Engineering geological investigation and monitoring system based on geophysics

Abstract

The invention belongs to the technical field of intelligent early warning, and particularly relates to an engineering geological investigation and monitoring system based on geophysics. The system comprises an on-site acquisition terminal, a station end and a cloud end, wherein the on-site acquisition terminal comprises two flexible conductive polymer geogrid sections and four regular dodecahedron geophysical stress sensors which are arranged along a filling and digging junction, and a fixed acquisition window is used for synchronously acquiring a conductive response, an optical fiber scattering response and a multidirectional stress response; the station end is used for forming basic acquisition data indexed by a unique identifier after finishing time alignment, exception elimination and paragraph neatening of the response; and the cloud end is used for constructing a unified scene grid by taking the basic acquisition data as input, setting a support anchor point at a sensor, and outputting a shear expansion evolution description organized along with a time sequence. According to the invention, the accuracy and reliability of geological investigation and monitoring under complex geological conditions are obviously improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of intelligent sensing system, and particularly relates to the field of intelligent early warning technology, and specifically relates to an engineering geological survey and monitoring system based on geophysics. BACKGROUND

[0002] Deep cutting and high filling embankment engineering widely exists in the construction of traffic infrastructure, especially in the construction of high-speed railway, highway and urban expressway. Since such engineering often involves large-scale excavation and filling of earthwork, the structure itself has complex mechanical response characteristics. In the prior art, the engineering field has generally adopted various monitoring and early warning means to control the risk of foundation settlement and slope collapse, such as commonly used displacement meters, inclinometers, strain gauges, inclinometers and fiber Bragg grating sensors. These devices can obtain local displacement or stress information to some extent to judge the stability of the engineering. However, these traditional monitoring methods are mainly point or line monitoring, which is difficult to form a continuous information field covering the entire filling and cutting interface area and soft belt. Therefore, under the conditions of complex geological environment and rapid construction, the change of a single monitoring point often cannot reflect the evolution law of the overall mechanical behavior, which is easy to cause the lag of geological survey and monitoring identification.

[0003] In recent years, with the development of optical fiber sensing technology and flexible conductive materials, researchers have tried to introduce optical fiber grating, distributed optical fiber sensing and conductive polymer materials into soil monitoring to enhance the monitoring accuracy and coverage. To some extent, this method breaks through the limitations of traditional point monitoring and can realize the perception of strain field and temperature field in continuous areas. However, existing researches mostly focus on the improvement of single monitoring means, and lack of comprehensive perception of the interaction between reinforcement deformation and soil stress transmission. More importantly, existing optical fiber or conductive monitoring often relies on preset crack surface assumptions or boundary conditions, such as assuming that the potential sliding crack surface is a plane or a simplified curved surface, so as to solve it by back calculation method. Although such assumptions are convenient for calculation, they are easy to misjudge the complex sliding mechanism, especially in the presence of multi-layer filling and cutting interface or soft interlayer, the crack surface is often highly irregular and cannot be accurately characterized by preset geometry simplification. In addition, in the data processing and early warning triggering link, the existing technology mostly adopts threshold type judgment method, that is, triggering early warning when the monitoring parameter exceeds the set threshold. This method is simple and easy to operate, but has two obvious problems: first, the threshold is often set based on experience or a small amount of historical samples, which is difficult to meet the individual needs of different engineering sites; second, the threshold type judgment cannot reveal the mechanical mechanism behind the monitoring data, resulting in insufficient explanation of the early warning, which makes it difficult for engineers to provide clear risk causes and disposal suggestions. Some advanced researches try to introduce machine learning models to identify risks using big data, but such methods are mostly "black box" discrimination, lack of transparent physical explanation, and limit the promotion in major engineering. SUMMARY

[0004] The main purpose of the present application is to provide a geophysical-based engineering geological survey and monitoring system, which realizes the fusion perception of multi-source sensing data, breaks through the limitation of relying on the preset sliding surface assumption, can transparently analyze the interaction between the reinforcement and the soil and the sliding evolution mechanism, identifies the subsidence and landslide risk in advance, and provides traceable, interpretable and disposal-suggested early warning results, significantly improving the accuracy and reliability of geological survey and monitoring under complex geological conditions.

[0005] To solve the above technical problems, the present application provides a geophysical-based engineering geological survey and monitoring system, which comprises: a field acquisition terminal, a station end and a cloud end; wherein the field acquisition terminal comprises two flexible conductive polymer geogrid segments arranged along the filling-digging interface and four icosahedral geophysical stress sensors, to synchronously obtain conductive response, optical fiber scattering response and multi-directional stress response in a fixed acquisition window; the station end is used to time-align, abnormity-eliminate and paragraphize the above responses, and then form basic acquisition data with unique identification index; the cloud end is used to input the basic acquisition data, construct a unified scene grid and set support anchor points at the sensors; according to a morphological sheet template library, an initial state reinforcement deformation field is identified and a boundary consistency constraint set is applied; a candidate morphological graph is generated and evaluated by a consistency evaluator, and after selecting the current morphological graph, morphological sheet replacement and connection correction are implemented according to a sliding window, and the consistency score is increased under the condition of maintaining the boundary consistency constraint set; in the iteration process, a shear dilation component graph and a transfer component graph are generated, and the area in which a closed feedback is formed between the same continuous band and the support anchor point is marked as a potential sliding zone candidate; when the convergence decision maker determines that the score change is below the set change threshold, the potential sliding zone candidate is solidified into a potential sliding zone graph, the main stem of the transfer component graph is solidified into a load transfer path graph, and a shear dilation evolution description organized in time sequence is output; the station end is also used to receive the potential sliding zone graph, the load transfer path graph and the shear dilation evolution description, and generate a comprehensive early warning level.

[0006] Further, the cloud end generates a unified scene grid with the start and end positions of the two flexible conductive polymer geogrid segments as boundaries; the unified scene grid contains a main index column along the direction of the flexible conductive polymer geogrid segment and an auxiliary index column across the filling-digging interface, and support anchor points are established at the positions of the four icosahedral geophysical stress sensors.

[0007] Further, according to indoor loading test data and historical engineering samples, a morphological sheet template library is generated in the cloud end; the morphological sheet template library contains at least the following types of morphological sheets: uniform stretching sheet, shear expansion sheet, end constraint sheet, cross-weak zone sheet, local buckling sheet and rebound recovery sheet, and each morphological sheet gives a corresponding relationship description between the conductive path response, the optical fiber scattering response and the multi-directional stress response.

[0008] Further, the cloud takes the unified scene grid as the bottom plate, and maps the basic collected data to the initial state reinforcement deformation field and the initial state soil support field; in the mapping process, the unique identification and the collection window are taken as the anchor points to complete the homology alignment, and the conductive response, the optical fiber scattering response and the multi-directional stress response are spliced into continuous segments; the boundary consistency constraint set is established at both ends of the unified scene grid, which requires the initial state reinforcement deformation field to maintain the same direction and phase as the response of the four regular dodecahedron geophysical stress sensors at both ends, and to maintain the consistent connection at the filling and excavation junction.

[0009] Further, the cloud takes the morphological piece template library to identify the initial state reinforcement deformation field, and generates a candidate morphological graph; the candidate morphological graph is composed of combinations of a plurality of morphological pieces on the unified scene grid, and a consistency scorer is established for each combination; the consistency scorer takes the boundary consistency constraint set, the conductive path change consistency and the multi-directional stress response consistency as the criteria to generate a single score.

[0010] Further, the cloud selects the combination with the highest consistency score in the candidate morphological graph as the current morphological graph, and slides a window of a morphological piece length from top to bottom according to the main index column on the current morphological graph, and performs morphological piece replacement and connection correction in the sliding window; when replacing, the boundary consistency constraint set should be maintained, the local morphological piece in the current morphological graph is replaced with a morphological piece that can improve the consistency score, and connection correction is performed on the replacement boundary; connection correction makes the adjacent segments tend to be coordinated in conductive path change and multi-directional stress response by inserting a rebound recovery piece or an end constraint piece at the replacement boundary; after completing a sliding, the current morphological graph is updated and the consistency score is recalculated, if the consistency score is improved, the replacement is retained, and then the sliding window is advanced to continue the cycle.

[0011] Further, the cloud generates a shear dilation component graph and a transfer component graph based on the current morphological graph, wherein the shear dilation component graph represents the contribution of reinforcement deformation to the volume change trend of the soil, and the transfer component graph represents the continuous path of load propagation along the unified scene grid; the generation of the shear dilation component graph and the transfer component graph should be based on the physical process of reinforcement deformation and load propagation, and the unified scene grid should be analyzed as a whole; a continuous band that meets the conditions of long-term directional consistency of the shear dilation component graph and the transfer component graph in the same continuous band, and forms a closed feedback condition with the support anchor response, is marked as a potential slip band candidate.

[0012] Further, when the sliding window completely covers the uniform scene grid and the consistency score change of the two consecutive rounds is below the set change threshold, the convergence resolver is triggered; the potential sliding shear zone candidate with the highest stability is solidified as the potential sliding shear zone map, the main stem with the highest connectivity in the transmission component map is solidified as the load transmission path map, and the ordered set of shear dilation components over time is solidified as the shear dilation evolution description; the three are taken together as transparent output for subsequent hierarchical warning calls.

[0013] Further, the station end receives the potential sliding shear zone map, the load transmission path map and the shear dilation evolution description and generates a comprehensive warning level; when the comprehensive warning level reaches a preset level, a treatment suggestion is synchronously pushed to the vehicle end and the scene; the treatment suggestion at least includes a monitoring encryption position, a traffic organization adjustment prompt and a temporary reinforcement operation sequence; the cloud end archives the comprehensive warning level and the transparent output for full life cycle backtracking and method review.

[0014] The geophysical-based engineering geological survey and monitoring system has the following beneficial effects: by arranging the flexible conductive polymer geogrid segment and the regular dodecahedron geophysical stress sensor on the field acquisition terminal, a distributed sensing network capable of simultaneously acquiring conductive response, optical fiber scattering response and multi-directional stress response is formed, so that more comprehensive monitoring information is obtained at the filling and excavation junction and the weak zone position. The station end performs time alignment, abnormality elimination and paragraph neatness processing on the collected data, ensuring the stability and comparability of the input data. The cloud end further constructs a uniform scene grid and sets support anchor points, performs segment identification on the initial state reinforcement deformation field with the morphological sheet template library, and applies boundary consistency constraint sets, so that the reconstructed morphological evolution process has clear physical constraints. Through the consistency score of the candidate morphological map, the replacement and connection correction of the sliding window, the overall matching degree can be gradually improved, and the shear dilation component map and the transmission component map are generated in the iteration process. In particular, when the shear dilation component map and the transmission component map form a closed feedback with the support anchor points in the same continuous zone, it can be marked as a potential sliding shear zone candidate. Finally, the potential sliding shear zone map and the load transmission path map are solidified under the judgment of the convergence resolver, and the shear dilation evolution description organized in time sequence is output. The station end generates a comprehensive warning level accordingly, realizing the whole process transparency from monitoring, identification to warning. Compared with the existing methods relying on experience threshold or preset crack surface assumption, the beneficial effects of the present application lie in the fusion of multi-source data, dynamic inverse evolution without sliding crack surface assumption, whole process transparent analysis and traceable evidence chain, which significantly improves the accuracy and advance of geological survey and monitoring identification for deep excavation and high fill embankment, and can provide targeted and operable decision support for on-site disposal. BRIEF DESCRIPTION OF DRAWINGS

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

[0016] Figure 1 The system structure schematic diagram of the geophysical-based engineering geological survey and monitoring system provided by the embodiments of the present application is shown in the figure. Figure 2 The potential sliding zone evolution process multi-index collaborative monitoring curve schematic diagram provided by the embodiments of the present application is shown in the figure. Figure 3 The load transfer path and form sheet template library identification response curve schematic diagram provided by the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0017] The method of the present application will be further described in detail below in combination with the drawings and the embodiments of the present application.

[0018] REFERENCE Figure 1 The geophysical-based engineering geological survey and monitoring system comprises a field acquisition terminal, a station end and a cloud end. The field acquisition terminal comprises two flexible conductive polymer geogrid segments arranged along the filling-digging interface and four regular dodecahedron geophysical stress sensors, so as to synchronously obtain conductive responses, optical fiber scattering responses and multi-directional stress responses by fixing and collecting window. The station end is used to time-align the above responses, remove abnormalities and paragraphize, and then form basic acquisition data with unique identification index. The cloud end is used to input the basic acquisition data, construct a unified scene grid and set support anchor points at the sensors. According to the form sheet template library, segment identification is performed on the initial state reinforcement deformation field and boundary consistency constraint set is applied. A candidate form diagram is generated and evaluated by a consistency evaluator. After selecting the current form diagram, form sheet replacement and connection correction are implemented according to the sliding window, and the consistency score is increased under the condition of maintaining the boundary consistency constraint set. In the iteration process, a shear dilation component diagram and a transfer component diagram are generated, and the area forming a closed feedback with the same continuous band and support anchor points is marked as a potential sliding zone candidate. When the convergence decision maker determines that the score change is below the set change threshold, the potential sliding zone candidate is solidified as a potential sliding zone diagram, the main stem of the transfer component diagram is solidified as a load transfer path diagram, and the shear dilation evolution description organized in time sequence is output. The station end is also used to receive the potential sliding zone diagram, the load transfer path diagram and the shear dilation evolution description, and generate a comprehensive warning level.

[0019] In actual implementation, first, the two flexible conductive polymer geogrid segments at the filling-digging junction are arranged, including: laying one flexible conductive polymer geogrid segment on each side of the normal direction of the filling-digging junction line, so that the axial direction of the two flexible conductive polymer geogrid segments is substantially parallel to the junction line, and the start and end positions of the two flexible conductive polymer geogrid segments are aligned with each other in the plane projection, so as to form a pair of main index columns in the unified scene grid in the subsequent process. Each flexible conductive polymer geogrid segment has embedded a conductive path and a fiber channel in the same bundle when leaving the factory, and forms a continuous electrical connection and optical connection after end crimping. In order to avoid the influence of non-target disturbance at the boundary on the conductive response and the fiber scattering response, a buffer segment not participating in interpretation is extended at each end of each flexible conductive polymer geogrid segment, and the buffer segment is only used to stabilize the end stress and suppress the clamp effect. Four regular dodecahedron geophysical stress sensors are arranged near the center of the filling-digging junction, near the upstream end of the flexible conductive polymer geogrid segment, near the downstream end of the flexible conductive polymer geogrid segment, and at a relatively deep position below the junction line, so that the four positions can cover the potential continuous band and form a stable geometric support in space, which is used to establish a support anchor point in the unified scene grid in the subsequent process. The regular dodecahedron shape has a symmetrical surface distribution in all directions, which facilitates the internal direction-finding element to obtain similar sensitivity in different stress directions, thereby reducing the direction bias in the reconstruction of multi-directional stress response.

[0020] The two flexible conductive polymer geogrid segments and the four regular dodecahedron geophysical stress sensors are respectively connected to the acquisition channel of the field acquisition terminal through electrical connection and optical connection. All connection points are subjected to pouring and waterproof treatment, and the connection path avoids sharp bends and heavy pressure as much as possible, so as to reduce the abnormal fiber scattering and unstable conductive path contact caused by local bending. The built-in clock of the field acquisition terminal is unified through satellite time service or wired time service; after the time service is completed, the field acquisition terminal broadcasts the current sampling beat, so that the two flexible conductive polymer geogrid segments and the four regular dodecahedron geophysical stress sensors start the fixed acquisition window at the same beat, thereby ensuring the co-location of the conductive response, the fiber scattering response and the multi-directional stress response on the time axis.

[0021] The fixed acquisition window selects constant duration and constant step mode (overlap or no overlap can be used), which can obtain comparable response fragments at different time periods. Constant duration facilitates the division of conductive response, fiber scattering response and multi-directional stress response into records with consistent length, and constant step mode facilitates time alignment and paragraph alignment at the station end. The fixed acquisition window uses uniform numbering, which is incremented by the field acquisition terminal according to the beat order to avoid time label ambiguity across devices. The field acquisition terminal triggers sampling at the beginning of each fixed acquisition window, and the three responses are in the sampling state at the same time. The core value of synchronous acquisition is to reduce the time registration error in the post-processing stage. Once the time registration error is reduced, the subsequent candidate morphology diagram generation is more likely to meet the boundary consistency constraint set, because the phase relationship between responses is more stable, avoiding false phase difference caused by asynchronization, thereby improving the recognition ability of the consistency scorer to the real morphology piece. Within the fixed acquisition window, the field acquisition terminal polls and scans the conductive path and optical fiber channel of each flexible conductive polymer geogrid segment according to the preset order, and simultaneously samples the three stress directions of the four icosahedral geophysical stress sensors in parallel. In order to reduce electromagnetic interference and optical interference, stable excitation and stable sampling rhythm are used in the window, and the gain and pulse settings are not changed in the middle of the window. Stable settings avoid human-induced amplitude drift, so that different channels in the same fixed acquisition window remain comparable, thereby improving the effectiveness of station end exception rejection. After the fixed acquisition window is completed, the field acquisition terminal immediately completes the window closure verification, including whether the record is complete, whether the timestamp is continuous, whether the trigger number is consistent, and whether all channels return valid data. After the window closure verification is passed, the three types of responses in the window are packaged and labeled with a unique identifier, waiting for the station end to receive.

[0022] Each flexible conductive polymer geogrid segment is divided into several collection nodes along the length direction, and the field collection terminal polls all the collection nodes in turn within the fixed collection window. The polling order is consistent in each fixed collection window, which facilitates the station end segment alignment. When first put into operation, the field collection terminal records the initial conductive value of each collection node as the baseline. In order to reduce the influence of temperature change on the conductive response, a conductive path that is not subjected to external force is connected in parallel with the buffer segment of each flexible conductive polymer geogrid segment as a reference path. The reference path and the working path are read at the same time in the same fixed collection window, and the station end calculates the relative change and offsets the common drift caused by temperature and power supply fluctuations accordingly. The advantage of this arrangement is that the reference path reflects the common factors of the environment and equipment, and the working path reflects the characteristic changes of the flexible conductive polymer geogrid segment with stress, and the subtraction of the two is closer to the true stress-related conductive response. The field collection terminal performs a short excitation stability test on each collection node at the beginning of the fixed collection window. If there is a stepwise discontinuity or periodic jitter that is inconsistent with the reference path, it is determined that the contact is poor or the terminal is loose, and is marked as suspicious in the fixed collection window. The suspicious marking makes the station end not to take it as an anchor point for forced alignment when time alignment, avoiding lowering the overall consistency score due to false alignment. When the conductive response of a collection node appears a significant mutation and is irreversible, and the optical fiber scattering response at the corresponding position appears a decrease or loss of echo energy, the field collection terminal marks the position as a suspected fracture, so that the station end sets the segment boundary at the position when segment alignment, preventing the data before and after the fracture position from being incorrectly spliced.

[0023] The field acquisition terminal uses a wideband light pulse to excite the optical fiber channel in the two flexible conductive polymer geogrid segments, and collects the backscattering generated along the light path position. To ensure comparability between different fixed acquisition windows, the amplitude and width of the excitation pulse remain unchanged under the same working conditions. Within the same fixed acquisition window, the field acquisition terminal selects the main echo combination with stable position and maximum energy proportion from the echo sequence as the effective echo of this window. Selecting the echo combination with stable position can reduce the random peak interference caused by multipath and microbending, and selecting the echo combination with maximum energy proportion can improve the consistency of positioning. When the position is stable and the energy proportion meets the two conditions at the same time, it can be ensured that the echo is mapped to the stable main index column in the subsequent unified scene grid. When the optical fiber channel of the two flexible conductive polymer geogrid segments has a joint, the field acquisition terminal retains a certain number of echo samples on both sides of the joint for splicing verification. Splicing verification is performed between two consecutive fixed acquisition windows. If the main echo combinations on both sides of the joint remain continuous in position and shape, it is determined that the splicing is stable, thereby ensuring the continuity of the optical fiber scattering response across the window. This continuity directly improves the accuracy of the station segment alignment. For isolated peaks caused by external light pulsation or end reflection, the field acquisition terminal uses the stable median method to suppress the isolated peaks by repeatedly measuring multiple times within the window and taking the stable median. The stable median method can suppress accidental peaks without introducing complex operations, while retaining the slow-changing echo structure related to stress, which is convenient for subsequent shape identification.

[0024] The regular dodecahedron shape provides uniformly distributed stress surfaces, and the internal direction-finding elements are arranged along the symmetry direction. The field acquisition terminal simultaneously reads three independent direction signals within the fixed acquisition window and records the direction combination in this window. The purpose of choosing the regular dodecahedron shape is to balance the direction sensitivity and reduce the direction bias caused by minor deflection during installation. After installation, a direction calibration is performed by using a standard load with a known direction to align the direction combinations of the four regular dodecahedron geophysical stress sensors to a common reference. After direction alignment, even if there is a small attitude change in the field, the relative relationship of the three direction signals can still remain stable, thereby making it easier for the subsequent generated multi-directional stress response to be consistent with the conductive response and optical fiber scattering response on the two flexible conductive polymer geogrid segments. When the baseline slowly drifts due to temperature or humidity changes, the field acquisition terminal records a short reference segment as the zero reference of the window within the fixed acquisition window. The introduction of the short reference segment enables the multi-directional stress response of each fixed acquisition window to be compared at the same reference level, reducing false positives caused by environmental changes.

[0025] The two flexible conductive polymer geogrid segments can be arranged in a staggered manner with one high and one low, so that the upper layer is closer to the road disturbance, and the lower layer is closer to the soft zone response. The staggered arrangement has the advantages of improving the coverage probability of the continuous zone, while maintaining the consistency of the start and end positions of the two flexible conductive polymer geogrid segments in the plane projection, facilitating the formation of paired primary index columns in a unified scene grid. The positions of the four regular dodecahedron geophysical stress sensors can be arranged in a diamond shape, i.e. one on the upstream and downstream of the boundary line, and one on the upper and lower sides of the boundary line normal direction. Diamond arrangement can improve the resolution of direction change without increasing the number, which is conducive to forming a stable support anchor point set of stress change in a unified scene grid. The fixed acquisition window can be in a partially overlapping sliding manner, for example, in the case of a fixed acquisition window duration, the step is set to a smaller proportion of the fixed acquisition window duration. The partially overlapping sliding manner can enhance the capture probability of rapid change events, while still maintaining the simplicity of station paragraph alignment. Temperature compensation of conductive response can be achieved by setting an independent reference line in the same soil layer outside the two flexible conductive polymer geogrid segments. The independent reference line does not bear external force for a long time and is only used to reflect environmental changes. This way is more representative than the reference path next to the buffer segment in a site with a large temperature gradient. The echo selection of optical fiber scattering response can introduce a multi-threshold stability check within the fixed acquisition window, i.e. in addition to energy proportion and position stability, a morphological stability check item is added. The morphological stability check item is used to identify whether the overall morphology of the echo envelope remains consistent, which can further suppress false peaks caused by accidental micro-bending. When the field vibration is strong, a layer of high-damping pad can be added to the bottom of the four regular dodecahedron geophysical stress sensors, and a short-time vibration memory zero operation can be performed after installation. High-damping pad reduces the coupling of high-frequency noise to multidirectional stress response, and vibration memory zero makes the sensor return to a stable state before the start of the fixed acquisition window, thereby ensuring the repeatability of the multidirectional stress response within the window.

[0026] The station end receives the fixed acquisition window data packaged by the field acquisition terminal, the fixed acquisition window is numbered with Arabic numerals, and has a unique identifier. The station end first sorts the fixed acquisition window numbers vertically, and then arranges them horizontally according to the channel sequence, to ensure that the positions of the three types of responses in the same fixed acquisition window correspond one by one. The station end takes the start time and end time of the fixed acquisition window as the alignment boundary, and takes the fixed acquisition window number as the alignment anchor. This can avoid misalignment caused by slight drift of the internal clock of different devices, because the number of the fixed acquisition window comes from the same trigger sequence and is naturally continuous on the timeline. The station end verifies whether the three types of responses in the same fixed acquisition window are returned under the same number one by one. If a channel is missing, the station end does not perform time extrapolation on the channel, but marks the channel as missing in the fixed acquisition window, and maintains the time integrity of the remaining channels. This can prevent false phases introduced by manual completion, and facilitate the formation of stable consistency relations subsequently. The station end checks whether the numbers of two adjacent fixed acquisition windows are continuous and the time interval is the expected value. If there is a repeated number or a number jump, the station end marks the position as suspicious and takes the position as a natural paragraph boundary when paragraphing. The purpose of this is to localize potential synchronization abnormalities and reduce the spread of abnormalities to a longer time range.

[0027] The station end compares the common changes of the conductive response and the optical fiber scattering response of the two flexible conductive polymer geogrid segments in the same fixed acquisition window. If the conductive response at a certain position shows a one-time irreversible jump, and the optical fiber scattering response at the same position shows energy decline or echo missing, the station end marks the position as a break and takes the break position as a segmentation point for paragraphing. The reason for using double evidence judgment is that the mechanisms of environmental interference of the electrical channel and the optical channel are different, and simultaneous abnormalities of the two can better indicate structural events. The station end classifies the abnormalities into three categories: missing, jump and isolated peak. Missing means that a channel does not return valid data in a fixed acquisition window; jump means that the amplitude changes irreversibly in a short time; isolated peak means that only a non-continuous peak appears in a single or a small number of samples. Different categories have different processing strategies: missing remains marked without interpolation, jump is disconnected when paragraphing, and isolated peak is replaced by the median value in the same window neighborhood without changing the overall shape. Such classification can neither over-modify the data nor over-magnify the occasional noise. The station end checks the direction consistency of the multi-directional stress response of the four regular dodecahedron geophysical stress sensors. If the conductive response and the optical fiber scattering response show significant changes, while the multi-directional stress response remains stable at the corresponding time and position, the changes are marked as suspicious environmental factors and are not used as segmentation points. This can avoid misjudging non-stress factors as structural changes. The station end deposits the types, positions and times of all abnormalities as quality marks, which are bound to the unique identifier, to ensure that subsequent processing is always carried out with the same abnormal view.

[0028] The station end takes the fracture mark, joint position, suspicious synchronization position, and the projection position of the four regular dodecahedron geophysical stress sensors in the unified scene grid as the priority boundary. These positions are taken as the reason for the start and end of the paragraph because they correspond to the key points on the structure or the uncertain points in time, and as the boundary, it can minimize the distortion caused by cross-border splicing. After determining the priority boundary, the station end makes the length of adjacent paragraphs as close as possible without breaking the continuity of the real event, so that subsequent mapping in the cloud according to the unified scene grid can obtain uniform coverage. Paragraph length specification is conducive to the cloud to perform smooth segment splicing on the main index column, reducing the morphological piece matching deviation caused by paragraph length difference. The station end ensures that the conductive response, fiber scattering response and multi-directional stress response in the same paragraph are co-located in time and position, and the sample count of the three types of responses is consistent. The co-location of the three types of responses enables the cloud to compare the corresponding relationship of the three types of information in the same spatial unit, thereby improving the reliability of the piece identification. The station end generates a unique identifier for each paragraph, including the field collection terminal identifier, fixed collection window number range, channel number and paragraph number. The unique identifier is used for one-to-one mapping and backtracking in the cloud to avoid losing the source relationship in multiple processing processes. The station end organizes the paragraphs after the alignment, their quality marks, unique identifiers, time ranges and position ranges into the basic collection data and delivers them to the cloud, ensuring that the cloud can directly build a unified scene grid without further cleaning.

[0029] The cloud takes the start and end positions of the two flexible conductive polymer geogrid segments as boundaries to form the outer frame of the unified scene grid. The reason for this setting is that the two flexible conductive polymer geogrid segments are consistent with the direction of the potential continuous zone, and the outer frame is consistent with the main propagation direction of the data, which helps to form a stable main index column subsequently. The unified scene grid contains a main index column along the direction of the two flexible conductive polymer geogrid segments and an auxiliary index column across the fill-dig interface. The main index column is used to carry the continuous change along the direction of the two flexible conductive polymer geogrid segments, and the auxiliary index column is used to describe the connectivity between the two sides. The setting of the main index column and the auxiliary index column can convert the linearly laid data into a planar expression, which is convenient for identifying the combination of morphological pieces across the fill-dig interface. The four regular dodecahedron geophysical stress sensors each occupy a unique position in the unified scene grid, and the cloud sets support anchor points at these positions. The support anchor points have two functions, one is to provide a directional reference for subsequent consistency judgment, and the other is to provide stable anchoring for the connection of morphological pieces near the boundary. When the unified scene grid has stable anchoring at these positions, the combination across the fill-dig interface is more likely to form closed feedback, reducing drift. The cloud maps the paragraph position of the basic acquisition data to the main index column and the auxiliary index column of the unified scene grid, ensuring that each record can be positioned in a specific grid cell. The meaning of coordinate parity is to allow data from different sources to be compared under the same spatial reference, so that the subsequent piece identification and boundary consistency constraint set application have a common coordinate context.

[0030] The cloud maps the two flexible conductive polymer geogrid segments at the same fixed collection window segment to the corresponding primary index column, and describes the continuous deformation of the reinforcement by the common change of the conductive response and the optical fiber scattering response. The reason for using common change instead of a single indicator is that the conductive response is more sensitive to path continuity and stretching, and the optical fiber scattering response is more sensitive to local micro-bending and echo shape. The superposition of the two can distinguish between local disturbance and overall deformation, so that the initial state reinforcement deformation field is more sensitive to the true stress. The cloud preferentially starts segment identification near the support anchor point. The three types of responses near the support anchor point are more stable, and as starting segments they can reduce the influence of mispairing global propagation. The specific method is to try to match the end constraint piece and the uniform stretching piece in the shape piece template library at the position of the primary index column where the support anchor point is located, to form a starting combination that can be constrained by the two end boundaries at the same time. The shape piece template library at least includes uniform stretching piece, shear expansion piece, end constraint piece, cross-weak zone piece, local buckling piece and rebound recovery piece. The cloud uses the order from near to far on the primary index column to divide the initial state reinforcement deformation field into candidate segments, and for each candidate segment, it finds the shape piece in the shape piece template library that best explains the common change of the conductive response and the optical fiber scattering response. The basis for selecting the "best explanation" is that the two types of responses are consistent in direction, consistent in change trend and consistent in amplitude under the shape piece. This can improve the correspondence between the segment and the true stress path, and reduce the overfitting to noise.

[0031] When a segment on the main index column needs to cross to another segment of flexible conductive polymer geogrid at the fill-dig interface, the cloud first tries to maintain continuity during the crossing process by combining the cross-weak zone segment and the rebound recovery segment. The reason for trying these two forms of segments first is that both weakened continuous zones and segments that recover to their original stretched state can occur when crossing the fill-dig interface. Trying the combination of segments that can cover these two typical situations first can get a stable crossing result faster. The cloud applies a set of boundary consistency constraints at the two end boundaries and the fill-dig interface, requiring the initial state of the deformed field of the tendon to be consistent with the multi-directional stress response direction and phase of the four icosahedral geophysical stress sensors at both ends, and to maintain continuity at the fill-dig interface. The significance of applying the above constraints is to limit the combination of segments in a clear and verifiable boundary behavior, so that the segments will not appear unreasonable fractures or reversals in the global due to local optimization. When two adjacent segments are inconsistent under the set of boundary consistency constraints, the cloud prefers to keep the segment closer to the support anchor and replace the segment farther away from the support anchor that is not coordinated with the common changes of the three types of responses. The strategy of giving priority to the support anchor can allow the high-trust area to pull the low-trust area, thereby maintaining the stability of the overall form. After completing the segment identification and applying the set of boundary consistency constraints, the cloud obtains a preliminary combination composed of several form segments and labels it based on the unified scene grid. This preliminary combination provides a direct input for generating candidate form maps and conducting further consistency evaluation.

[0032] The uniform stretching piece is defined as: applying balanced stretching at the start and end positions of two flexible conductive polymer geogrid segments, so that the response along the main index column presents continuous and consistent slow change. Selecting this working condition can obtain a simple form and long coverage reference segment, which is used for subsequent definition and comparison of other forms of pieces. The end constraint piece is defined as: setting strong constraints at both ends, so that the proximal end appears obvious boundary strengthening effect. This setting emphasizes the locality of end effect, which is convenient for forming the characteristic profile of rapid rise at both ends and stable in the middle, which is used as stable anchoring when connecting segments. The rebound recovery piece is defined as: after experiencing a load, it is unloaded quickly, so that the response returns to the stable zone in a short time. The rebound recovery process reflects the elastic recovery and hysteresis characteristics of materials and soil, which can provide a transition piece from strong to weak and from disorder to stability, which is beneficial to eliminate adjacent discontinuity in the candidate form diagram. The shear expansion piece is defined as: applying lateral disturbance on one side of the filling and excavation junction, and the other side is relatively stable, so that the response along the auxiliary index column presents continuous change from one side to the other side. This process captures the propagation characteristics of "originating from local to expanding outward", which is used to identify the directionality and expansibility of continuous bands. The cross-weak band piece is defined as: setting a controllable weak band between two flexible conductive polymer geogrid segments, inducing the responses on both sides to appear different but connectable continuous structures. The weak band makes the cross behavior more significant, which is convenient for summarizing the "first attenuation and then continuation" cross rule, which is used to guide the connection judgment at the filling and excavation junction. The local buckling piece is defined as: setting local disturbance in the middle of a flexible conductive polymer geogrid segment, which induces short-scale concentrated change without forming large-scale transmission. This form emphasizes the characteristics of "strong local and weak global", which is used to exclude the uniform stretching piece and the shear expansion piece.

[0033] The segment positioning includes: in a fixed acquisition window sequence, according to the window number and the response time sequence of the four regular dodecahedron geophysical stress sensors, positioning the most stable response interval as a candidate segment, and preferentially selecting the interval close to the four regular dodecahedron geophysical stress sensors, because the time consistency and direction consistency of these positions are higher, and the interference of drift on the shape judgment can be reduced. The feature description generation includes: for each candidate segment, respectively extracting five types of descriptions including direction consistency, amplitude continuity, phase sequence, echo stability and continuity across two flexible conductive polymer geogrid segments from the conductive response, optical fiber scattering response and multi-directional stress response to form a shape description list. The advantage of using multi-source description without relying on a single source is mutual verification, which can distinguish environmental noise from structural events. Similarity aggregation and representative selection include: in each target type, after aligning the candidate segments in time, similarity aggregation is performed, the segment closest to the overall sample is selected as the initial template, and the segments significantly different from the initial template are removed, and a number of stable median segments are retained for merging to obtain the template entry of the shape. By using this convergent way from inside to outside, the dominance of extreme samples on the template can be avoided, and the availability of the template in more scenarios can be ensured. The template entry compilation includes: for each template entry record, the generation condition, the applicable range, the corresponding relationship description between the conductive response, the optical fiber scattering response and the multi-directional stress response, and the typical distribution position in the unified scene grid are recorded, which are directly called by the subsequent consistency scorer.

[0034] With the basic acquisition data output by the station as input, the data is mapped to a unified scene grid in the cloud, and the support anchor point at the position of the four regular dodecahedron geophysical stress sensors is taken as the starting point, and the interval that remains direction consistent and phase stable in multiple fixed acquisition windows is expanded to both ends. Preferably, it starts from the vicinity of the support anchor point, because the trigger sequence in this area is the most reliable, and can reduce mislabeling. In the candidate interval, respectively delete single-source data for playback, and observe the change of the interval recognition by the consistency scorer. If the recognition result is unstable after deleting a source, it means that the source has a key contribution to the shape piece, and the description points of the source should be retained and written into the template entry. In this way, the necessity of multi-source description is clear, and the single-point failure of the template in the field is prevented. The difference characteristics of the same shape under different working conditions are written into the same template entry in the form of optional description, for example, the expansion speed of the shear expansion piece is slower in the rainy season, and the expansion speed is faster in the construction disturbance stage. The difference is described as an optional description in the same entry, which helps to reduce the number of templates and expand the coverage range.

[0035] A digital twin scene is established in the cloud, using a unified scene grid, the start and end positions of the two flexible conductive polymer geogrid segments, and the support anchor point positions of the four regular dodecahedron geophysical stress sensors. A variety of soft belt distributions and external disturbance sequences are set to generate response sequences covering extreme boundaries. From the synthesized response, intervals that meet the consistent direction, stable echo, and coherence across the two flexible conductive polymer geogrid segments are selected as candidate segments, and rare combinations such as multi-layer weakening and alternating recovery of soft belt pieces and multi-point distribution of local bending pieces are highlighted. The synthesized candidate segments are compared with indoor and field samples, and after confirming the consistency of the key descriptions, they are included in the morphological segment template library. By comparing the simulation deviation within an acceptable range, the template is made to be transferable on real data.

[0036] Several fixed acquisition windows are reserved as closed validation sets and are not involved in template extraction. On the closed validation set, template segments are positioned according to the unified scene grid, and the three-way consistency of conductive response, optical fiber scattering response, and multi-directional stress response is checked to see if it meets the description of the template item. If the consistency is insufficient, go back to the extraction stage to revise the description or re-aggregate. At the two end boundaries and the interface position of filling and excavation, verify whether the template item meets the boundary consistency constraint set. If the direction or phase of the four regular dodecahedron geophysical stress sensors is inconsistent, the template item is marked for use in specific areas only to prevent misuse in the full range. Assign a unique identification number to each template item, record the version, generation source, applicable range, and mutual exclusion relationship. For example, uniform stretching segments and local bending segments are mutually exclusive within the same paragraph in the same main index column to avoid contradictory combinations in the candidate morphological diagram. During engineering operation, the cloud continuously receives basic acquisition data, and if a new morphology that does not match existing templates but is stable and repetitive appears in the high-scoring zone of the consistency evaluator, start the incremental acquisition process to generate new template items or extend the optional description of existing items. All updates retain rollback records to ensure that in abnormal situations, the system can be rolled back to the previous stable version.

[0037] Uniform stretch piece: acquisition source: indoor uniform stretching, field long-term steady-state interval, digital twin steady-state scenario. Key description: consistent along the main index column direction, smooth amplitude variation, stable echo, the direction of the four regular dodecahedron geophysical stress sensors is consistent with the two ends. Effect and benefit: as a reference segment, it is used to correct the overall offset in the candidate morphology map, and improve the satisfiability of the boundary consistency constraint set. Shear expansion piece: acquisition source: indoor unilateral disturbance expansion, field construction side disturbance period, digital twin lateral propagation scenario. Key description: continuous advancement from one side to the other, the direction is consistent first, then differentiated and then consistent again, the transmission of the four regular dodecahedron geophysical stress sensors on the expansion path is more obvious. Effect and benefit: reveal the origin and propagation of continuous belts, which is beneficial to determine the optimal growth direction in the candidate morphology map. End constraint piece: acquisition source: indoor strong constraint at both ends, field rigid connection part, digital twin rigid end scenario. Key description: rapid lifting at the end, slow change in the middle, fiber scattering echoes are more concentrated at the end, the direction and amplitude of the four regular dodecahedron geophysical stress sensors near the end are stronger. Effect and benefit: used for segment connection and boundary stability, making the candidate morphology map more easily closed at both ends. Cross-weak belt piece: acquisition source: indoor controllable weakening belt, field soft interlayer interval, digital twin multi-layer weakening scenario. Key description: the continuity of the two flexible conductive polymer geogrid segments appears to decay and recover before and after the weakening belt, the fiber scattering echo pattern is rearranged within the weakening belt, and the four regular dodecahedron geophysical stress sensors form a response difference from near to far. Effect and benefit: provide a reliable template for crossing the fill-dig interface, making the cross-index connection more stable. Local buckling piece: acquisition source: indoor local disturbance, field local anomaly point, digital twin short-scale concentrated scenario. Key description: high amplitude concentration appears in a short range and does not continue to transmit to the far end, conductive response and fiber scattering response appear sharp changes at the same location, the overall direction of the four regular dodecahedron geophysical stress sensors changes little. Effect and benefit: help the candidate morphology map distinguish between "strong local" and "strong continuous", and avoid mistaking local events as continuous belts. Rebound recovery piece: acquisition source: indoor loading-unloading process, field recovery segment after short-time impact, digital twin rapid unloading scenario. Key description: from high to stable, amplitude gradually converges, echo configuration from complex to simple, the direction of the four regular dodecahedron geophysical stress sensors is consistent with the uniform stretch piece at the end of recovery. Effect and benefit: as a connection buffer, it reduces the boundary discontinuity of adjacent segments and improves the scoring stability of the consistency evaluator in the sliding window.

[0038] The cloud expands from the position of the support anchor on the uniform scene grid to both ends of the two flexible conductive polymer geogrid segments corresponding to the main index column to generate a candidate configuration graph. The support anchor is selected as the starting point because the conductive response, fiber scattering response and multi-directional stress response at this position are more stable, which can provide consistent direction and stable phase reference for the initial combination, and reduce the chain deviation caused by early mispairing. The candidate configuration graph is composed of configuration slices in the configuration slice template library on the uniform scene grid in the order of the segments. Each combination records the corresponding relationship with the two end boundaries and the filling and excavation junction to ensure that it can accept the test of the boundary consistency constraint set later.

[0039] The consistency scorer gives a single score to the candidate configuration graph. The score is based on three types of verifiable evidence: boundary consistency constraint set satisfaction: check the direction consistency, phase consistency and connectivity consistency of the two end boundaries and the filling and excavation junction. This can bind the candidate configuration graph with the actual boundary behavior and prevent local optimal slice combination from appearing in reverse or breaking on the global. Conductive path change consistency: compare the segment order of the candidate configuration graph on the main index column with the continuous change of the conductive response to see if they are in the same direction and have no repetition. The conductive response is sensitive to on-off and stretching, which can constrain the effectiveness of the arrangement of the configuration slices in the length direction. Multi-directional stress response consistency: near the support anchor and along the auxiliary index column direction, the direction combination of the multi-directional stress response is compared to require the segment transition and direction combination of the candidate configuration graph to match. This can avoid misjudgment caused by in-plane information alone and make the candidate configuration graph closer to the real stress path.

[0040] The cloud scores the candidate morphology map for consistency and selects the highest-scoring one as the current morphology map. The highest-scoring one is directly selected as the baseline for subsequent local optimization to avoid unnecessary adjustments from a weaker scheme. On the current morphology map, a sliding window of a morphology piece length is slid from top to bottom along the main index column. Setting the sliding window length to a morphology piece length helps to limit local adjustments within a single piece, which can be refined to a controllable scale without damaging the larger-scale structure that has passed the boundary consistency constraint set test. Each sliding window advance only affects the selection or connection of one piece, which facilitates accurate evaluation of the true gain in consistency score from a replacement. At the piece covered by the sliding window, the cloud enumerates replaceable morphology pieces from the morphology piece template library: first verify whether the replacement still satisfies the boundary consistency constraint set. If not, the replacement is immediately abandoned, because the boundary behavior is the first constraint for global stability, and cannot be traded for boundary distortion in exchange for local score improvement. If the boundary consistency constraint set is satisfied, the conductive path change consistency and multi-directional stress response consistency before and after the replacement are compared. Only when both are not reduced and at least one is improved is the replacement determined to be effective. This way, the apparent score improvement by sacrificing a source consistency can be avoided. For the replacement determined to be effective, the current morphology map is updated and the score change is recorded. When the score change is positive, the replacement is retained; when the score change is zero, the replacement is retained only if it can simplify subsequent connection correction; and when the score change is negative, the replacement is rolled back to the state before the replacement. This strategy takes score increment as the basic principle and structure simplification as the secondary benefit.

[0041] After the sliding window completes an effective replacement, the cloud performs connection correction on both sides of the replacement boundary to eliminate small discontinuities at the piece junction: when there is a slight misalignment in phase on both sides of the replacement boundary, a rebound recovery piece is inserted as a buffer piece. The rebound recovery piece has a transition feature from strong to weak and from chaotic to stable, which can smoothly connect two pieces without changing the overall direction, reducing the penalty in consistency score due to phase misalignment. When the replacement causes a stiffness mutation near the end, an end constraint piece is introduced to stabilize the response morphology at the end. This way, the end lifting or sinking can be avoided to mislead the middle piece, maintaining the continuity of the main index column. When the replacement involves crossing the fill-dig interface, the combination of a cross-weak band piece and a rebound recovery piece is used to connect the pieces, making the crossing section exhibit the feature of "first attenuation and then continuation", so as to smoothly transition with the continuous pieces on both sides. The common goal of connection correction is to preserve the score gain obtained, while eliminating the local seam effect under the boundary consistency constraint set.

[0042] After each round of sliding window replacement and junction correction, the cloud generates a shear-dilation component map according to the continuous changes of the current morphology map on the primary index column and the auxiliary index column, according to the expansion trend of the segment, the separation and reunion trend of the adjacent segment, and the complexity of the echo configuration. The specific method is: detecting the change from tight to loose or from loose to tight in the segment group on the primary index column as the sign of the volume change trend. Detecting the continuous advance from one side to the other on the auxiliary index column as the sign of the expansion of the shearing effect. Identify the conversion from complex to simple or from simple to complex in the echo configuration as the sign of the structure from disturbance to stability or from stability to disturbance. The three types of signs described by the scene superimposed constitute the shear-dilation component map. The advantage of such construction is to unify the observable phenomena from multiple sources to the intuitive semantics of "expansion and convergence", which is convenient for comparison with the subsequent transmission component map.

[0043] The cloud takes the connected order of the segments in the current morphology map as the skeleton, finds the continuous path from one end boundary to multiple segments, and corresponds with the direction combination of the positions where the four regular dodecahedron geophysical stress sensors are located, forming a transmission component map. The specific method is: On the primary index column, preferentially select continuous bands with consistent segment direction and consecutive amplitude, to avoid including repeated rebound and flexion in the path. When the path needs to cross the fill-dig interface, only when the cross-section can match the direction combination of the supporting anchor point, the path can be extended to prevent the path from losing physical support when crossing. When there are two or more feasible paths for the same position, preferentially retain the path that is closer to the supporting anchor point and has a higher matching degree with the multi-directional stress response, to reduce ambiguity. The transmission component map obtained through the above steps embodies the continuous propagation skeleton of the load on the unified scene grid. The cloud superimposes the shear-dilation component map and the transmission component map, and finds the continuous band that meets the following conditions and marks it as a potential slip zone candidate: In the same continuous band, the shear-dilation component map remains in the same direction for a long time, showing stable expansion or stable convergence, to avoid short-term fluctuations. Stable changes in the same direction indicate that the internal state of the continuous band is continuously accumulating, with a tendency to form a through structure. The continuous band forms a closed feedback with the supporting anchor point. The meaning of closed feedback is: the continuous band departs from the vicinity of a supporting anchor point, and after passing through the segment junctions on the primary index column and the auxiliary index column, another supporting anchor point or the same supporting anchor point appears in time and amplitude consistent with the continuous band, and the sequence and direction combination between them match. Forming a closed feedback means that the continuous band is not just a local disturbance, but has interacted with the overall stress system, with penetration and traceability. The continuous band can be covered by the main stem of the transmission component map in the current morphology map. Only when the load transmission path can cross or extend along the continuous band, does it have the possibility to evolve to a higher risk state.

[0044] The region satisfying the above three conditions is marked as a potential candidate of the slip zone, and is fixed in the form of a continuous band on the unified scene grid for subsequent convergence decision and solidification.

[0045] After each sliding window replacement is completed, the cloud only accepts modifications that can bring consistent score gains and do not violate the boundary consistency constraint set. When the score gain tends to be small after several consecutive sliding window advances, the cloud keeps the current morphology map and no longer performs aggressive replacement, but only performs necessary splicing corrections to consolidate the continuous bands and closed feedback that have been formed. This avoids the back-and-forth fluctuations caused by frequent replacement in the convergence phase, and keeps the superimposed relationship between the dilatation component map and the transfer component map stable. For fragments near the support anchor points and stable contribution scores, the cloud sets a freeze flag, and they no longer participate in replacement when subsequent sliding windows pass. The benefit of the freeze strategy is to provide long-term stability for the core skeleton and reduce repeated modifications to high-confidence areas. If the score decreases after replacement in a non-frozen area, the cloud immediately rolls back to the state before replacement and records an unfavorable attempt at that location, and subsequently prioritizes different types of morphology combinations at the same location to shorten the convergence path. After each replacement or splicing correction is completed, the cloud synchronously updates the dilatation component map and the transfer component map, without delayed batch updates. The benefit of synchronous updating is that it can immediately capture the growth of continuous bands and the appearance of closed feedback, making the marking of potential slip zone candidates closer to the real-time evolution.

[0046] On the basis of top-down advancement, bottom-up rounds are added, and the side with a higher score is ultimately selected. Bidirectional advancement can reduce path preferences caused by single-direction traversal order. Smaller step sizes are used near support anchor points, and larger step sizes are used in areas far from support anchor points, to increase the number of local optimizations at critical locations while controlling overall computational load. In the case of close scores, several parallel branches of the current morphology map are temporarily retained, and only one is determined after clear differentiation between the dilatation component map and the transfer component map. Multiple candidates can reduce the risk of path locking caused by early selection.

[0047] Anchor point priority splicing: When there are multiple splicing schemes across the fill-dig interface, the scheme that can form a closed feedback between the two support anchor points is preferred, and the scheme with a suboptimal score but insufficient closed feedback is considered. This strategy directly serves the reliable marking of potential slip zone candidates.

[0048] The cloud continuously receives the score changes of the consistency score calculator after each round of the sliding window is completed. The cloud maintains a sequence of score changes and a time sequence for each continuous band. When the score change is lower than the set change threshold for a continuous number of rounds, the convergence decider enters the judgment stage. The continuous observation method can filter accidental single jitter and avoid premature termination of effective local optimization due to one-time score fluctuations. The convergence decider adopts dual judgment of amplitude stability and direction stability. The amplitude stability requires the absolute value of the score change to be continuously lower than the change threshold. The direction stability requires the score change to no longer frequently flip between adjacent rounds. The reason for adopting dual judgment is that looking at the amplitude alone may misjudge high-frequency back-and-forth adjustments as stable, and adding direction constraints can ensure that the current pattern chart has entered a stable plateau, and further replacement is difficult to bring structural benefits. When the judgment passes, the convergence decider identifies the fragment set that contributes most to the score, determines the frozen range in combination with the support anchor neighborhood, and determines the frozen range. The frozen range no longer performs pattern fragment replacement, and only allows necessary splicing correction. By freezing the core area, the established continuous band and closed feedback can be fixed to prevent subsequent minor attempts from destroying the stable structure that has been obtained.

[0049] The cloud performs a one-time cleaning on the continuous bands marked as potential candidate for the sliding fracture zone. The cleaning includes three operations: short gap bridging, isolated branch pruning, and boundary end trimming. Short gap bridging is used to fill small gaps with consistent directions at both ends, making the continuous band form a coherent expression on the unified scene grid; isolated branch pruning is used to delete branches that only appear locally and do not form closed feedback with support anchors, to avoid mistaking random disturbances as main structures; boundary end trimming is used to align the end of the continuous band to the nearest support anchor or the boundary at both ends, so that the end behavior is consistent with the boundary consistency constraint set. After cleaning, the cloud rechecks the closed feedback for each continuous band. The closed feedback requires that the propagation of the continuous band along the main index column and the auxiliary index column has a clear chronological correspondence with the response of at least one support anchor, and the amplitude trend and direction combination are coordinated. Through rechecking, pseudo-closed feedback introduced by local bridging can be removed, ensuring that the solidified object can be repeatedly verified by external observation. The cloud extracts the geometric skeleton of each continuous band on the unified scene grid, and determines the band width on both sides of the skeleton according to the response continuity. The geometric skeleton facilitates subsequent superposition of load transmission path diagrams, and the band width is used to express the spatial influence range. Using the skeleton plus band width approach can meet the needs of intuitive display and accurate superposition. The cloud assigns a unique identification number to each solidified continuous band, records the generation time, the range of the fixed collection window involved, the set of support anchors involved, and the quality label, forming a potential sliding fracture zone map. The introduction of version solidification and time stamp backtracking facilitates the comparison of evolution and treatment effect at different times by the station end and subsequent operation and maintenance.

[0050] The cloud finds the most coherent path set in the transfer component graph that connects the two ends of the boundary or the two support anchors. Coherence means that the direction of the segment is consistent, the amplitude changes smoothly, and the direction can be combined with the support anchor when crossing the interface. Coherence is preferred because load propagation tends to spread along the path with smaller impedance and more stable direction in real scenarios. For sections with branches, the cloud uses three decision rules: support anchor priority, early response priority, and cross-border stability priority. Support anchor priority allows the path to fit the most reliable part of the observation; early response priority captures the earliest main road of propagation; cross-border stability priority ensures that there is no direction break when crossing the interface. The combination of the three rules can quickly determine the representative main road without introducing complex operations. When short-scale discontinuities appear on the main road, the cloud performs minimal repair without violating the boundary consistency constraint set. After repair and confirmation, the cloud solidifies the main road into the load transfer path graph and records the start point, end point, passing support anchor, cross-border auxiliary index column, and fixed acquisition window range involved.

[0051] The cloud annotates three types of states for the dilatancy component graph on each grid cell of the unified scene grid: expansion, stability, and recovery. Expansion indicates a trend of continuous growth, stability indicates a small change and consistent direction, and recovery indicates a fall from high to stable state. Using three types of state encoding, complex continuous changes can be compressed into reusable time sequence labels, facilitating joint interpretation with potential slip band graph and load transfer path graph. The cloud identifies three key events in the state encoding sequence: start, accelerate, and breakthrough. Start indicates the first appearance of sustained expansion; accelerate indicates that expansion significantly increases in a short period of time; breakthrough indicates that expansion is spatially aligned with the potential slip band graph and load transfer path graph and remains continuous. Binding events and spatial alignment can depict the evolution of dilatancy from both time and space dimensions. The cloud generates dilatancy evolution description at grid level, continuous band level, and global level respectively. Grid level is used for high-resolution positioning, continuous band level is used to describe the overall trend within the band, and global level is used to reflect the overall risk background. The multi-scale approach facilitates the station to reference the same set of descriptions at different decision levels without repeated calculation. The cloud organizes events and states in chronological order into records to form the dilatancy evolution description. Each record contains time, location range, state category, related potential slip band graph identification number, and related load transfer path graph identification number. By explicitly associating identification numbers, the station can spatially locate and play back single evolution records.

[0052] After receiving the potential slip band map, load transfer path map and shear dilation evolution description, the station end first verifies the consistency of the time range and unique identification number of the three at the local. Verification passed can confirm that the cloud output and the basic collection data of the station end are in the same spatio-temporal context, avoiding judgment deviation caused by inconsistent versions. The station end superimposes the potential slip band map and the load transfer path map on the base map of the unified scene grid, and locates the events of the shear dilation evolution description to the corresponding continuous band and path. Superimposed interpretation focuses on three types of relationships: the degree of coincidence of continuous bands and paths, the time sequence of events at the coincidence, and the diffusion trend of events at the non-coincidence. The three types of relationships jointly determine whether the risk has connectivity, whether it develops along the main stress channel, and whether it is expanding to new areas. The station end makes layered judgments according to threshold rules, trend rules and mechanism rules. The threshold rule checks whether the spatial range of the potential slip band map and the main length of the load transfer path map reach the range corresponding to the given level. The trend rule checks whether the acceleration and breakthrough in the shear dilation evolution description appear and remain in a short period of time. The mechanism rule checks whether a stable closed feedback is formed between the continuous band and the support anchor and is in the same direction as the load transfer path map. Using layered judgment can combine the three types of concerns into a clear judgment process. When the three types of rules jointly meet the conditions of a certain level, the station end generates a comprehensive early warning level, and writes the comprehensive early warning level and the corresponding potential slip band map identification number, load transfer path map identification number, and key events of the shear dilation evolution description into the unique identification index system. The purpose of writing back is to provide traceable reference points for subsequent disposal suggestions, post-mortem analysis and cross-period comparison.

[0053] As Figure 2As shown, the potential slip surface evolution monitoring graph of the present application shows the time evolution characteristics of three key technical indicators in the geophysical-based engineering geological survey and monitoring system. The abscissa represents the monitoring time (days), and the ordinate represents the dimensionless index, the value range is 0-1.4. The graph contains three main curves: the shear dilation component evolution curve (thick solid line), the transfer component evolution curve (medium thick solid line) and the boundary consistency score curve (thin solid line). The shear dilation component evolution curve reflects the contribution degree of reinforcement deformation to the volume change trend of soil body, which keeps at a low value level below 0.1 in the early stage of monitoring, and rises rapidly around the 15th day, reaches the peak value 1.35 at the 30th day, and then falls slightly and tends to be stable. The transfer component evolution curve represents the continuous path strength of load propagation along the unified scene grid, and its change rule is basically the same as that of the shear dilation component, but the peak value appears slightly earlier, about the 28th day reaches the maximum value 1.25. The boundary consistency score curve represents the satisfaction degree of boundary constraint condition in the morphological sheet replacement process, and the curve changes relatively gently, reaches the peak value 1.15 at the 25th day and remains relatively stable. Two key threshold lines are set in the graph: the critical threshold 1.2 (thick dotted line) and the early warning threshold 0.6 (thin dotted line). When any monitoring indicator exceeds the critical threshold 1.2, the system determines that the potential slip surface enters the critical state; when the indicator exceeds the early warning threshold 0.6, the early warning program is started. The whole monitoring process can be divided into four typical stages: initial development period (0-10 days), rapid development period (10-25 days), critical state period (25-32 days) and stable period (after 32 days). In the critical state period, the convergence decider continuously evaluates the score change, and when the consistency score change of two consecutive rounds is below the set threshold, the current potential slip surface candidate is solidified as the final potential slip surface graph.

[0054] As Figure 3As shown, the load transfer path and shape template library recognition response curves of this invention detail the core working mechanism of the cloud-based shape template recognition algorithm. The horizontal axis represents the shape template recognition sequence (dimensionless), and the vertical axis represents the deformation response (mm), ranging from -10mm to +40mm. The figure contains four characteristic curves, corresponding to four typical shape templates in the shape template library: uniformly stretched shape template (thickest solid line), shear expansion shape template (relatively thick solid line), end-constrained shape template (medium-thick solid line), and locally buckled shape template (thin solid line). The uniformly stretched shape template response curve shows the typical tensile deformation characteristics of the reinforcement under uniform load. The deformation increases monotonically from zero, reaching a maximum tensile deformation of 40mm at the 12th-13th point of the recognition sequence, and then slightly decreases. The shear expansion shape template response curve reflects the expansion deformation law of the reinforcement under shear load. Initially, it shows downward compression deformation, with a maximum compression of about -8mm. It begins to reverse at the 8th point of the recognition sequence, eventually stabilizing at a tensile state of about 20mm. The end-constraint plate response curve reflects the deformation characteristics of the reinforcement under constrained conditions, with relatively small variation amplitude and a maximum deformation of approximately 25 mm. The local buckling plate response curve shows the buckling deformation mode of the reinforcement under locally unstable conditions, initially rising rapidly to approximately 30 mm before stabilizing. The figure shows an upper limit of +35 mm and a lower limit of -10 mm (thick dashed line), forming an important component of the boundary consistency constraint set. The entire identification process is divided into four key stages: the initial identification stage (sequences 1-4) mainly involves basic data mapping and preliminary segment identification; the morphological plate replacement stage (sequences 4-8) performs morphological plate replacement and connection correction within the sliding window; the path solidification stage (sequences 8-12) determines the main load transfer path; and the stable transfer stage (sequences 12 and later) completes the final path map solidification. Through continuous evaluation by the consistency scorer, the system can accurately identify the morphological plate type corresponding to the current deformation state, providing reliable technical support for subsequent slip zone prediction.

[0055] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result according to substantially the same method falls within the scope of the present invention. Therefore, the scope of the present invention is defined only by the appended claims.

Claims

1. A geophysical-based engineering geological surveying and monitoring system, characterized in that, The system comprises a field acquisition terminal, a station terminal and a cloud terminal; the field acquisition terminal comprises two flexible conductive polymer geogrid segments and four icosahedral geophysical stress sensors arranged along the filling-digging interface to synchronously acquire conductive response, optical fiber scattering response and multi-directional stress response; the station terminal is used to time-align, remove abnormal data and edit the above responses to form basic acquisition data with unique identification index; the cloud terminal is used to input the basic acquisition data to construct a unified scene grid and set support anchor points at the sensors; according to the morphological template library, the initial state reinforcement deformation field is identified and boundary consistency constraints are applied; a candidate morphological graph is generated and evaluated by a consistency evaluator; after selecting the current morphological graph, morphological piece replacement and connection correction are implemented in a sliding window to increase the consistency score under the condition of maintaining the boundary consistency constraints; in the iteration process, a shear-dilation component graph and a transfer component graph are generated, and the area forming a closed feedback with the support anchor points in the same continuous band is marked as a potential slip zone candidate; when the convergence decision maker determines that the score change is below the set change threshold, the potential slip zone candidate is solidified into a potential slip zone graph, the main stem of the transfer component graph is solidified into a load transfer path graph, and the shear-dilation evolution description organized in time sequence is output; the station terminal is also used to receive the potential slip zone graph, the load transfer path graph and the shear-dilation evolution description to generate a comprehensive warning level.

2. The geophysical-based engineering geological reconnaissance and monitoring system of claim 1, wherein, The cloud terminal generates a unified scene grid with the start and end positions of the two flexible conductive polymer geogrid segments as boundaries; the unified scene grid contains a main index column along the direction of the flexible conductive polymer geogrid segment and an auxiliary index column across the filling-digging interface, and support anchor points are established at the positions of the four icosahedral geophysical stress sensors.

3. The geophysical-based engineering geological surveying and monitoring system of claim 2, wherein, According to indoor loading test data and historical engineering samples, a morphological template library is generated on the cloud terminal; the morphological template library contains at least the following types of morphological pieces: uniform stretching piece, shear expansion piece, end constraint piece, cross-weak zone piece, local buckling piece and rebound recovery piece, and each morphological piece gives a corresponding relationship description between the conductive path response, the optical fiber scattering response and the multi-directional stress response.

4. The geophysical-based engineering geological surveying and monitoring system of claim 3, wherein, The cloud terminal maps the basic acquisition data to the initial state reinforcement deformation field and the initial state soil support field with the unified scene grid as the bottom plate; during the mapping process, the same position is aligned with the unique identification and the acquisition window as the anchor point, and the conductive response, the optical fiber scattering response and the multi-directional stress response are spliced into continuous segments; boundary consistency constraints are established at both ends of the unified scene grid, which require the initial state reinforcement deformation field to maintain directional consistency and phase consistency with the responses of the four icosahedral geophysical stress sensors at both ends, and to maintain connectivity consistency at the filling-digging interface.

5. The geophysical-based engineering geological surveying and monitoring system of claim 4, wherein, The cloud terminal identifies the initial state reinforcement deformation field by the morphological template library to generate a candidate morphological graph; the candidate morphological graph is composed of combinations of morphological pieces on the unified scene grid, and a consistency evaluator is established for each combination; the consistency evaluator produces a single score based on the boundary consistency constraints, the conductive path change consistency and the multi-directional stress response consistency.

6. The geophysical-based engineering geological surveying and monitoring system of claim 5, wherein, The cloud selects a combination with the highest consistency score in the candidate morphology graph as the current morphology graph, slides a morphology piece length window from top to bottom along the primary index column in the current morphology graph, and performs morphology piece replacement and joint correction in the sliding window; the replacement should maintain the boundary consistency constraint set, replace the local morphology piece in the current morphology graph with a morphology piece that can improve the consistency score, and perform joint correction on the replacement boundary; the joint correction makes the adjacent segments tend to be coordinated in the conductive path change and multi-directional stress response by inserting a rebound recovery piece or an end constraint piece at the replacement boundary; after completing a sliding, the current morphology graph is updated and the consistency score is recalculated, if the consistency score is improved, the replacement is retained, and then the sliding window is pushed forward to continue the cycle.

7. The geophysical-based engineering geological surveying and monitoring system of claim 6, wherein, The cloud generates a dilatancy component graph and a transfer component graph based on the current morphology graph, wherein the dilatancy component graph represents the contribution of the deformation of the reinforcement to the volume change trend of the soil, and the transfer component graph represents the continuous path of the load propagation along the unified scene grid; The generation of the dilatancy component graph and the transfer component graph should be based on the physical process of the deformation of the reinforcement and the propagation of the load, and the unified scene grid should be analyzed as a whole; a continuous band that meets the conditions of long-term consistent direction of the dilatancy component graph and the transfer component graph in the same continuous band, and forms a closed feedback condition with the support anchor response is marked as a potential sliding band candidate.

8. The geophysical-based engineering geological surveying and monitoring system of claim 7, wherein, When the sliding window completely covers the unified scene grid and the consistency score changes of two consecutive rounds are below the set change threshold, the convergence arbitrator is triggered; the potential sliding band candidate with the highest stability is solidified as a potential sliding band graph, the main stem with the highest connectivity in the transfer component graph is solidified as a load transfer path graph, and the ordered set of the dilatancy component graph over time is solidified as a dilatancy evolution description; the three are used as transparent output for subsequent hierarchical warning calls.

9. The geophysical-based engineering geological surveying and monitoring system of claim 8, wherein, The station end receives the potential sliding band graph, the load transfer path graph and the dilatancy evolution description and generates a comprehensive warning level; when the comprehensive warning level reaches a preset level, a treatment suggestion is pushed to the vehicle end and the scene simultaneously; the treatment suggestion at least includes monitoring encryption positions, traffic organization adjustment prompts and temporary reinforcement operation sequences; The cloud archives the comprehensive warning level and the transparent output for full life cycle backtracking and method review.

Images