A tourism island bedrock shoreline safety monitoring and early warning system and method

Abstract

The present application belongs to the technical field of tourism route safety monitoring, and discloses a kind of tourism island bedrock coast line safety monitoring early warning system and method.The system includes: data acquisition module is used to collect fissure activity, globular weathering body activity, surface displacement, stress and strain, rainfall, sea wave power and other monitoring data information.Data processing module is used to receive the data information collected, and the data is stored and calculated.Danger alarm module sets different types of observation value change threshold according to different monitoring object entities and in combination with actual conditions.Safety hazard prediction module divides the frequently occurring alarm area into key monitoring areas, predicts the future changes in the area, and gives safety hazard removal and protection repair prompts in combination with the actual situation of tourism development of island bedrock coast line.The present application provides real-time monitoring data for island coast protection, repair and island ecological civilization construction.

Description

Technical Field

[0001] This invention belongs to the field of shoreline safety monitoring technology, and in particular relates to a safety monitoring and early warning system and method for bedrock shorelines of tourist islands. Background Technology

[0002] The bedrock coastline of the island boasts abundant tourism resources, including various symbolic rocks, geological relics such as sea cliffs and sea caves, beautiful coastal zones, cultural landscapes, and recreational fishing. It is the most developed and utilized section of the island tourism route and the most popular island tourism route among tourists.

[0003] However, the bedrock coastline of islands is constantly subjected to the combined effects of wave erosion and weathering, making it prone to geological disasters such as coastline erosion, expansion of faults, and rockfalls. Developing tourist routes on the bedrock coastline requires manual excavation, which can create the risk of rockfalls. Faulting activity can also lead to spheroidal weathering in areas with dense fault development. Over time, these weathered deposits may become unstable and suddenly roll down, causing unexpected harm to tourists and island residents. Limited monitoring equipment along island tourist routes can effectively monitor key areas of tourist activity, but it is difficult to effectively monitor and predict relatively slow changes such as landslides, rockfalls, faulting activity, island-land degradation, and the amount of bedrock coastline erosion.

[0004] Based on the above analysis, the problems and defects of the existing technology are as follows: the existing technology mainly monitors landslide bodies on slopes and constructs slope monitoring systems, but its video monitoring effect is poor; the existing technology does not utilize different monitoring methods for different monitoring, and cannot achieve the purpose of timely early warning; the existing technology has poor prediction effect on safety hazards and is inaccurate in predicting sudden safety hazards. Summary of the Invention

[0005] To overcome the problems existing in related technologies, the present invention discloses an embodiment of a safety monitoring and early warning system and method for bedrock coastlines of tourist islands.

[0006] The technical solution is as follows: A safety monitoring and early warning system for bedrock coastlines of tourist islands, which combines UAV oblique photography technology to establish a real-scene 3D model, realizing visualized monitoring, early warning, and prediction of safety hazards, specifically including:

[0007] The data acquisition module is used for monitoring fracture activity, spheroidal weathering body activity, surface displacement, stress and strain, rainfall, and ocean wave dynamics.

[0008] The data processing module is used to receive the collected data information, store and calculate the data, and compare the obtained spatial coordinate position, stress value, deformation value, rainfall and other observed index values ​​with the initial value to obtain the change value 1; compare it with the previous recorded value to obtain the change value 2, and display the two types of change in the form of charts;

[0009] The danger alarm module sets different thresholds for changes in observed values ​​based on different monitored entities. When the change in the monitored value exceeds the threshold, a danger alarm is triggered, and abnormal values ​​are reported and archived.

[0010] The safety hazard prediction module divides areas with frequent alarms into key monitoring areas, combines historical monitoring data to predict the future trend of these areas, and provides prompts for safety hazard removal and protection / repair. If the safety hazard is removed, a new prediction is made.

[0011] Furthermore, in the data acquisition module, fracture activity monitoring is used to monitor fractures developed along the bedrock shoreline and tourist routes, including fracture displacement, width expansion and contraction variables, and newly formed fractures.

[0012] Surface displacement monitoring involves installing monitoring equipment on slopes, steep exposed rock surfaces, and weathered spheroidal weathered bodies along tourist routes that are prone to collapse. This equipment is combined with a positioning system to monitor displacement changes. For tourist roads built along the coast, distributed optical fibers are used for full-coverage monitoring.

[0013] Stress and strain monitoring is used to monitor the stress state and strain of bridges, viewing platforms and tourist facilities in fracture distribution areas built on tourist roads.

[0014] Rainfall monitoring involves installing monitoring equipment in exposed slopes and vegetated areas along tourist routes to obtain information on changes in rainfall.

[0015] Wave dynamics monitoring involves installing monitoring equipment in areas of bedrock shorelines developed for tourism that are eroded by waves and have dense fractures, to obtain dynamic data on the impact force of waves.

[0016] Furthermore, the data acquisition stations and monitoring stations in the data acquisition module are displayed on the real-world 3D model according to their actual locations.

[0017] Another objective of this invention is to provide a method for safety monitoring and early warning of bedrock coastlines on tourist islands. This method operates within the aforementioned safety monitoring and early warning system for bedrock coastlines on tourist islands and includes:

[0018] S1 collects monitoring information on fracture activity, spheroidal weathering body activity, surface displacement, stress and strain, rainfall, and ocean wave dynamics.

[0019] S2 transmits the collected data to the data processing module via optical fiber, stores and calculates the data, compares the obtained spatial coordinates, stress values, and rainfall values ​​with the initial values, and compares them with the previous recorded values ​​to obtain the changes and display them in the form of charts.

[0020] S3: Set different thresholds for changes in observed values ​​based on different monitored entities. When the change in monitored values ​​exceeds the threshold, a danger alarm is triggered, and abnormal values ​​are reported and archived.

[0021] S4 divides areas where alarms frequently occur into key monitoring areas, predicts future changes in these areas based on historical monitoring data, and provides prompts for safety hazard removal and protection / repair. If the safety hazard is removed, a new prediction is made.

[0022] In step S2, the obtained spatial coordinates, stress values, and rainfall values ​​are compared with the initial values ​​and with the previously recorded values. Specifically, this includes:

[0023] Step 1: Input the current observed value X and the initial value T of the current observed index; obtain the change X1 relative to the initial value and the change X2 relative to the previous recorded value;

[0024] Step 2: Compare and verify the changes X1 and X2 with the set threshold Y for a certain observation element;

[0025] Step 3: Based on the specified number of decision trees N for the current stage. stage Using the label as the fitting threshold, each decision tree is built sequentially. Fit the change value of the change function of the current model;

[0026] Step 4: The model fitting is complete in the current stage. Calculate the fitting change value res in the current stage, where res = actual - pred, and pred is the predicted value of the training data X in the current stage; the expression is:

[0027]

[0028] In the formula, rate is the learning rate. For the i-th decision tree in the current stage The predicted values ​​on the training set X are updated to the values ​​of the data to be fitted, actual = res; n_estimators are the parameters of the random forest;

[0029] Step 5: Perform a Jarque-Bera test on the change in the previous recorded value X and the fitted change value res.

[0030] Step Six: When the evaluation function on the given test set T no longer decreases within the specified number of early_stop stages, the model training stops in time, the optimal number of iteration stages N is obtained, the model training is completed, and the process jumps to Step Seven. Otherwise, if the evaluation function on the test set T continues to decrease, the process jumps to Step Four and continues the stage iteration.

[0031] Step 7: Predict the test set. Based on the optimal number of iteration stages N for the model, obtain the test set prediction result pred.

[0032] Steps three through seven in this invention are actually an intermediate calculation process. The threshold of the change is obtained through calculation, which is used for the next step S3. Different types of observation value change thresholds are set according to different monitoring entities. When the change of the monitored value exceeds the threshold, a danger alarm is issued, providing technical support.

[0033] In step two, the original threshold y of the change X of the previous recorded value is examined. The specific implementation steps are as follows:

[0034] (1) Calculate the data distribution skewness S based on the threshold of change in the previous recorded value using the following formula:

[0035]

[0036] Where actual is the threshold of the data to be fitted, mean is the mean of the sample threshold, size is the sample size, and std is the standard deviation of the sample threshold;

[0037] (2) Calculate the kurtosis K of the data distribution based on the threshold of the change in the previous recorded value using the following formula:

[0038]

[0039] (3) Construct the chi-square statistic JB according to the following formula:

[0040]

[0041] Where size is the sample size, S is the skewness of the data distribution, and K is the kurtosis of the data distribution;

[0042] Test the significance of the chi-square statistic JB statistic, examine the degree of bias of the chi-square statistic JB to determine whether to make data changes; if the chi-square statistic JB exceeds the set threshold, proceed to (4), otherwise proceed to (5).

[0043] (4) If the data exceeds the set threshold, determine whether the upper limit of the number of data changes has been reached. Let the upper limit of the number of changes be diff. If diff≤0, jump to (5). Otherwise, update the diff value to diff-1 and perform a Boxcox transformation on the data. The λ value is traversed by the following formula. Select the λ that maximizes the likelihood function L of the data threshold label after the transformation to generate the transformation expression:

[0044]

[0045] Where actual is the threshold of the data to be fitted, a new data threshold is obtained, and it is used as the first stage fitting threshold label. The current stage judgment variable jud[1] = True is set.

[0046] (5) If the data is not obviously biased or the number of data threshold changes reaches the target upper limit, take the original data threshold y as the first stage fitting threshold label, and set the current stage judgment variable jud[1] = False.

[0047] In step seven, the test set prediction result pred is obtained based on the optimal number of iteration stages N of the model, specifically including:

[0048] (1) Calculate the predicted value pred of the stage model on the test set T up to the current stage. The expression is:

[0049]

[0050] In the formula, pred represents the prediction result on the test set, and rate represents the learning rate. For stage j decision tree The prediction results The inverse transform of the threshold pretransform for stage j is expressed as follows:

[0051]

[0052] In the formula, F stage -1 Here, jud[j] is the inverse transform function of the current stage model, sigmoid[j] is the judgment variable for the current stage j, and sigmoid[j] is the inverse transform function of the current stage model. -1 The inverse transform of the sigmoid(·) transform, boxcos -1 (·) is the inverse transformation of the boxcos(·) transformation;

[0053] Prediction results for stage j boxcos -1 The expression (·) is shown below:

[0054]

[0055] sigmod -1 The expression (·) is shown below:

[0056]

[0057] (2) Calculate the change function value of the current stage: loss = L(pred, y T ), y T Let T be the original threshold for the test set, and L(·) be the variation function used by the system. For regression problems, this is a squared variation function, where loss = (pred - y) T ) 2 .

[0058] In step S4, areas where alarms frequently occur are designated as key monitoring areas. Based on historical monitoring data, future changes in these areas are predicted, including:

[0059] The image sequence of frequently alarmed areas taken from different angles is used as the input set. Feature matching point pairs of the images are obtained through feature extraction and matching, and these pairs are then subjected to evolutionary processing.

[0060] Based on the geological evolution and security threat model, feature points of candidate images are selected as seed points to be matched and filtered in their surrounding neighborhoods to obtain evolution matching point pairs;

[0061] The image acquisition instrument is calibrated, and its intrinsic and extrinsic parameters are obtained by combining the matching point pairs; the three-dimensional model points are then recovered based on the image acquisition instrument parameters and the matching point pairs.

[0062] A geological evolution and security threat model is used for reconstruction. Seed model points are selected to generate initial points, which then evolve within their grid neighborhood.

[0063] By filtering out errors based on constraints, an accurate evolutionary 3D point cloud model is obtained.

[0064] Furthermore, the geological evolution and security threat model specifically includes:

[0065] For each feature point f in the reference image, find the corresponding candidate matching point f′ in the candidate image according to the epipolar constraint; using the geological evolution and security threat model, select the zero-mean normalized cross-correlation coefficient ZNCC as the objective function, calculate the ZNCC values ​​of the matching point pairs, and sort them by size:

[0066]

[0067] Where x is the coordinate information of image feature point f in the image, and x′ is the coordinate information of image feature point f′ in the image; I(x) and I(x′) represent the pixel brightness at x coordinate and x′ coordinate; and This represents the average pixel brightness of the image window centered at x and the image window W centered at x′;

[0068] Feature points greater than a threshold μ1 are selected as seed points for neighborhood evolution, and feature points greater than a threshold μ2 are selected as reserve matching points (μ1>μ2). For all matching points in the reference image, a one-to-many match is established in the center of the candidate image with a fixed window size. For points in the reference image, points in other images are matched, and a mixed match is established for all points within the window. Under the premise of satisfying the disparity gradient constraint and confidence constraint, the ZNCC of the evolving matching point pairs is calculated, and evolution points greater than a threshold μ3 are selected as seed points for secondary evolution, and evolution points greater than a threshold μ4 are selected as reserve matching points (μ3>μ4).

[0069] Assuming u′ and u are a pair of image matching points, and x′ and x are another adjacent pair of image matching points, the disparity gradient constraint formula is:

[0070] ||(u′-u)-(x′-x)|| ∞ ≤ε

[0071] Where ε is the threshold of the disparity gradient; the disparity gradient constraint reduces the ambiguity of image matching.

[0072] The formula for confidence constraints is:

[0073] s(x)=max{|I(x+Δ)-I(x)l, Δ∈{(1,0), (-1,0), (0,1), (0,-1)}}

[0074] Using confidence constraints can improve the reliability of matching evolution and obtain evolutionary matching point pairs.

[0075] Furthermore, the calibration process for the image acquisition instrument involves calculating its internal parameters based on the imaging principle; selecting two input images as references based on the feature points and matching of the image sequence; calculating the fundamental matrix F of the reference image point pair, where F satisfies the equation x′Fx=0, and x′ and x are a pair of image matching points; estimating the initial values ​​K′ and K of the intrinsic parameter matrix of the reference image pair; calculating the essential matrix of the image point pair and extracting rotation and translation components; and using triangulation to find the three-dimensional model points corresponding to the feature points, given the intrinsic and extrinsic parameters of the image acquisition instrument and the feature matching point pairs.

[0076] Combining all the above technical solutions, the advantages and positive effects of this invention are as follows: In light of the actual situation of tourism development along the bedrock coastline of islands, this invention proposes a safety monitoring and early warning system for the bedrock coastline of tourist islands, so as to effectively protect the safety of tourists and island residents and provide real-time monitoring data for coastline protection.

[0077] Innovation of the Safety Monitoring System: Previous safety monitoring of islands has largely focused on slopes after artificial quarrying, primarily monitoring landslides and constructing slope monitoring systems. The monitoring and early warning system proposed in this invention targets tourist routes developed along the bedrock coastline of tourist islands. In addition to real-time monitoring, the system also has predictive capabilities. It also effectively compensates for the shortcomings of many tourist islands that rely solely on video surveillance for monitoring.

[0078] Improvements to the data acquisition module: (1) More comprehensive monitoring content. The monitoring content has been expanded to include elements such as fissure activity and wave dynamics, making the monitoring content more systematic. Fissure activity and wave dynamics are key influencing factors causing bedrock shoreline erosion. In particular, the stability of tourist facilities and viewing platforms that are significantly affected by faults is monitored; the stability of spheroidal weathered bodies under the dual influence of faults and weathering is monitored. (2) Combination of point and area data acquisition. Point monitoring stations are set up for key landslide areas, rock collapse areas, spheroidal weathered rock blocks, tourist facilities, etc.; for areas with densely distributed faults and tourist roads, distributed optical fibers are used for continuous deployment to achieve timely early warning.

[0079] Improvements to the Safety Hazard Prediction and Data Processing Modules: Most monitoring and early warning systems lack embedded safety hazard prediction functionality. Spheroidal weathering, fissure activity, and bedrock shoreline erosion are long-term processes; quantitative changes are the result of long-term accumulation, but abrupt changes are sudden. Therefore, predicting potential safety hazards based on long-term monitoring data is crucial. To achieve this, on the one hand, accurate data and long-term data accumulation are required, thus placing high demands on the data processing module; on the other hand, it is necessary to construct a process model of the occurrence and development of safety hazards. This invention constructs a geological evolution and safety threat model of spheroidal weathering. This invention obtains an accurate three-dimensional point cloud model of evolution by calculating the coordinate positions obtained from the data, the changes relative to the initial value and the previous recorded value, and dividing frequently alarmed areas into key monitoring areas. Combined with historical monitoring data, it predicts future changes in these areas. Attached Figure Description

[0080] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure;

[0081] Figure 1 This is a schematic diagram of a safety monitoring and early warning system for bedrock coastlines of tourist islands provided in an embodiment of the present invention;

[0082] Figure 2 This is a schematic diagram of the safety monitoring and early warning system for bedrock coastlines of tourist islands provided in an embodiment of the present invention;

[0083] Figure 3 This is a flowchart of the safety monitoring and early warning method for bedrock coastlines of tourist islands provided in an embodiment of the present invention;

[0084] In the diagram: 1. Data acquisition module; 2. Data processing module; 3. Danger alarm module; 4. Safety hazard prediction module. Detailed Implementation

[0085] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0086] Example 1, such as Figure 1 As shown in the figure, this embodiment of the invention provides a safety monitoring and early warning system for bedrock coastlines of tourist islands, including a data acquisition module 1, a data processing module 2, a hazard alarm module 3, and a safety hazard prediction module 4. It utilizes a real-scene 3D model established using UAV oblique photography technology to conduct visualized safety hazard monitoring and early warning.

[0087] The data acquisition module 1 is used for monitoring fracture activity, spheroidal weathering body activity, surface displacement, stress and strain, rainfall, and ocean wave dynamics, etc.; all data acquisition stations are displayed on a real-world 3D model according to their actual locations.

[0088] (1) Fracture activity monitoring: mainly targeting large-scale fractures developed along the bedrock shoreline, monitoring fracture displacement and width expansion / contraction variables;

[0089] (2) Surface displacement monitoring: Monitoring equipment is installed on slopes and steep exposed rock surfaces that may collapse along tourist routes, as well as spheroidal weathered bodies that are highly weathered but difficult to remove. Combined with a high-precision positioning system, the displacement changes are monitored. For tourist roads built along the coast, distributed optical fiber is used for full-coverage monitoring.

[0090] (3) Stress and strain monitoring: The main focus is on monitoring the stress state and strain of bridges, viewing platforms and tourist facilities in the fracture distribution area built on tourist roads.

[0091] (4) Rainfall monitoring: Rainfall has a significant impact on island tourism routes, especially on newly built green belts and bare slopes. Monitoring equipment is installed in the vegetated slope areas of the tourism routes to obtain information on rainfall changes.

[0092] (5) Wave dynamic monitoring: Install monitoring equipment in the depressions of the bedrock coastline where waves erode the sea to obtain dynamic data on the impact force of the waves.

[0093] The data processing module 2 is used to receive the collected data information, store and calculate the data, compare the obtained spatial coordinate position, stress value, deformation value, rainfall and other observed index values ​​with the initial value to obtain the change value 1; compare it with the previous recorded value to obtain the change value 2, and display the two types of change in the form of a chart.

[0094] The hazard alarm module 3 sets different threshold values ​​for different types of observed values ​​based on different monitored entities. When the change in a monitored value exceeds the threshold, a hazard alarm is triggered, and the abnormal value is reported and archived. Relevant management personnel promptly verify the alarm information on-site and carry out necessary measures such as protection and repair or elimination of safety hazards. After the measures are completed, the initial values ​​for the next stage of monitoring are set.

[0095] The safety hazard prediction module 4: divides areas where alarms frequently occur into key monitoring areas, combines historical monitoring data to predict the future trend of these areas, and provides prompts for safety hazard removal and protection repair; if the safety hazard is removed, a new prediction is made.

[0096] like Figure 2 The diagram illustrates the principle of the safety monitoring and early warning system for bedrock coastlines of tourist islands provided in this embodiment of the invention.

[0097] Example 2, as Figure 3 As shown in the embodiment of the present invention, the method for safety monitoring and early warning of bedrock coastlines on tourist islands includes:

[0098] S1 collects monitoring information on fracture activity, surface displacement, stress and strain, rainfall, and ocean wave dynamics.

[0099] S2 transmits the collected data to the data processing terminal via optical fiber, stores the data and performs calculations, calculating the changes in the obtained spatial coordinates, stress values, rainfall and other observation indicators compared to the initial observation values ​​and the previous observation record values; and displays the changes in the form of charts.

[0100] S3: Set different types of numerical change thresholds according to different monitored entities; when the change in monitored values ​​exceeds the threshold, a danger alarm will be triggered.

[0101] S4 identifies areas with frequent alarms as key monitoring areas, and, based on historical monitoring data, predicts future changes in these areas, providing prompts for safety hazard removal and protection / repair.

[0102] In this embodiment of the invention, step S2, calculating the coordinate position of the acquired data and the change relative to the initial value and the previous recorded value, includes:

[0103] Step 1: Input the current observed value X and the initial value T of the current observed index; obtain the change X1 relative to the initial value and the change X2 relative to the previous recorded value;

[0104] Step 2: Compare and verify the changes X1 and X2 with the set threshold Y for a certain observation element;

[0105] Step 3: Based on the specified number of decision trees N for the current stage. stage Using the label as the fitting threshold, each decision tree is built sequentially. Fit the change value of the change function of the current model;

[0106] Step 4: The model fitting is complete in the current stage. Calculate the fitting change value res in the current stage, where res = actual - pred, and pred is the predicted value of the training data X in the current stage.

[0107]

[0108] In the formula, rate is the learning rate. For the i-th decision tree in the current stage The predicted values ​​on the training set X are updated to the values ​​of the data to be fitted, actual = res; n_estimators are the parameters of the random forest;

[0109] Step 5: Perform a Jarque-Bera test on the change in the previous recorded value X and the fitted change value res.

[0110] Step Six: When the evaluation function on the given test set T no longer decreases within the specified number of early_stop stages, the model training stops in time, the optimal number of iteration stages N is obtained, the model training is completed, and the process jumps to Step Seven. Otherwise, if the evaluation function on the test set T continues to decrease, the process jumps to Step Four and continues the stage iteration.

[0111] Step 7: Predict the test set. Based on the optimal number of iteration stages N for the model, obtain the test set prediction result pred.

[0112] In this embodiment of the invention, the specific implementation steps of step two are as follows:

[0113] (1) Calculate the data distribution skewness S based on the threshold of change in the previous recorded value using the following formula:

[0114]

[0115] Where actual is the threshold of the data to be fitted, mean is the mean of the sample threshold, size is the sample size, and std is the standard deviation of the sample threshold;

[0116] (2) Calculate the kurtosis K of the data distribution based on the threshold of the change in the previous recorded value using the following formula:

[0117]

[0118] (3) Construct the chi-square statistic JB according to the following formula:

[0119]

[0120] Where size is the sample size, S is the skewness of the data distribution, and K is the kurtosis of the data distribution;

[0121] Test the significance of the chi-square statistic JB statistic, examine the degree of bias of the chi-square statistic JB to determine whether to make data changes; if the chi-square statistic JB exceeds the set threshold, proceed to step (4), otherwise proceed to step (5).

[0122] (4) If the data exceeds the set threshold, determine whether the upper limit of the number of data changes has been reached. Let the upper limit of the number of changes be diff. If diff≤0, jump to (5). Otherwise, update the diff value to diff-1 and perform a Boxcox transformation on the data. The λ value is traversed by the following formula. Select the λ that maximizes the likelihood function L of the data threshold label after the transformation to generate the transformation expression:

[0123]

[0124] Where actual is the threshold of the data to be fitted, a new data threshold is obtained, and it is used as the first stage fitting threshold label. The current stage judgment variable jud[1] = True is set.

[0125] (5) If the data is not obviously biased or the number of data threshold changes reaches the target upper limit, take the original data threshold y as the first stage fitting threshold label, and set the current stage judgment variable jud[1] = False.

[0126] In step seven, the test set prediction result pred is obtained based on the optimal number of iteration stages N of the model, specifically including:

[0127] (1) Calculate the predicted value pred of the stage model on the test set T up to the current stage. The expression is:

[0128]

[0129] In the formula, pred represents the prediction result on the test set, and rate represents the learning rate. For stage j decision tree The prediction results The inverse transform of the threshold pretransform for stage j is expressed as follows:

[0130]

[0131] In the formula, F stage -1 Here, jud[j] is the inverse transform function of the current stage model, sigmoid[j] is the judgment variable for the current stage j, and sigmoid[j] is the inverse transform function of the current stage model. -1 For the inverse sigmoid (·) transform, boxcos -1 (·) is the inverse transformation of the boxcos(·) transformation;

[0132] Prediction results for stage j boxcos -1 The expression (·) is shown below:

[0133]

[0134] sigmod -1 The expression (·) is shown below:

[0135]

[0136] (2) Calculate the change function value of the current stage: loss = L(pred, y T ), y T Let T be the original threshold for the test set, and L(·) be the variation function used by the system. For regression problems, this is a squared variation function, where loss = (pred - y) T ) 2 .

[0137] In this embodiment of the invention, step S4, which involves dividing areas where alarms frequently occur into key monitoring areas and, in conjunction with historical monitoring data, predicting future changes in these areas, includes:

[0138] Image sequences of frequently alarmed areas acquired from different angles are used as the input set. Feature matching point pairs are obtained through feature extraction and matching, and these pairs undergo evolution processing. Based on the geological evolution and security threat model, feature points of candidate images are selected as seed points for matching and filtering in their surrounding neighborhoods to obtain evolved matching point pairs. The image acquisition instrument is calibrated, and its intrinsic and extrinsic parameters are obtained by combining the matching point pairs. The three-dimensional model points are recovered based on the image acquisition instrument parameters and the matching point pairs. Reconstruction is performed using the geological evolution and security threat model, seed model points are selected to generate initial points, and these points evolve within their grid neighborhoods. Errors are filtered according to constraints to obtain an accurate evolved three-dimensional point cloud model.

[0139] In one embodiment, the geological evolution and security threat model specifically includes:

[0140] For each feature point f in the reference image, a corresponding candidate matching point f' is found in the candidate image based on the epipolar constraint; then, using the geological evolution and security threat model, the zero-mean normalized cross-correlation coefficient ZNCC is selected as the objective function to calculate the ZNCC values ​​of the matching point pairs and sort them by size:

[0141]

[0142] Where x is the coordinate information of image feature point f in the image, and x′ is the coordinate information of image feature point f′ in the image; I(x) and I(x′) represent the pixel brightness at x coordinate and x′ coordinate; and This represents the average pixel brightness of the image window centered at x and the image window W centered at x′;

[0143] Feature points greater than a threshold μ1 are selected as seed points for neighborhood evolution, and feature points greater than a threshold μ2 are selected as reserve matching points (μ1>μ2). For all matching points in the reference image, a one-to-many match is established in the center of the candidate image with a fixed window size. For points in the reference image, points in other images are matched, and a mixed match is established for all points within the window. Under the premise of satisfying the disparity gradient constraint and confidence constraint, the ZNCC of the evolving matching point pairs is calculated, and evolution points greater than a threshold μ3 are selected as seed points for secondary evolution, and evolution points greater than a threshold μ4 are selected as reserve matching points (μ3>μ4).

[0144] Assuming u′ and u are a pair of image matching points, and x′ and x are another adjacent pair of image matching points, the disparity gradient constraint formula is:

[0145] ||(u′-u)-(x′-x)|| ∞ ≤ε

[0146] Where ε is the threshold of the disparity gradient; the disparity gradient constraint reduces the ambiguity of image matching.

[0147] The formula for confidence constraints is:

[0148] s(x)=max{|I(x+Δ)-I(x)|,Δ∈{(1,0),(-1,0),(0,1),(0,-1)}}

[0149] Using confidence constraints can improve the reliability of matching evolution and obtain evolutionary matching point pairs.

[0150] In this embodiment of the invention, the calibration process of the image acquisition instrument involves calculating the internal parameters of the image acquisition instrument based on its imaging principle; selecting two input images as references based on the feature points and matching of the image sequence; calculating the fundamental matrix F of the reference image point pair, where F satisfies the equation x′Fx=0, and x′ and x are a pair of image matching points; estimating the initial values ​​K′ and K of the intrinsic parameter matrix of the reference image pair; calculating the essential matrix of the image point pair and extracting rotation and translation components; and using triangulation to find the three-dimensional model points corresponding to the feature points, given the internal and external parameters of the image acquisition instrument and the feature matching point pairs.

[0151] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0152] The information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of the present invention. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0153] Those skilled in the art will readily understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the functions described above can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this invention. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments.

[0154] Based on the technical solutions described in the above embodiments of the present invention, the following application examples can be further proposed.

[0155] According to embodiments of this application, the present invention also provides a computer device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the steps in any of the above-described method embodiments.

[0156] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps described in the various method embodiments above.

[0157] This invention also provides an information data processing terminal, which, when executed on an electronic device, provides a user input interface to implement the steps described in the above method embodiments. The information data processing terminal is not limited to mobile phones, computers, or switches.

[0158] This invention also provides a server that, when executed on an electronic device, provides a user input interface to implement the steps described in the above method embodiments.

[0159] This invention also provides a computer program product that, when run on an electronic device, enables the electronic device to implement the steps described in the various method embodiments above.

[0160] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0161] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention and within the spirit and principles of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A safety monitoring and early warning system for bedrock coastlines of tourist islands, characterized in that, This system combines UAV oblique photography technology to create a realistic 3D model, enabling visualized monitoring, early warning, and prediction of safety hazards. Specifically, it includes: The data acquisition module (1) is used for monitoring fracture activity, spheroidal weathering body activity, surface displacement, stress and strain, rainfall, and ocean wave dynamics. The data processing module (2) is used to receive the collected data information, store and calculate the data, compare the obtained spatial coordinate position, stress value, deformation value and rainfall value with the initial value to obtain the change amount 1; compare with the previous recorded value to obtain the change amount 2, and display the two types of change amounts in the form of charts; The danger alarm module (3) sets different types of observation value change thresholds according to different monitoring entities. When the change of the monitored value exceeds the threshold, a danger alarm is triggered, and abnormal values ​​are fed back and archived. The safety hazard prediction module (4) divides the areas where alarms frequently occur into key monitoring areas, combines historical monitoring data to predict the future trend of the area, and provides prompts for safety hazard removal and protection and repair; if the safety hazard is removed, a new prediction is made. A method for monitoring and warning the safety of bedrock coastlines of tourist islands, operating within the aforementioned safety monitoring and warning system, comprising: S1 collects monitoring information on fracture activity, spheroidal weathering body activity, surface displacement, stress and strain, rainfall, and ocean wave dynamics. S2, the collected data is transmitted to the data processing module (2) through optical fiber, and the data is stored and calculated. The obtained spatial coordinate position, stress value, and rainfall observation index values ​​are compared with the initial values ​​and compared with the previous recorded value to obtain the change and display it in the form of a chart. S3: Set different thresholds for changes in observed values ​​based on different monitored entities. When the change in monitored values ​​exceeds the threshold, a danger alarm is triggered, and abnormal values ​​are reported and archived. S4: Areas with frequent alarms are designated as key monitoring areas. Based on historical monitoring data, future changes in these areas are predicted, and prompts are issued for the removal and protection of safety hazards. If the safety hazards are removed, a new prediction is made. In step S2, the obtained spatial coordinates, stress values, and rainfall values ​​are compared with the initial values ​​and with the previously recorded values. Specifically, this includes: Step 1: Enter the current observation value and the initial values ​​of the current observed indicators. ; to obtain the change relative to the initial value and the change relative to the previous recorded value ; Step 2: Calculate the change and With a set threshold for a certain observation element Conduct comparative testing; Step 3: Based on the specified current stage Number of decision trees ,by As the fitting threshold, each decision tree is constructed sequentially. Fit the change value of the change function of the current model; Step 4: The model fitting is complete in the current stage; calculate the fitting change value in the current stage. , , Let X be the predicted value for the training data X at the current stage; the expression is: ； In the formula, For learning rate, For the i-th decision tree in the current stage In the training set The predicted values ​​are updated, and the values ​​of the data to be fitted are updated. ; These are the parameters for the random forest. Step 5: Fit the change value to the change amount X of the previous recorded value. Perform the Jarque-Bera test; Step Six: Given the test set The evaluation function is specified If the function of change no longer decreases within the specified number of stages, then model training should be stopped promptly to obtain the optimal number of iteration stages for the model. Complete model training and proceed to step seven; otherwise, if the test set... The function continues to decrease, so proceed to step four and continue the stage iteration; Step 7: Predict the test set based on the optimal number of iterations for the model. The test set prediction results are obtained. .

2. The safety monitoring and early warning system for bedrock coastlines of tourist islands according to claim 1, characterized in that, In the data acquisition module (1), the fracture activity monitoring includes: monitoring fracture displacement, width expansion and contraction variables, and newly formed fractures in bedrock shorelines and tourist routes; The surface displacement monitoring includes: installing monitoring equipment on slopes, steep exposed rock surfaces, and weathered spheroidal weathered bodies along tourist routes that are prone to collapse, and monitoring displacement changes in conjunction with a positioning system; and using distributed optical fibers for full-coverage monitoring of tourist roads built along the coast. The stress and strain monitoring includes: monitoring the stress state and strain of bridges, viewing platforms and tourist facilities in the fracture distribution area built on tourist roads; The rainfall monitoring includes: installing monitoring equipment in bare slope areas and vegetated areas along tourist routes to obtain information on rainfall changes; The wave dynamics monitoring includes installing monitoring equipment in areas of bedrock shoreline developed for tourism routes that are eroded by waves and have dense fractures, to obtain dynamic change data on wave impact force.

3. The safety monitoring and early warning system for bedrock coastlines of tourist islands according to claim 1, characterized in that, The data acquisition stations and monitoring stations in the data acquisition module (1) are displayed on the real-world 3D model according to their actual locations.

4. The method for safety monitoring and early warning system of bedrock coastline of tourist islands according to claim 1, characterized in that, In step two, the original threshold y of the change X of the previous recorded value is examined. The specific implementation steps are as follows: (1) Calculate the data distribution skewness according to the threshold of change of the previous recorded value using the following formula. : ； in, The threshold value for the data to be fitted. The mean of the sample threshold. For sample size, The standard deviation of the sample threshold; (2) Calculate the kurtosis of the data distribution according to the threshold of the change in the previous recorded value using the following formula. : ； (3) Construct the chi-square statistic according to the formula shown below. : ； in, For sample size, For data distribution skewness, Kurtosis of the data distribution; Test chi-square statistic The significance of the statistic is examined by considering the chi-square statistic. The degree of bias determines whether data changes should be made; if the chi-square statistic... If the threshold is exceeded, proceed to step (4); otherwise proceed to step (5). (4) If the data exceeds the set threshold, determine whether the upper limit of the preset number of data changes has been reached. Let the upper limit of the number of changes be 1. ,if Jump to (5), otherwise update The value is , to process the data The change is carried out by the following formula. Value traversal, selecting the data threshold after transformation. Likelihood function The largest Generate change expressions: ； in, To obtain a new data threshold, we use the threshold value of the data to be fitted as the first-stage fitting threshold. Set the current stage judgment variable ; (5) If the data bias is not obvious or the number of data threshold changes reaches the upper limit of the target, take the original data threshold. As the first-stage fitting threshold Set the current stage judgment variable .

5. The safety monitoring and early warning system for bedrock coastlines of tourist islands according to claim 1, characterized in that, In step seven, the optimal number of iteration stages for the model is determined. The test set prediction results are obtained. Specifically, it includes: (1) Calculation up to the current stage Model predictions on test set T The expression is: ； In the formula, For the test set prediction results, For learning rate, For the stage Decision Tree The prediction results For the stage The inverse transform of the threshold pretransform is expressed as follows: ； In the formula, For the current stage The inverse transform function of the model, For the current stage Determine the variable. for inverse transform of the transform, for Inverse transform of a transform; stage Prediction results , The expression is as follows: ； The expression is as follows: ； (2) Calculate the value of the change function in the current stage. , For the original threshold of test set T, The change function used by the system is the squared change function for regression problems. .

6. The safety monitoring and early warning system for bedrock coastlines of tourist islands according to claim 1, characterized in that, In step S4, areas where alarms frequently occur are designated as key monitoring areas. Based on historical monitoring data, future changes in these areas are predicted, including: The image sequence of frequently alarmed areas taken from different angles is used as the input set. Feature matching point pairs of the images are obtained through feature extraction and matching, and these pairs are then subjected to evolutionary processing. Based on the geological evolution and security threat model, feature points of candidate images are selected as seed points to be matched and filtered in their surrounding neighborhoods to obtain evolution matching point pairs; The image acquisition instrument is calibrated, and its intrinsic and extrinsic parameters are obtained by combining the matching point pairs; the three-dimensional model points are then recovered based on the image acquisition instrument parameters and the matching point pairs. A geological evolution and security threat model is used for reconstruction. Seed model points are selected to generate initial points, which then evolve within their grid neighborhood. By filtering out errors based on constraints, an accurate evolutionary 3D point cloud model is obtained.

7. The safety monitoring and early warning system for bedrock coastlines of tourist islands according to claim 6, characterized in that, The geological evolution and security threat model specifically includes: For each feature point of the reference image Based on the epipolar constraint, find the corresponding candidate matching point in the candidate image. Using a geological evolution and security threat model, the zero-mean normalized cross-correlation coefficient (ZNCC) is selected as the objective function. The ZNCC values ​​of matched point pairs are calculated and sorted by their magnitude. ； in, Image feature points The corresponding coordinate information in the image. Image feature points The corresponding coordinate information in the image; and express coordinates and Pixel brightness at coordinates; and Indicated by The image window centered and with The average pixel brightness of the image window W centered on the image; Feature points greater than a threshold μ1 are selected as seed points for neighborhood evolution, and feature points greater than a threshold μ2 are selected as reserve matching points (μ1>μ2). For all matching points in the reference image, a one-to-many match is established in the center of the candidate image with a fixed window size. For points in the reference image, points in other images are matched, and a mixed match is established for all points within the window. Under the premise of satisfying the disparity gradient constraint and confidence constraint, the ZNCC of the evolving matching point pairs is calculated, and evolution points greater than a threshold μ3 are selected as seed points for secondary evolution, and evolution points greater than a threshold μ4 are selected as reserve matching points (μ3>μ4). Assumption and It is a pair of image matching points. and For adjacent image matching point pairs, the disparity gradient constraint formula is: ； in, It is the threshold of the disparity gradient; the disparity gradient constraint reduces the ambiguity of image matching; The formula for confidence constraints is: ； Using confidence constraints can improve the reliability of matching evolution and obtain evolutionary matching point pairs.

8. The safety monitoring and early warning system for bedrock coastlines of tourist islands according to claim 6, characterized in that, The calibration process for an image acquisition instrument involves calculating its internal parameters based on its imaging principle; selecting two input images as references based on feature points and matching from the image sequence; and calculating the fundamental matrix F of the reference image point pairs, where F satisfies the equation... , and It is a pair of image matching points; the initial value of the intrinsic parameter matrix for estimating the reference image pair. and The essential matrix of the image point pairs is calculated and the rotation and translation components are extracted; the intrinsic and extrinsic parameters of the instrument and the feature matching point pairs are obtained from the known image, and the three-dimensional model points corresponding to the feature points are obtained by triangulation.

Images