A highway slope comprehensive monitoring and early warning system based on multi-source information fusion

Abstract

The invention belongs to the technical field of highway slope monitoring, and discloses a highway slope comprehensive monitoring and early warning system with multi-source information fusion, comprising: collecting slope geometry data and soil data; obtaining soil properties based on a constructed soil viscosity prediction model; dividing the slope into blocks to obtain slope blocks, performing stability calculation on the slope blocks to obtain stability coefficients, constructing a minimum stability coefficient objective function based on the stability coefficients and determining a minimum stability coefficient partitioning scheme, performing block aggregation on the minimum stability coefficient partitioning scheme through a clustering algorithm to obtain edge block clustering clusters; performing safety analysis on the edge block clustering clusters to obtain safe areas and dangerous areas; constructing a stability prediction model based on the safe areas, and performing future safety analysis on the safe areas through the prediction results of the stability prediction model; and greatly improving the accuracy and timeliness of slope monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of highway slope monitoring, and more specifically to a highway slope comprehensive monitoring and early warning system with multi-source information fusion. Background Art

[0002] The patent application with the publication number CN118397563A discloses a monitoring method and a monitoring system for slopes, and the monitoring method comprises the following steps: setting a plurality of targets on the slope; establishing a coordinate system on the target surface of the target to obtain the world coordinates of the target; obtaining an image containing each target surface and establishing a pixel coordinate system on the image; obtaining the displacement of the target in the world coordinates by the change of the target in the pixel coordinates; controlling the triggering of the slope warning by the displacement of the target; solving the problems of insufficient accuracy and high cost of existing slope monitoring; realizing simultaneous monitoring of multiple points on the slope, flexibly arranging the positions of the monitoring points according to actual conditions, and realizing large-scale monitoring of the slope at a relatively low cost; the acquired monitoring data can be directly associated with the displacement of the slope, and when responding to an ongoing warning, the real state of the slope can also be directly observed based on the collected images.

[0003] Although the current highway slope monitoring system has achieved a certain degree of automated monitoring, it is still highly dependent on human intervention in data analysis, early warning and emergency response. Most systems can only provide raw data or simple alarm signals, and specific risk assessment and decision-making still require manual analysis, which not only increases the burden on staff, but may also lead to delayed early warnings or false alarms due to human misjudgment. The geological environment of highway slopes is usually dynamically changing, especially under severe climatic conditions, the stability of the slope may deteriorate rapidly. However, the current slope monitoring systems mostly use fixed thresholds or timed data collection methods, which cannot achieve real-time dynamic monitoring and rapid response.

[0004] In view of this, the present invention proposes a highway slope comprehensive monitoring and early warning system with multi-source information fusion to solve the above problems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solutions: a highway slope comprehensive monitoring and early warning system with multi-source information fusion, comprising: a slope data acquisition module: a preset sampling time, and collecting slope geometry data and soil data at each sampling time;

[0006] Soil prediction module: obtain soil properties based on the constructed soil viscosity prediction model;

[0007] Slope division module: divide the slope into blocks to obtain slope blocks, calculate the stability of the slope blocks based on soil properties, slope geometry data and soil data to obtain the stability coefficient, build the minimum stability coefficient objective function based on the stability coefficient, minimize the minimum stability coefficient objective function through the optimization algorithm to obtain the minimum stability coefficient division scheme, and perform block aggregation on the minimum stability coefficient division scheme through the clustering algorithm to obtain C edge block clustering clusters;

[0008] Analysis and alarm module: performs safety analysis on edge block clusters to obtain safe areas and dangerous areas;

[0009] Block warning module: Build a stability prediction model based on the safety area, and conduct future safety analysis on the safety area through the prediction results of the stability prediction model.

[0010] Furthermore, the slope geometry data includes: slope inclination, slope length and slope width; the soil data includes: soil density and soil moisture.

[0011] Furthermore, the soil viscosity prediction model is constructed in the following manner:

[0012] Collect E group of historical soil viscosity data, historical soil viscosity data: historical soil moisture and historical soil properties, historical soil properties include soil cohesion and soil internal friction angle, based on historical soil viscosity data, take historical soil moisture as the input of soil viscosity prediction model, take predicted soil properties as the output of soil viscosity prediction model, take historical soil properties as the prediction target of soil viscosity prediction model for model training, construct soil viscosity prediction model, soil viscosity prediction model is linear regression model, the target loss function of soil viscosity prediction model is: ;in, is the mean square error, is the total amount of historical soil viscosity data, No. The historical soil properties It is The predicted soil properties are obtained by minimizing the value of the target loss function and obtaining a soil viscosity prediction model with the minimum value of the target loss function.

[0013] Furthermore, the method of performing stability calculation includes:

[0014] The slope is divided into blocks based on the slope geometry data. The block division method includes: dividing the slope into A slope blocks, and the lengths of all slope blocks satisfy: ;in, Representative The length of the slope block, Represents the index of the slope block, represents the slope length, Represents the total amount of slope blocks; the stability of the slope blocks is calculated according to the length of the slope blocks. The formula for stability calculation is: in, Representative The stability factor of the slope block is represents the soil cohesion, represents the soil density, represents the slope inclination, represents the acceleration due to gravity, represents the internal friction angle of soil, Represents the slope width.

[0015] Furthermore, the minimum stability coefficient objective function .

[0016] Furthermore, the method of obtaining the minimum stability coefficient partitioning scheme includes:

[0017] Step 1: preset K groups of slope division schemes, each group of slope division schemes represents a different combination of slope lengths, a group of slope division schemes is represented as a chromosome, and the genes in the chromosome correspond to the length combination of slope blocks in the slope division scheme;

[0018] Step 2: For each chromosome, calculate the fitness of each chromosome according to the fitness calculation formula. The fitness calculation formula is: ;in, Represents fitness; the parent chromosome is selected using the tournament selection algorithm based on the fitness value;

[0019] Step 3: According to the combination of slope block lengths in the parent chromosome, take the length of the first slope block as the starting value, and add up the sums one by one to obtain the local length. When the local lengths in the parent chromosome are consistent, record the gene as an equal-length node gene. All equal-length node genes constitute an equal-length node set. Determine the crossover gene segment by the permutation and combination method according to the equal-length node set. All crossover gene segments constitute a crossover set. Traverse the crossover gene segments in the crossover set and try to exchange genes corresponding to the crossover gene segments in the parent chromosome to obtain a first-generation daughter chromosome. Calculate the fitness of the first-generation daughter chromosome according to the fitness calculation formula. When there is a chromosome in the first-generation daughter chromosome whose fitness value is greater than the fitness value of the parent chromosome, exchange genes corresponding to the crossover gene segment in the parent chromosome to obtain a first-generation daughter chromosome, and repeat the gene exchange with the first-generation daughter chromosome as the parent chromosome until the crossover gene segments in the crossover set are completely traversed, and the first-generation daughter chromosome obtained in the last traversal is used as the final daughter chromosome;

[0020] Step 4: Select the sub-chromosome with the largest fitness value as the target chromosome, and randomly select a gene of the sub-chromosome to replace with the target chromosome through the random mutation method to generate a mutant chromosome;

[0021] Step 5: Repeat steps 1 to 4 until the fitness value of the mutated chromosome no longer changes. The output mutated chromosome is the partitioning scheme with the minimum stability coefficient.

[0022] Furthermore, the method of performing block aggregation includes:

[0023] Randomly select data from C minimum stability coefficient sets as initial cluster centers, where C represents the optimal clustering value, and perform data aggregation on the minimum stability coefficient set according to the initial cluster centers. The data aggregation method includes: taking each initial cluster center as a cluster, taking the stability coefficients other than the initial cluster center as detection nodes, traversing the detection nodes, and calculating the distance from the stability coefficient in the detection node to the stability coefficient in each initial cluster center in turn by the Euclidean distance formula, obtaining a detection node distance list, filtering out the cluster corresponding to the minimum distance value according to the detection node distance list, and merging the detection node into the cluster corresponding to the minimum distance value, until all detection nodes are traversed to form C edge block clusters;

[0024] The data in each cluster is sorted from large to small according to the stability coefficient, and the median data in each cluster is selected as the new initial cluster center. When the median data is not unique, the data with a smaller stability coefficient is selected as the new initial cluster center. The data aggregation operation is performed according to the new initial cluster center, and repeated until the value of the new initial cluster center no longer changes, thereby obtaining C edge block clusters.

[0025] Furthermore, the method for obtaining the optimal clustering value includes:

[0026] Based on the minimum stability coefficient division scheme, each slope block is uniquely numbered to obtain the slope code, and the stability of the coded slope block is calculated to obtain the stability coefficient of each slope block. All stability coefficients and slope codes corresponding to the stability coefficients constitute the minimum stability coefficient set; define the grouping set , is a positive integer and satisfies the condition and ,Right now Divide by The remainder is 0, where is the total number of slope blocks in the minimum stability coefficient partitioning scheme; traverse the grouping set in turn and assign values ​​to the grouping values , sort the stability coefficients in the stability coefficient set in ascending order from small to large, and divide the sorted stability coefficient set into The group entropy of each stability coefficient group is calculated according to the coefficient entropy formula. The coefficient entropy formula is: ;in, The representative group value is The group entropy of Represents the grouping index, Representative The probability of a group, Representative The mean of the stability coefficients of the groups; record the group entropy corresponding to each group value, calculate the entropy difference of each group value according to the group entropy, the entropy difference is the difference between the group entropy of each group value and the group entropy of the previous group value, and take the group value with the largest entropy difference as the optimal clustering value.

[0027] Furthermore, the method of performing security analysis includes:

[0028] A stability coefficient threshold is preset, and the mean of the stability coefficient of the data in each edge block cluster is calculated to obtain the average stability coefficient. When the average stability coefficient is less than or equal to the stability coefficient threshold, the slope block corresponding to the data in the cluster is divided into a dangerous area and an alarm is triggered. When the average stability coefficient is greater than the stability coefficient threshold, the slope block corresponding to the data in the cluster is divided into a safe area.

[0029] Furthermore, the stability prediction model is constructed in the following manner:

[0030] Through the recorded The slope geometry data and soil data of the sampling time are used to calculate the safe area The historical average stability coefficient;

[0031] Depend on The historical average stability coefficient of the safety area within the sampling time and the sampling time are used to obtain the historical average stability coefficient sequence and time series ;

[0032] The historical average stability coefficient sequence and time series are used as the training sample set, the LSTM model is used as the initial model, the LSTM model is trained using the training sample set, the historical average stability coefficient sequence samples and time series samples are used as the input data for training the stability prediction model, and the predicted average stability coefficient is used as the output data for training the stability prediction model; minimizing the error between the actual historical average stability coefficient and the predicted historical average stability coefficient is used as the training goal, the recall rate function is used as the loss function of the stability prediction model, and when the loss function converges, the training is stopped to obtain the stability prediction model.

[0033] The technical effects and advantages of the present invention are as follows:

[0034] The present invention comprehensively collects slope geometry data and soil data, and combines it with a soil viscosity prediction model. The system can provide accurate stability analysis results. This multi-source information fusion method improves the reliability of the assessment. The preset sampling time allows the state of the slope to be updated in real time, reflecting the changes in the slope in a timely manner, ensuring that early warnings are provided before potential dangers occur. Advanced machine learning and cluster analysis techniques are used to achieve intelligent analysis of the safety status of the slope. Dynamic monitoring of the stability coefficient and deep mining of historical data can automatically identify dangerous areas and issue alarms. The optimization algorithm and cluster analysis are used to divide the slope into blocks, so that the stability of each slope block can be more accurately identified, facilitating the concentration of efforts for key monitoring and maintenance. The stability prediction mechanism based on historical data and the LSTM model can predict future slope stability changes, provide early warnings, and enhance the foresight of management measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of a highway slope comprehensive monitoring and early warning system with multi-source information fusion according to the present invention;

[0036] Figure 2 It is a schematic diagram of a comprehensive monitoring and early warning method for highway slopes based on multi-source information fusion according to the present invention. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] Example 1

[0039] See also Figure 1 As shown, the highway slope comprehensive monitoring and early warning system with multi-source information fusion described in this embodiment includes: a slope data acquisition module: a sampling time is preset, and slope geometry data and soil data are collected at each sampling time;

[0040] Soil prediction module: obtain soil properties based on the constructed soil viscosity prediction model;

[0041] Slope division module: divide the slope into blocks to obtain slope blocks, calculate the stability of the slope blocks based on soil properties, slope geometry data and soil data to obtain the stability coefficient, build the minimum stability coefficient objective function based on the stability coefficient, minimize the minimum stability coefficient objective function through the optimization algorithm to obtain the minimum stability coefficient division scheme, and perform block aggregation on the minimum stability coefficient division scheme through the clustering algorithm to obtain C edge block clustering clusters;

[0042] Analysis and alarm module: performs safety analysis on edge block clusters to obtain safe areas and dangerous areas;

[0043] Block early warning module: build a stability prediction model based on the safety area, and conduct future safety analysis on the safety area through the prediction results of the stability prediction model;

[0044] Each module is connected by wired and / or wireless means to achieve data transmission between modules;

[0045] Slope geometry data include: slope inclination, slope length and slope width; soil data include: soil density and soil moisture;

[0046] The slope inclination is the angle between the slope and the horizontal ground, which is collected by an inclinometer, and the soil moisture is collected by a soil moisture detector;

[0047] Collect E group of historical soil viscosity data, historical soil viscosity data: historical soil moisture and historical soil properties, historical soil properties include soil cohesion and soil internal friction angle, based on historical soil viscosity data, take historical soil moisture as the input of soil viscosity prediction model, take predicted soil properties as the output of soil viscosity prediction model, take historical soil properties as the prediction target of soil viscosity prediction model for model training, construct soil viscosity prediction model, soil viscosity prediction model is linear regression model, the target loss function of soil viscosity prediction model is: ;in, is the mean square error, is the total amount of historical soil viscosity data, No. The historical soil properties It is The predicted soil properties; taking the minimization of the target loss function as the training goal, the soil viscosity prediction model with the minimum target loss function value is obtained;

[0048] Based on the established soil viscosity prediction model, soil moisture is used as the input of the soil viscosity prediction model to predict soil properties.

[0049] The soil cohesion is the bonding force or adhesion between soil particles. It reflects the ability of soil to resist shear failure under effective stress, especially the intrinsic adhesion characteristics between particles, and does not depend on external stress. The internal friction angle of soil is the friction between soil particles. Its size determines the stability of soil under shear. The methods for obtaining historical soil properties include:

[0050] Collect soil samples of the same humidity and size, and place the prepared soil samples in the sample room of the triaxial testing machine to ensure that the soil samples are in the center. Apply confining pressure to the soil samples through the triaxial testing machine, which is usually divided into several stages, such as 20kPa, 40kPa, etc., record the confining pressure, initial height and diameter of the soil samples, calculate the initial volume and stress state of the soil samples, and apply axial loads at a constant speed on the triaxial testing machine, gradually increasing to the failure point (that is, the soil sample is in a critical state between the whole and the failure). During this process, continuously monitor the stress and displacement of the soil sample, record the data, and when the soil sample fails, record the axial stress and confining pressure at this time as stress data, conduct D tests to obtain reliable data, and according to the Mohr-Coulomb failure criterion, use the stress data of the failure point to draw the Mohr circle, and calculate the soil cohesion and soil internal friction angle;

[0051] The slope is divided into blocks based on the slope geometry data. The block division method includes: dividing the slope into A slope blocks, and the lengths of all slope blocks satisfy: ;in, Representative The length of the slope block, Represents the index of the slope block, represents the slope length, Represents the total amount of slope blocks; the stability of the slope blocks is calculated according to the length of the slope blocks. The formula for stability calculation is: in, Representative The stability factor of the slope block is represents the soil cohesion, represents the soil density, represents the slope inclination, represents the acceleration due to gravity, represents the internal friction angle of soil, Represents the slope width; constructs the minimum stability coefficient objective function based on the stability coefficient of each slope block ;

[0052] By comprehensively collecting slope geometry data and soil data and combining them with soil viscosity prediction models, the system can provide accurate stability analysis results. This multi-source information fusion method improves the reliability of the assessment.

[0053] The slope blocks are strips that are approximately trapezoidal. The slope is regarded as a right triangle in the plane. The slope block uses a right-angled side of the right triangle as a dividing side. The dividing side is divided into different lengths. Each divided figure is a trapezoid, that is, a slope block.

[0054] Based on the minimum stability coefficient objective function, a minimum stability coefficient partitioning scheme is obtained. The methods for obtaining the minimum stability coefficient partitioning scheme include:

[0055] Step 1: preset K groups of slope division schemes, each group of slope division schemes represents a different combination of slope lengths, a group of slope division schemes is represented as a chromosome, and the genes in the chromosome correspond to the length combination of slope blocks in the slope division scheme;

[0056] Step 2: For each chromosome, calculate the fitness of each chromosome according to the fitness calculation formula. The fitness calculation formula is: ;in, Represents fitness; the larger the fitness value, the better the partitioning scheme. The parent chromosome is selected using the tournament selection algorithm based on the fitness value;

[0057] Step 3: According to the combination of slope block lengths in the parent chromosome, take the length of the first slope block as the starting value, and add up the sums one by one to obtain the local length. When the local lengths in the parent chromosome are consistent, record the gene as an equal-length node gene. All equal-length node genes constitute an equal-length node set. Determine the crossover gene segment by the permutation and combination method according to the equal-length node set. All crossover gene segments constitute a crossover set. Traverse the crossover gene segments in the crossover set and try to exchange genes corresponding to the crossover gene segments in the parent chromosome to obtain a first-generation daughter chromosome. Calculate the fitness of the first-generation daughter chromosome according to the fitness calculation formula. When there is a chromosome in the first-generation daughter chromosome whose fitness value is greater than the fitness value of the parent chromosome, exchange genes corresponding to the crossover gene segment in the parent chromosome to obtain a first-generation daughter chromosome, and repeat the gene exchange with the first-generation daughter chromosome as the parent chromosome until the crossover gene segments in the crossover set are completely traversed, and the first-generation daughter chromosome obtained in the last traversal is used as the final daughter chromosome;

[0058] Step 4: Select the sub-chromosome with the largest fitness value as the target chromosome, and use the random mutation method to randomly select a gene of the sub-chromosome to replace it with the target chromosome to generate a mutant chromosome;

[0059] Step 5: Repeat steps 1 to 4 until the fitness value of the mutated chromosome no longer changes. The output mutated chromosome is the partitioning scheme with the minimum stability coefficient.

[0060] Based on the minimum stability coefficient division scheme, each slope block is uniquely numbered to obtain the slope code (a unique value that can determine the slope position), and the stability of the coded slope blocks is calculated to obtain the stability coefficient of each slope block. All stability coefficients and slope codes corresponding to the stability coefficients constitute the minimum stability coefficient set; define the grouping set , is a positive integer and satisfies the condition and ,Right now Divide by The remainder is 0, where is the total number of slope blocks in the minimum stability coefficient partitioning scheme; traverse the grouping set in turn and assign values ​​to the grouping values , sort the stability coefficients in the stability coefficient set in ascending order from small to large, and divide the sorted stability coefficient set into The group entropy of each stability coefficient group is calculated according to the coefficient entropy formula. The coefficient entropy formula is: ;in, The representative group value is The group entropy of Represents the grouping index, Representative The probability of a group, Representative The mean of the stability coefficients of the groups; record the group entropy corresponding to each group value, calculate the entropy difference of each group value according to the group entropy, the entropy difference is the difference between the group entropy of each group value and the group entropy of the previous group value, and take the group value with the largest entropy difference as the optimal clustering value;

[0061] Based on the optimal clustering value, the minimum stability coefficient set is similarly grouped. The similar grouping methods include:

[0062] Based on the best clustering value, randomly select the data in C minimum stability coefficient sets as the initial clustering centers, and perform data aggregation on the minimum stability coefficient sets according to the initial clustering centers. The data aggregation method includes: taking each initial clustering center as a cluster, taking the stability coefficients other than the initial clustering center as detection nodes, traversing the detection nodes, and calculating the distance from the stability coefficient in the detection node to the stability coefficient in each initial clustering center in turn by the Euclidean distance formula, obtaining the detection node distance list, filtering out the cluster corresponding to the minimum distance value according to the detection node distance list, and merging the detection node into the cluster corresponding to the minimum distance value, until all detection nodes are traversed, forming C edge block clusters;

[0063] The data in each cluster are sorted from large to small according to the stability coefficient, and the median data in each cluster is selected as the new initial cluster center. When the median data is not unique, the data with a smaller stability coefficient is selected as the new initial cluster center. The data aggregation operation is performed according to the new initial cluster center, and the operation is repeated until the value of the new initial cluster center no longer changes, and C edge block clusters are obtained; the slope is divided into blocks through optimization algorithm and cluster analysis, so that the stability of each slope block can be more accurately identified, which is convenient for concentrating efforts on key monitoring and maintenance;

[0064] A stability coefficient threshold is preset, and the mean of the stability coefficient of the data in each edge block cluster is calculated to obtain the average stability coefficient. When the average stability coefficient is less than or equal to the stability coefficient threshold, the slope block corresponding to the data in the cluster is divided into a dangerous area and an alarm is triggered. When the average stability coefficient is greater than the stability coefficient threshold, the slope block corresponding to the data in the cluster is divided into a safe area.

[0065] A stability prediction model is constructed based on the safety area to predict the change of the average stability coefficient of the safety area in the future time period and to provide future warning. The methods of constructing the stability prediction model include:

[0066] Through the recorded The slope geometry data and soil data of the sampling time are used to calculate the safe area The historical average stability coefficient;

[0067] Depend on The historical average stability coefficient of the safety area within the sampling time and the sampling time are used to obtain the historical average stability coefficient sequence and time series ;

[0068] The historical average stability coefficient sequence and time series are used as the training sample set, the LSTM model is used as the initial model, the LSTM model is trained using the training sample set, the historical average stability coefficient sequence samples and time series samples are used as the input data for training the stability prediction model, and the predicted average stability coefficient is used as the output data for training the stability prediction model; the training goal is to minimize the error between the actual historical average stability coefficient and the predicted historical average stability coefficient, the recall rate function is used as the loss function of the stability prediction model, and when the loss function reaches convergence, the training is stopped to obtain the stability prediction model;

[0069] The stability prediction mechanism built based on historical data and LSTM model can predict future slope stability changes, provide early warning, and enhance the foresight of management measures;

[0070] Through the trained stability prediction model, the time series of the future time period is used as the input of the stability prediction model to predict the average stability coefficient of the safety area in the future time period, and the future warning of the safety area is carried out according to the average stability coefficient of the safety area in the future time period;

[0071] In this embodiment, by comprehensively collecting slope geometry data and soil data and combining it with a soil viscosity prediction model, the system can provide accurate stability analysis results. This multi-source information fusion method improves the reliability of the assessment; the preset sampling time allows the state of the slope to be updated in real time, promptly reflecting the changes in the slope and ensuring that early warnings are provided before potential dangers occur; advanced machine learning and cluster analysis techniques are used to achieve intelligent analysis of the safety status of the slope; dynamic monitoring of the stability coefficient and deep mining of historical data can automatically identify dangerous areas and issue alarms; the optimization algorithm and cluster analysis are used to divide the slope into blocks, which can more accurately identify the stability of each slope block and facilitate concentrated monitoring and maintenance; the stability prediction mechanism based on historical data and the LSTM model can predict future slope stability changes, provide early warnings, and enhance the foresight of management measures.

[0072] Example 2

[0073] See also Figure 2 As shown, the part not described in detail in this embodiment is described in Example 1, and a comprehensive monitoring and early warning method for highway slopes by multi-source information fusion is provided, including:

[0074] S1, preset sampling time, collect slope geometry data and soil data at each sampling time;

[0075] S2, obtaining soil properties based on the constructed soil viscosity prediction model;

[0076] S3, dividing the slope into blocks to obtain slope blocks, performing stability calculation on the slope blocks based on soil properties, slope geometry data and soil data to obtain stability coefficients, constructing a minimum stability coefficient objective function based on the stability coefficients, minimizing the minimum stability coefficient objective function through an optimization algorithm to obtain a minimum stability coefficient partitioning scheme, and performing block aggregation on the minimum stability coefficient partitioning scheme through a clustering algorithm to obtain C edge block clustering clusters;

[0077] S4, perform safety analysis on the edge block clusters to obtain safe areas and dangerous areas;

[0078] S5. Construct a stability prediction model based on the safety area, and conduct future safety analysis on the safety area through the prediction results of the stability prediction model.

[0079] Example 3

[0080] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the above-mentioned highway slope comprehensive monitoring and early warning method with multi-source information fusion is implemented.

[0081] Since the electronic device introduced in this embodiment is an electronic device used to implement a comprehensive monitoring and early warning method for highway slopes with multi-source information fusion in the embodiment of this application, based on the comprehensive monitoring and early warning method for highway slopes with multi-source information fusion introduced in the embodiment of this application, the technical personnel of this field can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application is not introduced in detail here. As long as the technical personnel of this field implement the electronic device used in the comprehensive monitoring and early warning method for highway slopes with multi-source information fusion in the embodiment of this application, it belongs to the scope of protection of this application.

[0082] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0083] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A highway slope comprehensive monitoring and early warning system with multi-source information fusion, characterized in that: include: Slope data acquisition module: preset sampling time, and collect slope geometry data and soil data at each sampling time; Soil prediction module: obtain soil properties based on the constructed soil viscosity prediction model; Slope division module: divide the slope into blocks to obtain slope blocks, calculate the stability of the slope blocks based on soil properties, slope geometry data and soil data to obtain the stability coefficient, build the minimum stability coefficient objective function based on the stability coefficient, minimize the minimum stability coefficient objective function through the optimization algorithm to obtain the minimum stability coefficient division scheme, and perform block aggregation on the minimum stability coefficient division scheme through the clustering algorithm to obtain C edge block clustering clusters; Analysis and alarm module: performs safety analysis on edge block clusters to obtain safe areas and dangerous areas; Block early warning module: build a stability prediction model based on the safety area, and conduct future safety analysis on the safety area through the prediction results of the stability prediction model; The method of performing stability calculation includes: The slope is divided into blocks based on the slope geometry data. The block division method includes: dividing the slope into A slope blocks, and the lengths of all slope blocks satisfy: in, Representative The length of the slope block, Represents the index of the slope block, represents the slope length, Represents the total amount of slope blocks; the stability of the slope blocks is calculated according to the length of the slope blocks. The formula for stability calculation is: in, Representative The stability factor of the slope block is represents the soil cohesion, represents the soil density, represents the slope inclination, represents the acceleration due to gravity, represents the internal friction angle of soil, Represents the slope width.

2. The highway slope comprehensive monitoring and early warning system with multi-source information fusion according to claim 1 is characterized in that: The slope geometry data include: slope inclination, slope length and slope width; the soil data include: soil density and soil moisture.

3. The highway slope comprehensive monitoring and early warning system with multi-source information fusion according to claim 2 is characterized in that: The soil viscosity prediction model is constructed in the following manner: Collect E group of historical soil viscosity data, historical soil viscosity data: historical soil moisture and historical soil properties, historical soil properties include soil cohesion and soil internal friction angle, based on historical soil viscosity data, take historical soil moisture as the input of soil viscosity prediction model, take predicted soil properties as the output of soil viscosity prediction model, take historical soil properties as the prediction target of soil viscosity prediction model for model training, construct soil viscosity prediction model, soil viscosity prediction model is linear regression model, the target loss function of soil viscosity prediction model is: ;in, is the mean square error, is the total amount of historical soil viscosity data, No. The historical soil properties It is The predicted soil properties are obtained by minimizing the value of the target loss function and obtaining a soil viscosity prediction model with the minimum value of the target loss function.

4. The highway slope comprehensive monitoring and early warning system with multi-source information fusion according to claim 3 is characterized in that: The minimum stability coefficient objective function .

5. The highway slope comprehensive monitoring and early warning system with multi-source information fusion according to claim 4 is characterized in that: The method of obtaining the minimum partitioning scheme of the stability coefficient includes: Step 1: preset K groups of slope division schemes, each group of slope division schemes represents a different combination of slope lengths, a group of slope division schemes is represented as a chromosome, and the genes in the chromosome correspond to the length combination of slope blocks in the slope division scheme; Step 2: For each chromosome, calculate the fitness of each chromosome according to the fitness calculation formula. The fitness calculation formula is: ;in, Represents fitness; the parent chromosome is selected using the tournament selection algorithm based on the fitness value; Step 3: According to the combination of slope block lengths in the parent chromosome, take the length of the first slope block as the starting value, and add up the sums one by one to obtain the local length. When the local lengths in the parent chromosome are consistent, record the gene as an equal-length node gene. All equal-length node genes constitute an equal-length node set. Determine the crossover gene segment by the permutation and combination method according to the equal-length node set. All crossover gene segments constitute a crossover set. Traverse the crossover gene segments in the crossover set and try to exchange genes corresponding to the crossover gene segments in the parent chromosome to obtain a first-generation daughter chromosome. Calculate the fitness of the first-generation daughter chromosome according to the fitness calculation formula. When there is a chromosome in the first-generation daughter chromosome whose fitness value is greater than the fitness value of the parent chromosome, exchange genes corresponding to the crossover gene segment in the parent chromosome to obtain a first-generation daughter chromosome, and repeat the gene exchange with the first-generation daughter chromosome as the parent chromosome until the crossover gene segments in the crossover set are completely traversed, and the first-generation daughter chromosome obtained in the last traversal is used as the final daughter chromosome; Step 4: Select the sub-chromosome with the largest fitness value as the target chromosome, and randomly select a gene of the sub-chromosome to replace with the target chromosome through the random mutation method to generate a mutant chromosome; Step 5: Repeat steps 1 to 4 until the fitness value of the mutated chromosome no longer changes. The output mutated chromosome is the partitioning scheme with the minimum stability coefficient.

6. A highway slope comprehensive monitoring and early warning system with multi-source information fusion according to claim 5, characterized in that: The method of performing block aggregation includes: Randomly select data from C minimum stability coefficient sets as initial cluster centers, where C represents the optimal clustering value, and perform data aggregation on the minimum stability coefficient set according to the initial cluster centers. The data aggregation method includes: taking each initial cluster center as a cluster, taking the stability coefficients other than the initial cluster center as detection nodes, traversing the detection nodes, and calculating the distance from the stability coefficient in the detection node to the stability coefficient in each initial cluster center in turn by the Euclidean distance formula, obtaining a detection node distance list, filtering out the cluster corresponding to the minimum distance value according to the detection node distance list, and merging the detection node into the cluster corresponding to the minimum distance value, until all detection nodes are traversed to form C edge block clusters; The data in each cluster is sorted from large to small according to the stability coefficient, and the median data in each cluster is selected as the new initial cluster center. When the median data is not unique, the data with a smaller stability coefficient is selected as the new initial cluster center. The data aggregation operation is performed according to the new initial cluster center, and repeated until the value of the new initial cluster center no longer changes, thereby obtaining C edge block clusters.

7. A highway slope comprehensive monitoring and early warning system with multi-source information fusion according to claim 6, characterized in that: The method for obtaining the optimal clustering value includes: Based on the minimum stability coefficient division scheme, each slope block is uniquely numbered to obtain the slope code, and the stability of the coded slope block is calculated to obtain the stability coefficient of each slope block. All stability coefficients and slope codes corresponding to the stability coefficients constitute the minimum stability coefficient set; define the grouping set , is a positive integer and satisfies the condition and ,Right now Divide by The remainder is 0, where is the total number of slope blocks in the minimum stability coefficient partitioning scheme; traverse the grouping set in turn and assign values ​​to the grouping values , sort the stability coefficients in the stability coefficient set in ascending order from small to large, and divide the sorted stability coefficient set into The group entropy of each stability coefficient group is calculated according to the coefficient entropy formula. The coefficient entropy formula is: ;in, The representative group value is The group entropy of Represents the grouping index, Representative The probability of a group, Representative The mean of the stability coefficients of the groups; record the group entropy corresponding to each group value, calculate the entropy difference of each group value according to the group entropy, the entropy difference is the difference between the group entropy of each group value and the group entropy of the previous group value, and take the group value with the largest entropy difference as the optimal clustering value.

8. The highway slope comprehensive monitoring and early warning system with multi-source information fusion according to claim 7 is characterized in that: The methods for conducting safety analysis include: A stability coefficient threshold is preset, and the mean of the stability coefficient of the data in each edge block cluster is calculated to obtain the average stability coefficient. When the average stability coefficient is less than or equal to the stability coefficient threshold, the slope block corresponding to the data in the cluster is divided into a dangerous area and an alarm is triggered. When the average stability coefficient is greater than the stability coefficient threshold, the slope block corresponding to the data in the cluster is divided into a safe area.

9. The highway slope comprehensive monitoring and early warning system with multi-source information fusion according to claim 8 is characterized in that: The construction method of the stability prediction model includes: Through the recorded The slope geometry data and soil data of the sampling time are used to calculate the safe area The historical average stability coefficient; Depend on The historical average stability coefficient of the safety area within the sampling time and the sampling time are used to obtain the historical average stability coefficient sequence and time series ; The historical average stability coefficient sequence and time series are used as the training sample set, the LSTM model is used as the initial model, the LSTM model is trained using the training sample set, the historical average stability coefficient sequence samples and time series samples are used as the input data for training the stability prediction model, and the predicted average stability coefficient is used as the output data for training the stability prediction model; minimizing the error between the actual historical average stability coefficient and the predicted historical average stability coefficient is used as the training goal, the recall rate function is used as the loss function of the stability prediction model, and when the loss function converges, the training is stopped to obtain the stability prediction model.

Images