A deformation monitoring method and system for sheet pile foundation pit support

Abstract

This invention relates to the field of foundation pit support monitoring technology, specifically a deformation monitoring method and system for steel sheet pile foundation pit support. The method includes: acquiring steel sheet pile parameters, foundation pit soil profile information, and preset excavation stages; establishing a set of monitoring points for predicting lateral deformation of the foundation pit; deriving a monitoring data sampling strategy with staged sampling intervals and acquisition durations based on the excavation stages; monitoring and obtaining initial deformation data sequences according to the strategy; extracting deformation trend curves to identify key deformation stages; dynamically calling an adaptive data fusion model to fuse and reconstruct the data; and generating state data reflecting the overall spatial deformation of the support structure. This solution can be adapted to the characteristics of the foundation pit and the excavation stage to formulate monitoring plans, achieving dynamic fusion of deformation data and characterization of the overall spatial deformation state.

Description

Technical Field

[0001] This invention relates to the field of foundation pit engineering support monitoring technology, and in particular to a deformation monitoring method and system for steel sheet pile foundation pit support. Background Technology

[0002] Currently, deformation monitoring of sheet pile foundation pit support mostly adopts conventional monitoring modes. The layout of monitoring points is largely determined based on engineering experience, and the data sampling interval and acquisition duration use uniform and fixed parameters. The sheet pile body parameters, foundation pit soil profile information, and foundation pit excavation stage are not integrated and applied. On-site monitoring follows fixed layout schemes and sampling rules. Deformation monitoring data is mostly stored and analyzed in the form of single-point raw data, simply recording and statistically analyzing the deformation values ​​of individual monitoring points. Deformation stage division is not performed using deformation trend curves, nor are dynamic data processing operations implemented for deformation changes during the monitoring process.

[0003] Under conventional monitoring methods, the distribution of monitoring points cannot match the geological conditions and structural characteristics of the foundation pit itself. This can easily lead to situations where monitoring of critical deformation areas is missing or where monitoring points are redundant in non-critical areas. Fixed sampling strategies cannot adapt to the deformation change rates at different excavation stages, resulting in excessive data collection or omissions of key deformation data. Monitoring data can only present single-point deformation values ​​and cannot reflect the overall spatial deformation state of the support structure. The data processing method is disconnected from the dynamic changes in the monitoring process and cannot adjust the data processing logic according to the actual deformation stage.

[0004] It is necessary to develop a suitable monitoring point layout plan and sampling strategy based on the characteristics of the sheet piles and the foundation pit soil layers, as well as the excavation stage. At the same time, it is necessary to identify the key deformation stages based on the deformation trend of the monitoring data, and obtain the overall spatial deformation state of the support structure through dynamic data fusion and reconstruction. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and to propose a deformation monitoring method and system for steel sheet pile foundation pit support.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a deformation monitoring method for sheet pile foundation pit support, comprising: The pile parameters of the sheet pile, the soil profile information of the foundation pit, and the preset multiple excavation stages are obtained. The lateral deformation prediction structure of the foundation pit is established by integrating the pile parameters and soil profile information. Based on the lateral deformation prediction structure of the foundation pit, a set of monitoring points is generated according to the monitoring point deployment strategy. By combining the preset multiple excavation stages with the foundation pit lateral deformation prediction structure, a monitoring data sampling strategy is derived. The monitoring data sampling strategy includes a set of sampling intervals and a set of data collection durations for different excavation stages. Using the set of monitoring points and the monitoring data sampling strategy, on-site monitoring is initiated to obtain an initial deformation monitoring data sequence; Deformation trend curves of monitoring points are extracted from the initial deformation monitoring data sequence, and key deformation stages are identified through the deformation trend curves. During the monitoring process, an adaptive data fusion model is dynamically invoked based on the key deformation stages. The adaptive data fusion model is used to fuse and reconstruct the deformation monitoring data sequence to generate a fused spatial deformation state that reflects the overall spatial deformation state of the support structure.

[0007] As a further aspect of the present invention, a structure for predicting the lateral deformation of the foundation pit is established by integrating the pile parameters and soil profile information, including: The soil profile information includes the soil type and depth boundaries of each soil layer; The structure for predicting the lateral deformation of the foundation pit includes the estimated deformation values ​​for multiple soil layer boundary points. Based on the soil type of each soil layer in the soil profile information, the corresponding soil elastic modulus and cohesion parameters are obtained; The pile parameters are imported into the lateral deformation analysis framework, which uses the pile's moment of inertia and section modulus as the calculation basis. In the lateral deformation analysis framework, the soil elastic modulus and cohesion parameters are superimposed layer by layer according to depth to generate a soil-pile interaction model. A standard lateral load distribution defined by the excavation stage is applied to the soil-pile interaction model, and the estimated lateral displacement of multiple soil layer boundary points is calculated. The estimated lateral displacement values ​​of the multiple soil layer boundary points are integrated to construct the estimated lateral deformation structure of the foundation pit. The estimated lateral deformation structure of the foundation pit links the soil layer depth and the estimated displacement value in tabular form.

[0008] As a further aspect of the present invention, a monitoring point distribution set is generated based on the monitoring point deployment strategy of the foundation pit lateral deformation prediction structure, including: The monitoring point distribution set includes multiple monitoring points and the sensor type for each monitoring point. The sensor types include inclinometers, strain sensors, and displacement sensors. The structure for predicting the lateral deformation of the foundation pit is analyzed to locate the abrupt depth of the change in the predicted lateral displacement. Based on the pile length in the pile parameters, preliminary monitoring points are generated at the top, middle, and bottom of the pile, as well as at the location of the abrupt change depth. Based on the preset multiple excavation stages, determine the exposure time and stress change characteristics of each preliminary monitoring point during the excavation process; Based on the stress change characteristics, at least one sensor type is assigned to each preliminary monitoring point, wherein displacement sensors are associated with the top and bottom of the pile, inclinometer sensors are associated with the abrupt change depth location, and strain sensors are associated with the middle of the pile. The initial monitoring points and their assigned sensor types are integrated to form the monitoring point distribution set.

[0009] As a further aspect of the present invention, by combining the preset multiple excavation stages with the foundation pit lateral deformation prediction structure, a monitoring data sampling strategy is derived, including: Extract the stage depth and stage duration of the preset multiple excavation stages; In the lateral deformation prediction structure of the foundation pit, the lateral displacement prediction value corresponding to the depth of each stage is found. Based on the rate of change of the lateral displacement prediction value, the initial monitoring sampling interval for each excavation stage is set, where the stage with a large rate of change corresponds to a short sampling interval. A safety factor matrix is ​​introduced to calibrate each initial monitoring sampling interval, resulting in a calibrated sampling interval; Based on the stage duration and the post-calibration sampling interval, the data acquisition duration of each stage is calculated, and the post-calibration sampling interval and data acquisition duration of all stages are summarized into the monitoring data sampling strategy.

[0010] As a further aspect of the present invention, deformation trend curves of monitoring points are extracted from the initial deformation monitoring data sequence, and key deformation stages are identified through the deformation trend curves, including: With time as the horizontal axis and the displacement at the top of the pile, the strain at the middle of the pile, and the displacement at the bottom of the pile extracted from the initial deformation monitoring data sequence as the vertical axis, separate curves of displacement versus time and strain versus time are plotted. The trend decomposition algorithm is applied to process the displacement-time curve and the strain-time curve to separate the long-term trend component, periodic component and random noise component. Extract the long-term trend component and calculate the curvature value of the long-term trend component between adjacent sampling points; The time interval in which the sampling points with curvature values ​​exceeding the curvature threshold are marked is defined as a potential critical deformation segment; By combining the preset multiple excavation stages, the specific excavation stage to which the potential critical deformation section belongs is determined, and the specific excavation stage is marked as the critical deformation stage.

[0011] As a further aspect of the present invention, during the monitoring process, an adaptive data fusion model is dynamically invoked based on the key deformation stages, including: An adaptive data fusion model containing multiple data fusion operators is pre-constructed, including a moving average operator, a weighted fusion operator, and a spatial interpolation operator; Establish a stage-operator mapping table, which maps different excavation stages to different combinations of data fusion operators; When the monitoring process reaches the critical deformation stage, the target data fusion operator combination corresponding to the critical deformation stage is dynamically loaded according to the stage-operator mapping table. Read the active sensor types in the current stage from the set of monitoring point distributions, and configure the internal parameters of the target data fusion operator combination according to the sensor types.

[0012] As a further aspect of the present invention, an adaptive data fusion model is used to fuse and reconstruct deformation monitoring data sequences, including: Using the calibrated sampling interval of the current stage in the monitoring data sampling strategy as a time window, the deformation monitoring data sequence is extracted; The extracted deformation monitoring data sequence is input into the configured target data fusion operator combination; The extracted deformation monitoring data sequence is smoothed in the time dimension by using the moving average operator to obtain a smoothed data sequence. By using a spatial interpolation operator, based on the spatial positional relationship of the monitoring points in the monitoring point distribution set, the smoothed data sequence is interpolated in the spatial dimension to generate a spatially continuous deformation field. By using a weighted fusion operator, different weights are assigned to the sensor types at the monitoring points, and data from different types of sensors in a continuous spatial deformation field are weighted and synthesized to output preliminary fusion results.

[0013] As a further aspect of the present invention, generating a fused spatial deformation state reflecting the overall spatial deformation state of the support structure includes: The preliminary fusion results are spatially meshed, and the spatial region corresponding to the foundation pit support structure is discretized into regular spatial mesh nodes. Based on the Kriging spatial estimation algorithm, the deformation state estimate of each spatial grid node is calculated using data from known locations in the preliminary fusion results. The deformation state estimates of all spatial grid nodes are integrated to form the overall spatial deformation state field of the support structure; From the overall spatial deformation state field of the support structure, the deformation distribution on the preset feature profile is extracted to form the fused spatial deformation state. The feature profile includes the central longitudinal profile along the long side of the foundation pit and the transverse profile at the maximum excavation depth.

[0014] As a further aspect of the present invention, the method further includes a verification step of the spatial deformation state after fusion: From the initial deformation monitoring data sequence, separate the independent validation data subset that did not participate in the adaptive data fusion model fusion process; In the fused spatial deformation state, locate the spatial position corresponding to the monitoring point in the independent verification data subset, and read the deformation state estimate of the spatial position; Calculate the residuals between the measured values ​​and the corresponding deformation state estimates in the independent validation data subset; If the statistical characteristic value of the residual exceeds the preset residual threshold, a fusion parameter adjustment instruction is triggered, and the internal parameters of the target data fusion operator combination are optimized and updated in reverse based on the residual.

[0015] As a further aspect of the present invention, the present invention also includes a deformation monitoring system for sheet pile foundation pit support, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, it implements the steps of the deformation monitoring method for sheet pile foundation pit support as described above.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: By integrating the pile parameters of sheet piles, the soil profile information of the foundation pit, and multiple pre-set excavation stages, a lateral deformation prediction structure for the foundation pit is constructed. This structure generates a corresponding set of monitoring points. By combining the pre-set excavation stages with the lateral deformation prediction structure, a monitoring data sampling strategy is derived, which includes a set of sampling intervals and a set of data acquisition durations for different excavation stages. This ensures that the location of the monitoring points is consistent with the geological conditions and structural characteristics of the foundation pit, that the sampling intervals and data acquisition durations match the deformation change patterns corresponding to each excavation stage, that the monitoring point layout matches the actual foundation pit working conditions, and that the sampling parameters are adapted to the deformation development rhythm during the foundation pit excavation process. This avoids problems such as unreasonable monitoring point layout and mismatch between sampling parameters and deformation characteristics. The process of collecting the original monitoring data can closely match the actual changes in the deformation of the foundation pit support, ensuring the relevance and adaptability of the monitoring layout and data acquisition.

[0017] Deformation trend curves corresponding to monitoring points are extracted from the initial deformation monitoring data sequence. Key deformation stages are identified through these curves. During monitoring, an adaptive data fusion model is dynamically invoked according to the identified key deformation stages. The model performs fusion and reconstruction operations on the deformation monitoring data sequence to form a fused spatial deformation state that reflects the overall spatial deformation state of the support structure. This achieves synchronous adaptation between the data processing logic and the changes in deformation stages during monitoring, transforming scattered single-point monitoring data into holistic spatial deformation characterization information. This fully presents the spatial deformation distribution characteristics of the support structure, breaking through the limitations of independent analysis of single-point data. The deformation monitoring results can comprehensively reflect the overall spatial deformation state of the support structure, strengthening the overall characterization ability of the monitoring data on the deformation state of the support structure, and making the analysis results of deformation monitoring more consistent with the actual spatial deformation of the foundation pit support structure. Attached Figure Description

[0018] Figure 1 This is a flowchart of a deformation monitoring method for sheet pile foundation pit support according to the present invention; Figure 2 A flowchart for establishing a structure to predict the lateral deformation of the foundation pit; Figure 3 A flowchart for generating a set of monitoring point distributions. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0021] See Figure 1 This invention provides a deformation monitoring method for sheet pile foundation pit support, the specific method including: The process involves acquiring sheet pile parameters, soil profile information of the foundation pit, and multiple pre-defined excavation stages. A lateral deformation prediction structure for the foundation pit is established by integrating the pile parameters and soil profile information. Based on this structure, a monitoring point distribution set is generated using a deployment strategy. By combining the pre-defined excavation stages with the lateral deformation prediction structure, a monitoring data sampling strategy is derived, including a set of sampling intervals and data acquisition durations for different excavation stages. On-site monitoring is initiated using the monitoring point distribution set and the monitoring data sampling strategy to obtain an initial deformation monitoring data sequence. Deformation trend curves for the monitoring points are extracted from this initial sequence, and key deformation stages are identified using these curves. During monitoring, an adaptive data fusion model is dynamically invoked based on the key deformation stages. This model is used to fuse and reconstruct the deformation monitoring data sequence, generating a fused spatial deformation state reflecting the overall spatial deformation state of the support structure.

[0022] In one embodiment of the present invention, see [reference] Figure 2 The soil profile information includes the soil type and depth boundaries of each soil layer. The lateral deformation prediction structure for the foundation pit contains deformation prediction values ​​for multiple soil layer boundary points. Based on the soil type of each soil layer in the soil profile information, the corresponding soil elastic modulus and cohesion parameters are obtained. The pile parameters are imported into the lateral deformation analysis framework, which uses the pile's moment of inertia and section modulus as the calculation benchmark. In the lateral deformation analysis framework, the soil elastic modulus and cohesion parameters are layered and superimposed according to depth to generate a soil-pile interaction model. A standard lateral load distribution defined by the excavation stage is applied to the soil-pile interaction model, and the lateral displacement prediction values ​​for multiple soil layer boundary points are calculated. The lateral displacement prediction values ​​for multiple soil layer boundary points are integrated to construct the foundation pit lateral deformation prediction structure, which links the soil layer depth and displacement prediction values ​​in tabular form.

[0023] In practical implementation, establishing a lateral deformation prediction structure for the foundation pit requires specific soil profile information. Considering an example scenario, the soil profile information includes three soil layers: the first layer is miscellaneous fill, with a depth boundary of 0 to 2 meters; the second layer is silty clay, with a depth boundary of 2 to 8 meters; and the third layer is medium sand, with a depth boundary of 8 to 15 meters. The lateral deformation prediction structure for the foundation pit needs to include the deformation prediction values ​​at these three depth boundaries. Based on the soil type of each soil layer in the soil profile information, the corresponding soil elastic modulus and cohesion parameters are obtained. In practice, these parameters are obtained by consulting the engineering geological survey report. The elastic modulus of miscellaneous fill is 8 MPa, and the cohesion is 5 kPa; the elastic modulus of silty clay is 15 MPa, and the cohesion is 20 kPa; and the elastic modulus of medium sand is 30 MPa, and the cohesion is 0 kPa. These parameters are the input data for building the model. The pile parameters are imported into the lateral deformation analysis framework, which uses the pile's moment of inertia and section modulus as the calculation basis. In some embodiments, the pile parameters include the steel sheet pile model FSP-Ⅳ, with a moment of inertia of 46,800 cm⁴ per meter and a section modulus of 2,030 cm³ per meter. The lateral deformation analysis framework is constructed based on the elastic foundation beam theory.

[0024] In the lateral deformation analysis framework, the soil elastic modulus and cohesion parameters are layered and superimposed according to depth to generate a soil-pile interaction model. In specific implementation, the depth direction of the foundation pit is discretized into multiple calculation nodes. The soil resistance at each node is calculated using the elastic modulus and cohesion parameters of the soil layer at that depth, expressed by the following formula: , in: This represents the distributed resistance of the soil at depth z to the pile. This represents the subgrade coefficient of the soil at depth z, which is directly proportional to the elastic modulus of the soil. This represents the lateral displacement of the pile at depth z. The formula represents the equivalent cohesion contribution of the soil at depth z, reflecting the process of layering soil parameters into the model according to depth. A standard lateral load distribution defined by the excavation stage is applied to the soil-pile interaction model, and the estimated lateral displacements at multiple soil layer boundaries are calculated. In practice, the pre-defined excavation stages include three phases: the first phase excavates to a depth of 4 meters, with a triangular standard lateral load distribution and a maximum value of 30 kN / m²; the second phase excavates to a depth of 8 meters, with a trapezoidal load distribution; and the third phase excavates to a depth of 12 meters, with a rectangular load distribution. The load distribution for each phase is input into the soil-pile interaction model for solution, yielding the estimated lateral displacement at the soil layer boundaries for each phase. For example, after the first phase excavation, the estimated lateral displacement at the 2-meter depth boundary is 4.2 mm, and the estimated lateral displacement at the 8-meter depth boundary is 1.8 mm.

[0025] By integrating the estimated lateral displacements from multiple soil layer boundary points, a lateral deformation prediction structure for the foundation pit is constructed. This structure links soil layer depths with estimated displacements in a tabular format. The optional table format includes three columns: excavation stage number, soil layer boundary point depth, and estimated lateral displacement. Taking the first stage as an example, the table records a displacement of 0 mm at depth 0 meters, 4.2 mm at depth 2 meters, 1.8 mm at depth 8 meters, and 0 mm at depth 15 meters. This table provides quantified deformation prediction data for subsequent monitoring point deployment. In this way, the foundation pit lateral deformation prediction structure is established based on specific soil layer and pile parameters. Displacement estimates are calculated using standard loads and presented in tabular form.

[0026] In one embodiment of the present invention, see [reference] Figure 3 The monitoring point distribution set includes multiple monitoring points and the sensor type for each point, including inclinometers, strain sensors, and displacement sensors. The lateral deformation prediction structure of the foundation pit is analyzed to locate the abrupt change depths where the predicted lateral displacement changes. Based on the pile length parameters, preliminary monitoring points are generated at the top, middle, and bottom of the pile, as well as at the abrupt change depths. According to multiple pre-set excavation stages, the exposure time and stress change characteristics of each preliminary monitoring point during the excavation process are determined. Based on the stress change characteristics, at least one sensor type is assigned to each preliminary monitoring point: displacement sensors are associated with the top and bottom of the pile, inclinometers with the abrupt change depths, and strain sensors with the middle of the pile. The preliminary monitoring points and their assigned sensor types are integrated to form the monitoring point distribution set.

[0027] In practical implementation, the generation of the monitoring point distribution set depends on the lateral deformation prediction structure of the foundation pit. In an example scenario, the lateral deformation prediction structure records the estimated lateral displacement at different depths in tabular form. For example, the displacement is 0 mm at a depth of 0 meters, 4.2 mm at a depth of 2 meters, 7.1 mm at a depth of 4 meters, 1.8 mm at a depth of 8 meters, 0.5 mm at a depth of 12 meters, and 0 mm at a depth of 15 meters. The monitoring point distribution set includes multiple monitoring points and the sensor type for each monitoring point. The sensor types include inclinometers, strain sensors, and displacement sensors. It can be understood that these different types of sensors are used to capture deformation information of different properties. The lateral deformation prediction structure of the foundation pit is analyzed to locate the abrupt change depth where the lateral displacement prediction value changes. In some embodiments, the abrupt change is identified by calculating the absolute value of the difference between the lateral displacement prediction values ​​between adjacent depth points. For example, the absolute value of the displacement change from depth 2 meters to depth 4 meters is 2.9 mm, the absolute value of the displacement change from depth 4 meters to depth 8 meters is 5.3 mm, and the absolute value of the displacement change from depth 8 meters to depth 12 meters is 1.3 mm. A change threshold is set, and when the absolute value of the difference exceeds 3.0 mm, it is determined to be an abrupt change. In this example, the change of 5.3 mm between depth 4 meters and depth 8 meters exceeds the threshold. Therefore, the two boundary points at depths of 4 meters and 8 meters are identified as the abrupt change depth locations. This process can be formally expressed by a formula as follows: , in: This represents the change in the i-th depth segment. This represents the estimated lateral displacement at the (i+1)th depth point. This represents the estimated lateral displacement at the i-th depth point, and the values ​​of all adjacent depth segments are calculated. and will Points with depths greater than a preset threshold are marked as mutation depth locations.

[0028] Based on the pile length in the pile parameters, preliminary monitoring points are generated at the top, middle, and bottom of the pile, as well as at the abrupt change depth locations. In the specific implementation, the pile length is 15 meters, the top of the pile corresponds to a depth of 0 meters, the bottom of the pile corresponds to a depth of 15 meters, and the middle of the pile corresponds to a depth of 7.5 meters. The identified abrupt change depth locations are 4 meters and 8 meters. Therefore, the preliminary monitoring points are set at depths of 0 meters, 4 meters, 7.5 meters, 8 meters, and 15 meters. It can be understood that these points cover the key locations of the pile and the depths at which the predicted deformation characteristics change significantly. Based on the pre-set multiple excavation stages, the exposure time and stress change characteristics of each preliminary monitoring point during the excavation process are determined. The pre-set excavation stages include three stages: the first stage excavates to a depth of 4 meters, the second stage excavates to a depth of 8 meters, and the third stage excavates to a depth of 12 meters. For example, the monitoring point at a depth of 4 meters is fully exposed at the end of the first stage excavation, and its stress state changes from being constrained by earth pressure to a complex stress state near the free surface, with significant stress change characteristics. The monitoring point at a depth of 8 meters is only fully exposed at the end of the second stage excavation, and its stress change mainly occurs in the later part of the second stage. Based on the stress change characteristics, at least one sensor type is assigned to each initial monitoring point. Displacement sensors are associated with the top and bottom of the pile, inclinometers are associated with the abrupt depth locations, and strain sensors are associated with the middle of the pile. In some embodiments, a displacement sensor is installed at the top of the pile at a depth of 0 meters and at the bottom of the pile at a depth of 15 meters to monitor the absolute displacement of the pile tip. An inclinometer is installed at the abrupt depth locations at depths of 4 meters and 8 meters to monitor the inclination changes of the soil layer near that depth. A strain sensor is installed in the middle of the pile at a depth of 7.5 meters to monitor the pile stress. An optional configuration is to install displacement sensors at the points at depths of 4 meters and 8 meters simultaneously as redundant monitoring. The initial monitoring points and their assigned sensor types are integrated to form a monitoring point distribution set. In practice, this set is a list, and each record in the list includes the monitoring point depth and the type of sensor designed to be installed. For example, Record 1: Depth 0 meters, sensor type is displacement sensor; Record 2: Depth 4 meters, sensor types are inclinometer and displacement sensor; Record 3: Depth 7.5 meters, sensor type is strain sensor; Record 4: Depth 8 meters, sensor type is inclinometer sensor; Record 5: Depth 15 meters, sensor type is displacement sensor. This list will be used to guide the installation of sensors on site.

[0029] In one embodiment of the present invention, the stage depth and stage duration of multiple preset excavation stages are extracted. In the lateral deformation prediction structure of the foundation pit, the lateral displacement prediction value corresponding to each stage depth is found. Based on the rate of change of the lateral displacement prediction value, an initial monitoring sampling interval for each excavation stage is set, where stages with a large rate of change correspond to shorter sampling intervals. A safety factor matrix is ​​introduced to calibrate each initial monitoring sampling interval, resulting in a calibrated sampling interval. Based on the stage duration and the calibrated sampling interval, the data acquisition duration for each stage is calculated. The calibrated sampling intervals and data acquisition durations for all stages are summarized into a monitoring data sampling strategy.

[0030] Using time as the horizontal axis and the displacement at the top of the pile, strain at the middle of the pile, and displacement at the bottom of the pile extracted from the initial deformation monitoring data sequence as the vertical axis, independent displacement versus time curves and strain versus time curves are plotted respectively. A trend decomposition algorithm is applied to process the displacement versus time curves and strain versus time curves to separate the long-term trend component, periodic component, and random noise component. The long-term trend component is extracted, and the curvature value of the long-term trend component between adjacent sampling points is calculated. The time intervals of sampling points whose curvature values ​​exceed the curvature threshold are marked, and these time intervals are defined as potential critical deformation sections. Combining multiple preset excavation stages, the specific excavation stage to which the potential critical deformation section belongs is determined, and the specific excavation stage is marked as the critical deformation stage.

[0031] In practice, the derivation of the monitoring data sampling strategy is based on multiple pre-set excavation stages and the predicted lateral deformation structure of the foundation pit. Considering an example scenario, the three pre-set excavation stages are the first stage, the second stage, and the third stage. The stage depth and stage duration of each excavation stage are extracted. The first stage is excavated to a depth of 4 meters and the stage duration is 5 days. The second stage is excavated to a depth of 8 meters and the stage duration is 7 days. The third stage is excavated to a depth of 12 meters and the stage duration is 6 days. In the lateral deformation prediction structure of the foundation pit, the lateral displacement prediction value corresponding to the depth of each stage is found. The lateral displacement prediction value corresponding to the depth of the first stage is 7.1 mm for a depth of 4 meters, the lateral displacement prediction value corresponding to the depth of the second stage is 1.8 mm for a depth of 8 meters, and the lateral displacement prediction value corresponding to the depth of the third stage is 0.5 mm for a depth of 12 meters. Based on the rate of change of the lateral displacement prediction value, the initial monitoring sampling interval for each excavation stage is set. The rate of change is calculated as the ratio of the difference in the lateral displacement prediction value between adjacent stages to the difference in stage depth. For example, the rate of change from the first stage to the second stage is (1.8-7.1) / (8-4) = -1.325 mm per meter. Since the absolute value is relatively large, a shorter initial monitoring sampling interval is set for the first stage, such as 2 hours. The rate of change for the second stage is smaller, so a longer interval is set, such as 4 hours. A safety factor matrix is ​​introduced to calibrate each initial monitoring sampling interval, resulting in a calibrated sampling interval. The safety factor matrix is ​​a table of coefficients related to the risk level of each excavation stage. For example, in the first stage, with a high risk level and a safety factor of 0.8, the calibrated sampling interval is the initial sampling interval multiplied by the safety factor, i.e., 2 hours multiplied by 0.8 equals 1.6 hours. In the second stage, with a medium risk level and a safety factor of 1.0, the calibrated sampling interval is 4 hours multiplied by 1.0 equals 4 hours. Based on the stage duration and the calibrated sampling interval, the data collection duration for each stage is calculated. The data collection duration equals the stage duration multiplied by 24 hours and then divided by the calibrated sampling interval to obtain the number of data points required for that stage. The calibrated sampling intervals and data collection durations for all stages are summarized into a monitoring data sampling strategy. See Table 1 for a specific monitoring data sampling strategy table. Table 1: Monitoring Data Sampling Strategy Table Excavation stage Stage depth (meters) Phase duration (days) Initial monitoring sampling interval (hours) Safety factor Sampling interval after calibration (hours) Data collection duration Phase 1 4 5 2 0.8 1.6 75 Phase Two 8 7 4 1.0 4.0 42 Phase Three 12 6 6 1.2 7.2 20

[0032] In practical implementation, deformation trend curves of monitoring points are extracted from the initial deformation monitoring data sequence, and key deformation stages are identified through these curves. Independent displacement versus time curves and strain versus time curves are plotted with time as the horizontal axis and the pile top displacement, pile middle strain, and pile bottom displacement extracted from the initial deformation monitoring data sequence as the vertical axis. For example, the pile top displacement data sequence is [0, 1.2, 3.5, 7.1, 8.0] mm, corresponding to time points [0, 1, 2, 3, 4] days. A trend decomposition algorithm is applied to process the displacement versus time curves and strain versus time curves, separating the long-term trend component, periodic component, and random noise component. In some embodiments, the trend decomposition algorithm uses a moving average method, for example, taking a window size of 3 data points and performing a moving average processing on the pile top displacement data sequence to obtain the long-term trend component sequence [0, 1.6, 4.0, 6.2, 7.6] mm. Extract the long-term trend component and calculate the curvature value of the long-term trend component between adjacent sampling points. The curvature value reflects the degree of bending of the trend curve, and the calculation formula is as follows: , in: This represents the curvature value at the k-th sampling point. This represents the long-term trend component value of the (k+1)th sampling point. This represents the long-term trend component value of the k-th sampling point. The long-term trend component value of the (k-1)th sampling point is represented. The curvature values ​​of all sampling points are calculated, for example, the curvature value sequence is [0, 0.05, 0.12, 0.08, 0.03]. The time interval of sampling points whose curvature values ​​exceed the curvature threshold is marked, and the time interval is defined as a potential critical deformation segment. In some embodiments, the curvature threshold is set to 0.1. The curvature value of the third sampling point in the curvature value sequence is 0.12, which exceeds the threshold. Its corresponding time interval is from day 2 to day 3. Therefore, day 2 to day 3 is marked as a potential critical deformation segment. Combining multiple preset excavation stages, the specific excavation stage to which the potential critical deformation segment belongs is determined, and the specific excavation stage is marked as a critical deformation stage. It can be understood that the first stage in the preset excavation stages lasts for 5 days. The potential critical deformation segment from day 2 to day 3 is within the first stage. Therefore, the first stage is marked as a critical deformation stage. Optionally, when there are multiple potential critical deformation segments, they may correspond to multiple critical deformation stages. In this way, the monitoring data sampling strategy is obtained through specific calculations, and the key deformation stages are identified through curvature analysis, providing a basis for the subsequent use of the adaptive data fusion model.

[0033] In one embodiment of the present invention, an adaptive data fusion model containing multiple data fusion operators is pre-constructed. These operators include a moving average operator, a weighted fusion operator, and a spatial interpolation operator. A stage-operator mapping table is established, mapping different excavation stages to different combinations of data fusion operators. When the monitoring process reaches a critical deformation stage, the target data fusion operator combination corresponding to the critical deformation stage is dynamically loaded according to the stage-operator mapping table. The active sensor types for the current stage are read from the monitoring point distribution set, and the internal parameters of the target data fusion operator combination are configured according to the sensor types.

[0034] Using the calibrated sampling interval of the current stage in the monitoring data sampling strategy as a time window, a deformation monitoring data sequence is extracted. The extracted deformation monitoring data sequence is then input into a pre-configured target data fusion operator combination. A moving average operator is used to smooth the extracted deformation monitoring data sequence in the time dimension, resulting in a smoothed data sequence. A spatial interpolation operator is then used to interpolate the smoothed data sequence in the spatial dimension based on the spatial positional relationships of the monitoring points in the monitoring point distribution set, generating a spatially continuous deformation field. Finally, a weighted fusion operator is used to assign different weights to the data from different types of sensors in the spatially continuous deformation field according to the sensor type of the monitoring points, performing weighted synthesis and outputting a preliminary fusion result.

[0035] In practical implementation, an adaptive data fusion model containing multiple data fusion operators is pre-constructed. These operators include the moving average operator, the weighted fusion operator, and the spatial interpolation operator. A stage-operator mapping table is established, mapping different excavation stages to different combinations of data fusion operators. The stage-operator mapping table defines the recommended combinations of data fusion operators for different excavation stages. For example, in the initial stage of excavation when deformation is gentle, only the moving average operator may be used, while in the critical stage of severe deformation, all three operators are used. See Table 2 for a specific stage-operator mapping: Table 2: Stage-Operator Mapping Table Excavation stage Stage Feature Description Recommended data fusion operator combinations Phase 1 Initial excavation, deformation initiation Moving average operator Phase Two Key deformations, rapid changes Moving average operator, spatial interpolation operator, weighted fusion operator Phase Three Deformation converges and tends to stabilize. Moving average operator, weighted fusion operator

[0036] When the monitoring process reaches the critical deformation stage, the target data fusion operator combination corresponding to the critical deformation stage is dynamically loaded according to the stage-operator mapping table. In specific implementation, the critical deformation stage is identified as the second stage. The target data fusion operator combination corresponding to the second stage is found from the stage-operator mapping table as a moving average operator, a spatial interpolation operator, and a weighted fusion operator. The active sensor types for the current stage are read from the monitoring point distribution set, and the internal parameters of the target data fusion operator combination are configured according to the sensor types. The monitoring point distribution set indicates that displacement sensors, inclinometers, and strain sensors are installed at the current monitoring point. For example, the data weight of the displacement sensor is configured to 0.4, the data weight of the inclinometer is configured to 0.4, the data weight of the strain sensor is configured to 0.2, the time window size of the moving average operator is configured to 3 times the sampling interval after calibration for the current stage, and the search radius of the spatial interpolation operator is configured to 5 meters.

[0037] Using the post-calibration sampling interval of the current stage in the monitoring data sampling strategy as a time window, the deformation monitoring data sequence is extracted. In specific implementation, the post-calibration sampling interval of the key deformation stage, namely the second stage, is 4 hours. Therefore, a continuous 4-hour period is used as a time window to extract the reading sequence of all sensors within this time period. For example, the displacement sensor collects the data sequence D1: [2.1, 2.3, 2.6, 3.0] mm within 4 hours, the inclinometer collects the data sequence I1: [0.5, 0.7, 0.9, 1.2] degrees, and the strain sensor collects the data sequence S1: [150, 155, 165, 180] microstrain. The extracted deformation monitoring data sequence is input into the configured target data fusion operator combination. A moving average operator is used to smooth the extracted deformation monitoring data sequence in the time dimension, resulting in a smoothed data sequence. The moving average operator uses a three-point moving average. For example, smoothing the displacement sensor data sequence D1 yields the smoothed sequence D1_smooth: [None, (2.1+2.3+2.6) / 3, (2.3+2.6+3.0) / 3, None], i.e., D1_smooth: [None, 2.33, 2.63, None]. Using a spatial interpolation operator, based on the spatial positional relationship of the monitoring points in the monitoring point distribution set, the smoothed data sequence is interpolated in the spatial dimension to generate a spatially continuous deformation field. In some embodiments, the monitoring points are distributed at depths of 0 meters, 4 meters, 7.5 meters, 8 meters, and 15 meters. The smoothed data provides deformation values ​​for these discrete depth points. The spatial interpolation operator uses an inverse distance weighting method to calculate the deformation estimate for unmonitored depth points, using the following formula: , in: This represents the estimated deformation value at the depth point z to be determined. This represents the smoothed deformation value of the i-th known monitoring point. Let represent the weight of the i-th known monitoring point. The weight is inversely proportional to the square of the distance from the point z to be determined to the known monitoring point i. This formula can generate a deformation field at a continuous depth from the top to the bottom of the pile. Using a weighted fusion operator, different weights are assigned based on the sensor type at each monitoring point. Data from different types of sensors in the continuous spatial deformation field are weighted and synthesized, outputting a preliminary fusion result. For example, at a depth of 4 meters, there is a data value of 2.33 mm from a displacement sensor and an equivalent displacement value of 2.1 mm derived from an inclinometer. The weighted fusion operator performs a weighted average according to the pre-configured weights of 0.4 for the displacement sensor and 0.4 for the inclinometer sensor, calculating the preliminary fusion result at a depth of 4 meters as (0.4 + 2.33 + 0.4 + 2.1) / (0.4 + 0.4) = 2.215 mm. Similar calculations are performed for all depth points, outputting a preliminary fusion result sequence of depth-deformation values.

[0038] In one embodiment of the present invention, the preliminary fusion result is spatially meshed, discretizing the spatial region corresponding to the foundation pit support structure into regular spatial grid nodes. Based on the Kriging spatial estimation algorithm, the deformation state estimate of each spatial grid node is calculated using data from known locations in the preliminary fusion result. The deformation state estimates of all spatial grid nodes are integrated to form the overall spatial deformation state field of the support structure. From the overall spatial deformation state field of the support structure, the deformation distribution on a preset feature profile is extracted to form the fused spatial deformation state. The feature profile includes the central longitudinal section along the long side of the foundation pit and the transverse section at the maximum excavation depth.

[0039] From the initial deformation monitoring data sequence, an independent validation data subset that did not participate in the adaptive data fusion model fusion process is separated. In the fused spatial deformation state, the spatial locations corresponding to the monitoring points in the independent validation data subset are located, and the deformation state estimates for these locations are read. The residuals between the measured values ​​and the corresponding deformation state estimates in the independent validation data subset are calculated. If the statistical characteristic value of the residual exceeds a preset residual threshold, a fusion parameter adjustment command is triggered, and the internal parameters of the target data fusion operator combination are back-optimized and updated based on the residuals.

[0040] In the specific implementation, the preliminary fusion results are spatially meshed, and the spatial area corresponding to the foundation pit support structure is discretized into regular spatial grid nodes. In the example scenario, the foundation pit support structure is a continuous wall made of steel sheet piles. The corresponding spatial area is defined as a rectangular area of ​​20 meters along the length direction and 15 meters along the depth direction of the wall. It is divided in the length and depth directions at 1-meter intervals to generate a regular grid containing 300 spatial grid nodes of 20 by 15. Each spatial grid node is identified by its planar coordinates (x, y), where x represents the position along the length direction of the foundation pit and y represents the depth. Based on the Kriging spatial estimation algorithm, the deformation state estimate of each spatial grid node is calculated using data from known locations in the preliminary fusion results. In specific implementation, the preliminary fusion results are deformation data from different monitoring points, which correspond to known points in the spatial grid. For example, sensors are deployed at depths of 0 meters, 4 meters, 7.5 meters, 8 meters, and 15 meters, with coordinates of (10,0), (10,4), (10,7.5), (10,8), and (10,15), respectively. The first number 10 in the parentheses indicates the position 10 meters from the center in the length direction. The Kriging spatial estimation algorithm uses the deformation data from these known points as a basis to construct a variogram model to describe spatial correlation, thereby predicting the deformation state estimate of spatial grid nodes without deployed sensors. It can be understood that Kriging estimation is an optimal linear unbiased estimation. The deformation state estimates of all spatial grid nodes are integrated to form the overall spatial deformation state field of the support structure. The overall spatial deformation state field of the support structure is a two-dimensional matrix. The row index of the matrix corresponds to the grid number in the depth direction, and the column index corresponds to the grid number in the length direction. The matrix element value is the deformation state estimate of the corresponding spatial grid node. For example, the deformation state estimate of the spatial grid node at coordinate (10,2) is 3.5 mm, and the deformation state estimate of the spatial grid node at coordinate (5,10) is 1.2 mm. From the overall spatial deformation state field of the support structure, the deformation distribution on the preset characteristic profiles is extracted to form the fused spatial deformation state. The characteristic profiles include the central longitudinal profile along the long side of the foundation pit and the transverse profile at the maximum excavation depth. In some embodiments, the central longitudinal profile along the long side of the foundation pit refers to the entire vertical line with a length direction coordinate of x=10 meters. The deformation state estimates of all spatial grid nodes at x=10 meters in the overall spatial deformation state field of the support structure are extracted and arranged by depth to form the longitudinal profile deformation distribution curve. The transverse profile at the maximum excavation depth refers to the horizontal line with a depth of y=12 meters. The deformation state estimates of all spatial grid nodes at y=12 meters in the overall spatial deformation state field of the support structure are extracted and arranged by length to form the transverse profile deformation distribution curve. The longitudinal profile deformation distribution curve and the transverse profile deformation distribution curve together constitute the fused spatial deformation state, which is presented in the form of a visual chart or data list.

[0041] In practical implementation, the verification step of the fused spatial deformation state separates an independent verification data subset from the initial deformation monitoring data sequence that did not participate in the adaptive data fusion model fusion process. This separation is achieved through preset rules; for example, data from one sampling point is retained every five sampling points and not input into the adaptive data fusion model. These retained data points constitute the independent verification data subset, which includes the monitoring point location, timestamp, and measured deformation value. In the fused spatial deformation state, the spatial location corresponding to the monitoring point in the independent verification data subset is located, and the estimated deformation state value at that location is read. The fused spatial deformation state is a continuous spatial field. For any monitoring point in the independent verification data subset, its spatial coordinates (x, y) are known. The estimated deformation state value at that coordinate is read from the overall spatial deformation state field of the support structure using nearest neighbor interpolation or bilinear interpolation. For example, if a monitoring point in the independent verification data subset has coordinates (10, 2) and a measured displacement of 3.8 mm, the estimated deformation state value for that point from the overall spatial deformation state field of the support structure is 3.5 mm. Calculate the residuals between the measured values ​​and the corresponding deformation state estimates in the independent validation dataset subset. The residual calculation formula is as follows: , in: This represents the residual at the m-th verification point. This represents the measured value at the m-th verification point. This represents the deformation state estimate of the m-th verification point read from the fused spatial deformation state. The residuals of data points in all independent verification data subsets are calculated to obtain a set of residual sequences. If the statistical characteristic value of the residual exceeds a preset residual threshold, a fusion parameter adjustment command is triggered. Based on the residual, the internal parameters of the target data fusion operator combination are back-optimized and updated. In some embodiments, the statistical characteristic value of the residual is selected as the root mean square error of the residual sequence. The preset residual threshold is 0.5 mm. If the calculated root mean square error of the residual sequence is 0.6 mm, exceeding the threshold, a fusion parameter adjustment command is triggered. The adjustment command back-optimizes and updates the internal parameters of the target data fusion operator combination based on the magnitude of the residual sequence. For example, in the weighted fusion operator, the weights of the displacement sensor (0.4), the inclinometer (0.4), and the strain sensor (0.2) are minimized using the gradient descent method. It can be understood that the updated weights may become 0.45, 0.35, or 0.2. The adjusted weights will be used in the adaptive data fusion process during subsequent monitoring periods. Optionally, the interpolation model parameters of the spatial interpolation operator can also be adjusted based on the residual.

[0042] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A deformation monitoring method for sheet pile foundation pit support, characterized in that, The method includes: The pile parameters of the sheet pile, the soil profile information of the foundation pit, and the preset multiple excavation stages are obtained. The lateral deformation prediction structure of the foundation pit is established by integrating the pile parameters and soil profile information. Based on the lateral deformation prediction structure of the foundation pit, a set of monitoring points is generated according to the monitoring point deployment strategy. By combining the preset multiple excavation stages with the foundation pit lateral deformation prediction structure, a monitoring data sampling strategy is derived. The monitoring data sampling strategy includes a set of sampling intervals and a set of data collection durations for different excavation stages. Using the set of monitoring points and the monitoring data sampling strategy, on-site monitoring is initiated to obtain an initial deformation monitoring data sequence; Deformation trend curves of monitoring points are extracted from the initial deformation monitoring data sequence, and key deformation stages are identified through the deformation trend curves. During the monitoring process, an adaptive data fusion model is dynamically invoked based on the key deformation stages. The adaptive data fusion model is used to fuse and reconstruct the deformation monitoring data sequence to generate a fused spatial deformation state that reflects the overall spatial deformation state of the support structure.

2. The deformation monitoring method for sheet pile foundation pit support as described in claim 1, characterized in that, A structure for predicting the lateral deformation of the foundation pit is established by integrating the pile parameters and soil profile information, including: The soil profile information includes the soil type and depth boundaries of each soil layer; The structure for predicting the lateral deformation of the foundation pit includes the estimated deformation values ​​for multiple soil layer boundary points. Based on the soil type of each soil layer in the soil profile information, the corresponding soil elastic modulus and cohesion parameters are obtained; The pile parameters are imported into the lateral deformation analysis framework, which uses the pile's moment of inertia and section modulus as the calculation basis. In the lateral deformation analysis framework, the soil elastic modulus and cohesion parameters are superimposed layer by layer according to depth to generate a soil-pile interaction model. A standard lateral load distribution defined by the excavation stage is applied to the soil-pile interaction model, and the estimated lateral displacement of multiple soil layer boundary points is calculated. The estimated lateral displacement values ​​of the multiple soil layer boundary points are integrated to construct the estimated lateral deformation structure of the foundation pit. The estimated lateral deformation structure of the foundation pit links the soil layer depth and the estimated displacement value in tabular form.

3. The deformation monitoring method for sheet pile foundation pit support as described in claim 2, characterized in that, Based on the aforementioned strategy for predicting lateral deformation of the foundation pit, a set of monitoring point distributions is generated, including: The monitoring point distribution set includes multiple monitoring points and the sensor type for each monitoring point. The sensor types include inclinometers, strain sensors, and displacement sensors. The structure for predicting the lateral deformation of the foundation pit is analyzed to locate the abrupt depth of the change in the predicted lateral displacement. Based on the pile length in the pile parameters, preliminary monitoring points are generated at the top, middle, and bottom of the pile, as well as at the location of the abrupt change depth. Based on the preset multiple excavation stages, determine the exposure time and stress change characteristics of each preliminary monitoring point during the excavation process; Based on the stress change characteristics, at least one sensor type is assigned to each preliminary monitoring point, wherein displacement sensors are associated with the top and bottom of the pile, inclinometer sensors are associated with the abrupt change depth location, and strain sensors are associated with the middle of the pile. The initial monitoring points and their assigned sensor types are integrated to form the monitoring point distribution set.

4. The deformation monitoring method for sheet pile foundation pit support as described in claim 3, characterized in that, By combining the preset multiple excavation stages with the predicted lateral deformation structure of the foundation pit, a monitoring data sampling strategy is derived, including: Extract the stage depth and stage duration of the preset multiple excavation stages; In the lateral deformation prediction structure of the foundation pit, the lateral displacement prediction value corresponding to the depth of each stage is found. Based on the rate of change of the lateral displacement prediction value, the initial monitoring sampling interval for each excavation stage is set, where the stage with a large rate of change corresponds to a short sampling interval. A safety factor matrix is ​​introduced to calibrate each initial monitoring sampling interval, resulting in a calibrated sampling interval; Based on the stage duration and the post-calibration sampling interval, the data acquisition duration of each stage is calculated, and the post-calibration sampling interval and data acquisition duration of all stages are summarized into the monitoring data sampling strategy.

5. A deformation monitoring method for sheet pile foundation pit support as described in claim 4, characterized in that, Deformation trend curves of monitoring points are extracted from the initial deformation monitoring data sequence, and key deformation stages are identified through the deformation trend curves, including: With time as the horizontal axis and the displacement at the top of the pile, the strain at the middle of the pile, and the displacement at the bottom of the pile extracted from the initial deformation monitoring data sequence as the vertical axis, separate curves of displacement versus time and strain versus time are plotted. The trend decomposition algorithm is applied to process the displacement-time curve and the strain-time curve to separate the long-term trend component, periodic component and random noise component. Extract the long-term trend component and calculate the curvature value of the long-term trend component between adjacent sampling points; The time interval in which the sampling points with curvature values ​​exceeding the curvature threshold are marked is defined as a potential critical deformation segment; By combining the preset multiple excavation stages, the specific excavation stage to which the potential critical deformation section belongs is determined, and the specific excavation stage is marked as the critical deformation stage.

6. The deformation monitoring method for sheet pile foundation pit support as described in claim 5, characterized in that, During the monitoring process, an adaptive data fusion model is dynamically invoked based on the key deformation stages, including: An adaptive data fusion model containing multiple data fusion operators is pre-constructed, including a moving average operator, a weighted fusion operator, and a spatial interpolation operator; Establish a stage-operator mapping table, which maps different excavation stages to different combinations of data fusion operators; When the monitoring process reaches the critical deformation stage, the target data fusion operator combination corresponding to the critical deformation stage is dynamically loaded according to the stage-operator mapping table. Read the active sensor types in the current stage from the set of monitoring point distributions, and configure the internal parameters of the target data fusion operator combination according to the sensor types.

7. The deformation monitoring method for sheet pile foundation pit support as described in claim 6, characterized in that, An adaptive data fusion model is used to fuse and reconstruct deformation monitoring data sequences, including: Using the calibrated sampling interval of the current stage in the monitoring data sampling strategy as a time window, the deformation monitoring data sequence is extracted; The extracted deformation monitoring data sequence is input into the configured target data fusion operator combination; The extracted deformation monitoring data sequence is smoothed in the time dimension by using the moving average operator to obtain a smoothed data sequence. By using a spatial interpolation operator, based on the spatial positional relationship of the monitoring points in the monitoring point distribution set, the smoothed data sequence is interpolated in the spatial dimension to generate a spatially continuous deformation field. By using a weighted fusion operator, different weights are assigned to the sensor types at the monitoring points, and data from different types of sensors in a continuous spatial deformation field are weighted and synthesized to output preliminary fusion results.

8. The deformation monitoring method for sheet pile foundation pit support as described in claim 7, characterized in that, Generate a fused spatial deformation state that reflects the overall spatial deformation state of the support structure, including: The preliminary fusion results are spatially meshed, and the spatial region corresponding to the foundation pit support structure is discretized into regular spatial mesh nodes. Based on the Kriging spatial estimation algorithm, the deformation state estimate of each spatial grid node is calculated using data from known locations in the preliminary fusion results. The deformation state estimates of all spatial grid nodes are integrated to form the overall spatial deformation state field of the support structure; From the overall spatial deformation state field of the support structure, the deformation distribution on the preset feature profile is extracted to form the fused spatial deformation state. The feature profile includes the central longitudinal profile along the long side of the foundation pit and the transverse profile at the maximum excavation depth.

9. A deformation monitoring method for sheet pile foundation pit support as described in claim 8, characterized in that, The method also includes a verification step for the spatial deformation state after fusion: From the initial deformation monitoring data sequence, separate the independent validation data subset that did not participate in the adaptive data fusion model fusion process; In the fused spatial deformation state, locate the spatial position corresponding to the monitoring point in the independent verification data subset, and read the deformation state estimate of the spatial position; Calculate the residuals between the measured values ​​and the corresponding deformation state estimates in the independent validation data subset; If the statistical characteristic value of the residual exceeds the preset residual threshold, a fusion parameter adjustment instruction is triggered, and the internal parameters of the target data fusion operator combination are optimized and updated in reverse based on the residual.

10. A deformation monitoring system for sheet pile foundation pit support, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the deformation monitoring method for steel sheet pile foundation pit support as described in any one of claims 1 to 9.

Images