Based on the unified deformation segment set in the direction, identify the coordinate segment of the layer where the segment is located, extract the fluctuation pattern of the monitoring curve in the adjacent area, extract the response path offset position point at the continuous trend change node, analyze the spatial connection relationship between the offset points, and obtain the continuous response segment positioning index. S403: Based on the location index of the continuous response fragment, match the number and coordinate area in the layer, filter the path response sequence with continuous change pattern under the same number, remove the points that do not form periodic fragments, aggregate the corresponding index position and response direction information in the path sequence, and obtain the continuous abnormal trend distribution content.

9. The method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence as described in claim 8, characterized in that, The process of filtering path response sequences with continuous changes under the same number is as follows: extract the layer number and corresponding coordinate area included in the location index of the continuous response segment, compare the response direction of the monitoring points in the path with the same number point by point, identify the continuity characteristics of the response direction change between adjacent points, and determine whether a continuous path segment is formed based on the trend of direction maintenance and gradual change. The process of removing points that do not form periodic segments is as follows: identify response path segments, repeatedly compare the directional response changes of each point in the path sequence, determine whether there is a regular directional change, and remove the corresponding points from the sequence when there is a lack of directional continuity between adjacent points or random jumps in the path. The process of obtaining the index position and response direction information corresponding to the aggregated path sequence is as follows: based on the continuous points retained in the path, extract the spatial index and response direction in numerical order, compare the change state of the response direction, analyze the response segments with consistent directions, and obtain the corresponding spatial position sequence and direction information.

10. The method for predicting anchorage deformation during excavation of high rock slopes based on artificial intelligence according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Based on the continuous abnormal trend distribution, extract the segments with concentrated changes in the shear deformation rate sequence, call the corresponding index of the monitoring time series, analyze the continuity of adjacent data points in the time axis direction, filter the segments with the direction maintained, and obtain the shear rate continuity segment index. S502: Based on the shear rate continuation segment index, match the start and end times of the segment with the construction sequence node positions in the layer, analyze the response delay distribution points, filter segments with time difference structures with the preceding segment, and obtain the construction response time delay path set. S503: Call the construction response time delay path set, aggregate the spatial coordinate point sequence corresponding to the path, extract the path segments with consistent direction extension according to coordinate sorting, exclude segments with abrupt direction changes, and obtain the prediction result of rock slope anchorage deformation trend based on the relationship between trajectory continuity and position trend.

Images