Road slope stability prediction method and device, storage medium and electronic equipment

Abstract

The invention provides a road slope stability prediction method and device, a storage medium and electronic equipment, and the method comprises the steps: collecting the multi-dimensional environmental parameters of a road slope in real time, and forming an original parameter set; based on a Granger causal test algorithm, analyzing the original parameter set, determining a causal influence relationship among the parameters, constructing a parameter causal network, and screening out key parameters; calculating the similarity of each key parameter in time and space dimensions through a dynamic time warping algorithm, and generating a space-time incidence matrix; based on the space-time incidence matrix, performing weighted fusion on the key parameters to obtain fusion parameters; inputting the fusion parameters into a pre-trained slope stability prediction model to obtain a predicted value of the slope stability; and when the predicted value exceeds a dynamically updated early warning threshold value, outputting slope instability early warning information. According to the method, the defects of inaccurate prediction and data redundancy of a current road slope stability prediction method are overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a method, device, storage medium and electronic device for predicting the stability of road slopes. Background Art

[0002] The stability of road slopes is directly related to road traffic safety and the safety of personnel and property in surrounding areas. Timely and accurately predicting the slope stability is of great significance for preventing geological disasters such as landslides and collapses.

[0003] The existing methods for predicting the stability of road slopes mainly have the following deficiencies: In terms of data processing, the original monitoring data is mostly simply used or only conventional filtering and normalization processing are carried out, and the potential relationships between multi-dimensional environmental parameters cannot be fully explored, making it difficult to accurately reflect the actual impact of parameters on slope stability. At the same time, the key factors that truly affect slope stability cannot be effectively identified, resulting in data redundancy and waste of computing resources.

[0004] In addition, the threshold setting of existing prediction models usually adopts fixed values or simple adjustment methods based on experience, and cannot adapt to the dynamic changes of slope stability under different geological conditions and climatic environments, and it is easy to have false alarms or missed alarms.

[0005] Therefore, the current methods for predicting the stability of road slopes have the defects of inaccurate prediction and data redundancy. Summary of the Invention

[0006] The main object of the present invention is to provide a method, device, storage medium and electronic device for predicting the stability of road slopes, aiming to overcome the defects of inaccurate prediction and data redundancy in the current methods for predicting the stability of road slopes.

[0007] To achieve the above object, the present invention provides a method for predicting the stability of road slopes, including the following steps: Real-time collect multi-dimensional environmental parameters of the road slope to form an original parameter set; Based on the Granger causality test algorithm, analyze the original parameter set, determine the causal influence relationships between parameters, construct a parameter causal network, and screen out the key parameters affecting slope stability; Calculate the similarity of each key parameter in the time and space dimensions through the dynamic time warping algorithm to generate a spatio-temporal correlation matrix; Based on the spatio-temporal correlation matrix, perform weighted fusion on the key parameters to obtain a fusion parameter; Input the fusion parameter into a pre-trained slope stability prediction model to obtain a predicted value of slope stability; Dynamically update the warning threshold of the slope stability prediction model, and when the predicted value exceeds the dynamically updated warning threshold, output slope instability warning information.

[0008] Further, the environmental parameters include geological parameters, mechanical parameters, environmental parameters, and displacement parameters. Further, when performing weighted fusion on the key parameters, weights are assigned based on the degree of association on the spatio-temporal correlation matrix, and fusion parameters are generated through matrix dot product operations. Further, based on the Granger causality test algorithm, analyze the original parameter set, determine the causal influence relationships among the parameters, construct a parameter causal network, and screen out the key parameters affecting slope stability, including: Calculate the grey correlation degree between each parameter in the original parameter set and the historical data of slope stability, preliminarily sort the parameters according to the grey correlation degree, and preferentially perform Granger causality tests on the parameter pairs with high correlation degrees; When performing the Granger causality test, based on the change point detection algorithm, identify the structural change points of the parameters in the time series, and set the lag order of the Granger causality test for different mutation intervals respectively; Based on the causal relationship strength evaluation model, calculate the causal relationship strength scores between each parameter pair; retain the parameter pairs with causal relationship strength scores higher than the threshold, and construct a parameter causal network; Use the Louvain algorithm to perform community division on the parameter causal network, calculate the cohesion coefficient of each community and the contribution degree to slope stability; select the core parameters in the communities with contribution degrees higher than the set value, as well as the hub parameters connecting across communities, as the key parameters affecting slope stability.

[0009] Further, based on the Granger causality test algorithm, analyze the original parameter set, determine the causal influence relationships among the parameters, construct a parameter causal network, and screen out the key parameters affecting slope stability, including: Perform three-dimensional convolution on the original parameter set through a three-dimensional convolutional kernel to obtain a spatio-temporal parameter set with enhanced features; Adopt a dynamic time window division method, adaptively adjust the time window length according to the parameter fluctuation frequency, segment the spatio-temporal parameter set, and obtain multiple parameter subsequences; Perform Granger causality tests on each of the parameter subsequences respectively, and use the attention mechanism to calculate the weights of the test results of each parameter subsequence, and perform weighted fusion to obtain a comprehensive causal relationship matrix; Based on a recurrent neural network, learn the evolution law of the comprehensive causal relationship matrix over time, and predict the change trend of the causal influence relationships among the parameters in the future period; Based on the comprehensive causal relationship matrix and the changing trend, construct a parameter causal network with a time dimension, calculate the temporal importance of each node in the parameter causal network, and select the parameters with temporal importance higher than the preset value as the key parameters affecting slope stability.

[0010] Further, through the dynamic time warping algorithm, calculate the similarity of each of the key parameters in the time and space dimensions to generate a spatio-temporal correlation matrix, including: Use a convolutional neural network to extract the local features of the key parameters in the spatial dimension and splice them into spatial sequence features; Use a long short-term memory network to capture the long-term dependence features of the key parameters in the time dimension and splice them into time series features; Map the time series features to quantum states, and through the dynamic time warping algorithm, combine quantum parallel computing of multiple time warping paths, evaluate the similarity of each time warping path, and select the optimal matching path to obtain a time dimension similarity matrix; Based on a pre-constructed spatial similarity evaluation model, analyze the spatial sequence features to generate a spatial dimension similarity matrix; Through tensor product operation, fuse the time dimension similarity matrix and the spatial dimension similarity matrix to obtain the spatio-temporal correlation matrix.

[0011] Further, dynamically update the warning threshold of the slope stability prediction model, including: Based on historical prediction data, use the sliding window algorithm to dynamically update the warning threshold of the slope stability prediction model.

[0012] The present invention also provides a device for predicting the stability of a road slope, including: An acquisition unit for real-time acquisition of multi-dimensional environmental parameters of the road slope to form an original parameter set; A screening unit for analyzing the original parameter set based on the Granger causality test algorithm, determining the causal influence relationship between parameters, constructing a parameter causal network, and screening out the key parameters affecting slope stability; An association unit for calculating the similarity of each of the key parameters in the time and space dimensions through the dynamic time warping algorithm to generate a spatio-temporal correlation matrix; A fusion unit for weighted fusion of the key parameters based on the spatio-temporal correlation matrix to obtain a fusion parameter; A prediction unit for inputting the fusion parameter into a pre-trained slope stability prediction model to obtain a predicted value of slope stability; An early warning unit for dynamically updating the early warning threshold of the slope stability prediction model, and outputting slope instability early warning information when the predicted value exceeds the dynamically updated early warning threshold.

[0013] The present invention also provides an electronic device, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps of the method described in any one of the above are implemented.

[0014] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0015] The road slope stability prediction method, device, storage medium and electronic device provided by the present invention include: collecting multi-dimensional environmental parameters of the road slope in real time to form an original parameter set; analyzing the original parameter set based on the Granger causality test algorithm to determine the causal influence relationship between parameters, constructing a parameter causal network, and screening out key parameters affecting slope stability; calculating the similarity of each key parameter in the time and space dimensions through the dynamic time warping algorithm to generate a spatio-temporal correlation matrix; performing weighted fusion on the key parameters based on the spatio-temporal correlation matrix to obtain a fusion parameter; inputting the fusion parameter into a pre-trained slope stability prediction model to obtain a predicted value of slope stability; dynamically updating the early warning threshold of the slope stability prediction model, and outputting slope instability early warning information when the predicted value exceeds the dynamically updated early warning threshold. In the present invention, by analyzing the original parameter set, determining the causal influence relationship between parameters, constructing a parameter causal network, screening out key parameters affecting slope stability, calculating the similarity of each key parameter in the time and space dimensions, generating a spatio-temporal correlation matrix, and dynamically updating the early warning threshold of the slope stability prediction model at the same time; fully excavating the potential relationship between multi-dimensional environmental parameters, overcoming the defects of inaccurate prediction and data redundancy in the current road slope stability prediction method. Brief Description of the Drawings

[0016] Figure 1 is a schematic diagram of the steps of the road slope stability prediction method in an embodiment of the present invention; Figure 2 is a block diagram of the structure of the road slope stability prediction device in an embodiment of the present invention; Figure 3 is a schematic block diagram of the structure of an electronic device in an embodiment of the present invention.

[0017] The implementation, functional features and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. Detailed Embodiments

[0018] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0019] Referring to Figure 1 , in one embodiment of the present invention, a method for predicting the stability of a road slope is provided, including the following steps: Step S1, collecting multi-dimensional environmental parameters of the road slope in real time to form an original parameter set; Step S2, based on the Granger causality test algorithm, analyzing the original parameter set, determining the causal influence relationship between each parameter, constructing a parameter causal network, and screening out the key parameters affecting the slope stability; Step S3, calculating the similarity of each key parameter in the time and space dimensions through the dynamic time warping algorithm to generate a spatio-temporal correlation matrix; Step S4, based on the spatio-temporal correlation matrix, performing weighted fusion on the key parameters to obtain a fusion parameter; Step S5, inputting the fusion parameter into a pre-trained slope stability prediction model to obtain a predicted value of the slope stability; Step S6, dynamically updating the warning threshold of the slope stability prediction model, and when the predicted value exceeds the dynamically updated warning threshold, outputting a slope instability warning message.

[0020] In this embodiment, as described in step S1 above, by deploying various sensors on the road slope, such as displacement sensors, humidity sensors, pressure sensors, meteorological sensors, etc., multi-dimensional environmental parameters are collected in real time. The above environmental parameters cover geomechanical parameters (such as rock and soil density, shear strength), hydrological parameters (groundwater level, rainfall), meteorological parameters (wind speed, temperature), slope shape parameters (crack width, slope angle), etc. Multi-dimensional data collection can comprehensively reflect the environmental state and physical characteristics of the slope itself, providing a rich and accurate data source for subsequent analysis. The collected data directly constitutes the original parameter set. This original parameter set contains a large amount of original and unprocessed information, retaining the authenticity and integrity of the data, but there are also problems such as data redundancy and complex relationships between parameters, which need to be further processed in subsequent steps.

[0021] As described in step S2 above, based on the Granger causality test algorithm, the original parameter set is analyzed, the causal influence relationship between each parameter is determined, a parameter causal network is constructed, and the key parameters affecting the stability of the slope are screened out. The core of this step is to explore the causal relationship between parameters and screen the key parameters. The Granger causality test algorithm is a statistical method based on time series. Its principle is to determine whether there is a causal relationship between the two by testing whether the historical information of a variable can improve the prediction accuracy of another variable. Granger causality test is performed on each parameter combination in the original parameter set to calculate the degree of causal influence between the parameters. According to the test results, the parameters with causal relationships are connected with edges to form a parameter causal network. In this network, nodes represent various environmental parameters, edges represent causal relationships between parameters, and the weights of edges can reflect the strength of causal influence. By analyzing the connection of network structure and parameters, those parameters that are in key positions in the network and have a significant impact on slope stability are screened out. These key parameters play a major role in the change of slope stability. The subsequent analysis is based on these parameters, which can effectively reduce the amount of data processing and improve analysis efficiency and prediction accuracy.

[0022] As described in step S3 above, the similarity of each of the key parameters in the time and space dimensions is calculated by the dynamic time warping algorithm to generate a spatiotemporal association matrix. This step aims to explore the intrinsic connection of the key parameters in the time and space dimensions. The dynamic time warping algorithm was originally used to measure the similarity of time series data. It aligns two time series of different lengths or speeds by bending the time axis to calculate the similarity between them. This solution expands its application and considers both time and space dimensions. In the time dimension, the monitoring data sequence of the key parameters is arranged in chronological order, and the dynamic time warping algorithm is used to calculate the similarity of the time series of different parameters; in the spatial dimension, based on the spatial position relationship of the slope monitoring points, the spatial distribution characteristics of the parameters are included in the calculation, such as adjusting the similarity measurement by factors such as spatial distance and topological structure. The similarity calculation results of the time dimension and the spatial dimension are integrated to generate a matrix, namely the spatiotemporal association matrix. Each element in the matrix represents the similarity of two key parameters in the time and space dimensions. The matrix fully reflects the correlation characteristics of the key parameters in time and space, providing a basis for subsequent parameter fusion.

[0023] As described in step S4 above, based on the spatio-temporal correlation matrix, the key parameters are weighted and fused to obtain the fused parameters. This step realizes the fusion of key parameters based on the spatio-temporal correlation matrix. The spatio-temporal correlation matrix reflects the mutual relationship of each key parameter in the spatio-temporal dimension. According to the similarity degree between the parameters in the matrix, corresponding weights are assigned to each key parameter. Parameters with high similarity and close correlation are given higher weights, indicating that they play an important role in reflecting the slope stability; on the contrary, the weights are lower. By means of weighted summation, etc., each key parameter is fused to obtain the fused parameters. The fused parameters integrate the spatio-temporal information of the key parameters. Compared with the original parameters, they can more comprehensively and fully reflect the state of the slope, reduce information redundancy, enhance the representativeness and effectiveness of the data, and provide better input data for the subsequent prediction model.

[0024] As described in step S5 above, the fused parameters are input into a pre-trained slope stability prediction model to obtain the predicted value of slope stability. This step uses the pre-trained prediction model to process the fused parameters to obtain the prediction result. The pre-trained slope stability prediction model can be a machine learning model (such as a neural network, a support vector machine, etc.) or a deep learning model. In the model training stage, a large number of sample data containing historical environmental parameters and corresponding slope stability states are used, and the model parameters are continuously adjusted through an optimization algorithm so that the model can learn the mapping relationship between the environmental parameters and the slope stability. The fused parameters are input into the trained model, and the model analyzes and calculates the input data according to the rules it has learned, and outputs a quantified value, that is, the predicted value of slope stability. This predicted value reflects the stability degree of the slope based on the current environmental parameters, and provides an important reference for slope stability assessment.

[0025] As described in step S6 above, the warning threshold of the slope stability prediction model is dynamically updated. When the predicted value exceeds the dynamically updated warning threshold, a slope instability warning message is output. This step realizes the dynamic optimization and warning function of the prediction model. The environmental conditions of the slope are complex and changeable, and a fixed warning threshold is difficult to meet the requirements of slope stability assessment under different working conditions. Therefore, according to factors such as real-time monitoring data, historical slope instability situations, and geological environment changes, a dynamic algorithm is used to update the warning threshold of the prediction model. For example, time series analysis methods can be used to predict the future environmental change trend, and combined with the critical values of slope instability under different environmental conditions in historical data, the warning threshold is dynamically adjusted. When the predicted value exceeds the dynamically updated warning threshold, it indicates that the slope is in a state with a high risk of instability, and the system immediately outputs a slope instability warning message to remind relevant personnel to take corresponding preventive and treatment measures, so as to realize the real-time and effective monitoring and warning of the slope stability of the road and ensure the traffic safety of the road and the safety of the surrounding areas.

[0026] In this embodiment, by analyzing the original parameter set, determining the causal influence relationship between the parameters, constructing a parameter causal network, screening out key parameters that affect slope stability, calculating the similarity of each key parameter in time and space dimensions, generating a spatiotemporal correlation matrix, and dynamically updating the warning threshold of the slope stability prediction model; fully exploring the potential relationship between multi-dimensional environmental parameters, and overcoming the defects of inaccurate prediction and data redundancy in the current road slope stability prediction method.

[0027] In one embodiment, the environmental parameters include geological parameters, mechanical parameters, environmental parameters and displacement parameters.

[0028] In this embodiment, geological parameters are key data that reflect the geological structure and composition characteristics of the road slope. It covers information such as rock and soil type, rock thickness, geological structure (such as the location and characteristics of faults and folds). The rock and soil type determines the basic physical properties of the slope. For example, there are significant differences in the shear strength and permeability of different rocks and soils such as clay, sand, and rock, which directly affect the stability of the slope. The thickness of the rock layer and the geological structure reveal the geological history and potential weak links of the slope. The integrity of the rock mass near the fault is poor and it is easy to become a sliding surface. The shape and direction of the fold will also change the stress distribution. Accurately obtaining these geological parameters can provide a basic geological background for the subsequent analysis of the stability of the slope and help determine the innate stability conditions of the slope.

[0029] Mechanical parameters are used to describe the mechanical behavior and bearing capacity of the rock and soil of the slope. They mainly include the density, elastic modulus, Poisson's ratio, cohesion, internal friction angle, etc. The density of the rock and soil affects its own gravity, which in turn affects the magnitude of the sliding force; the elastic modulus and Poisson's ratio reflect the deformation characteristics of the material when subjected to stress; cohesion and internal friction angle are key indicators for measuring the shear strength of rock and soil, and determine the ability of the slope to resist shear failure. By monitoring and analyzing these mechanical parameters, a mechanical model of the slope can be established to simulate its stress state under different loads and environmental conditions, and to predict whether the slope will deform or become unstable, which is an important basis for quantitative stability analysis.

[0030] Environmental parameters focus on external environmental factors, mainly including meteorological conditions (such as rainfall, rainfall intensity, temperature, wind speed, sunshine duration, etc.) and hydrological conditions (groundwater level, surface runoff, pore water pressure, etc.). Rainfall and rainfall intensity directly affect the water content of rock and soil masses, saturate and soften the rock and soil masses, reduce the shear strength, increase the sliding force, and are common inducements for slope instability. Temperature changes can cause the rock and soil masses to expand and contract thermally, and cracks may be generated under long-term action, damaging the slope structure. The rise of the groundwater level will increase the pore water pressure, reduce the effective stress, and lower the slope stability; the scouring effect of surface runoff will erode the slope surface and change the slope shape. The real-time changes of these environmental parameters will continuously change the stress and physical state of the slope. Therefore, continuous monitoring and inclusion in the analysis can more accurately reflect the stability dynamics of the slope in the actual environment.

[0031] Displacement parameters are the most intuitive manifestation of the change in slope stability. The displacement situation on the surface or inside of the slope is monitored in real time through displacement sensors, including data such as horizontal displacement, vertical displacement, and displacement rate. Tiny displacement changes may be precursors of slope instability, and a continuous trend of displacement growth directly indicates that the slope stability is decreasing. By monitoring and analyzing the displacement parameters, the deformation signs of the slope can be captured in a timely manner, and the reasons and development trends of the deformation can be judged in combination with other parameters. For example, when the displacement rate suddenly increases, accompanied by the rise of the groundwater level and an increase in rainfall, the slope can be quickly evaluated to be in a high-risk state, providing key information for timely warning. The dynamic changes of displacement parameters are an important basis for judging slope stability and play an intuitive and crucial guiding role in the entire prediction process.

[0032] In one embodiment, when performing weighted fusion on the key parameters, weights are assigned based on the degree of association on the spatio-temporal correlation matrix, and the fusion parameters are generated through matrix dot multiplication operations. In this embodiment, the value of each element in the spatio-temporal correlation matrix represents the similarity or tightness of association between the corresponding two key parameters in the time and space dimensions. The higher the degree of association, the stronger the synergistic effect of these two parameters in reflecting the slope stability state, and the more significant the impact on the change of slope stability. Therefore, higher weights should be assigned; conversely, parameters with a low degree of association have a smaller synergistic impact on slope stability, and the weights are correspondingly reduced. This method of weight assignment based on the degree of association can fully explore the potential connections between parameters and make the weight setting more in line with the actual physical mechanism and data characteristics of the slope.

[0033] In this embodiment, first, the spatio-temporal correlation matrix is processed row by row or column by column (usually processed column by column, with each column corresponding to a key parameter). Normalization operations are performed on the elements of each column to ensure that the sum of weights is 1, thereby obtaining the weight vector corresponding to each key parameter. This weight vector reflects the influence of the correlation degree between other parameters and this key parameter in the spatio-temporal dimension on its importance. Then, the matrix composed of each key parameter (each key parameter is a column of the matrix, arranged in time series or spatial position) is subjected to matrix dot product operation with the above weight vector. During the matrix dot product operation, each element of the weight vector is multiplied by the elements of the corresponding column of the key parameter matrix one by one, realizing the weighting of the data of each key parameter at different times or spatial positions. Finally, the weighted column data are added together to obtain the final fusion parameter. This process organically integrates the information of each key parameter in the spatio-temporal dimension according to their correlation tightness through mathematical operations.

[0034] The above weighted fusion method based on the spatio-temporal correlation matrix can effectively highlight the role of parameters that have a greater impact on slope stability and are closely related to each other, while reducing the interference of parameters with low correlation and small influence. The generated fusion parameter not only contains the original information of each key parameter, but also incorporates the spatio-temporal correlation characteristics between parameters. Compared with simple parameter superposition or averaging processing, it can more comprehensively and accurately reflect the actual state of the slope. The fusion parameter, as the input of the subsequent slope stability prediction model, can significantly improve the sensitivity and prediction accuracy of the model to the changes in slope stability, make the prediction results more in line with the actual situation, and provide a more reliable basis for slope instability warning.

[0035] In one embodiment, based on the Granger causality test algorithm, the original parameter set is analyzed to determine the causal influence relationship between parameters, construct a parameter causal network, and screen out the key parameters that affect slope stability, including: Calculate the grey correlation degree between each parameter in the original parameter set and the historical data of slope stability, and perform a preliminary ranking of each parameter according to the size of the grey correlation degree. Priority is given to performing the Granger causality test on parameter pairs with high correlation degrees; When performing the Granger causality test, based on the change point detection algorithm, identify the structural change points of the parameter in the time series, and set the lag order of the Granger causality test for different mutation intervals respectively; Based on the causal relationship strength evaluation model, calculate the causal relationship strength scores between each parameter pair; retain the parameter pairs with causal relationship strength scores higher than the threshold, and construct a parameter causal network; Use the Louvain algorithm to perform community division on the parameter causal network, calculate the cohesion coefficient of each community and its contribution degree to slope stability; select the core parameters in the community with a contribution degree higher than the set value, and the hub parameters connecting across communities as the key parameters that affect slope stability.

[0036] In this embodiment, first, through grey relational analysis, the parameters in the original parameter set are preliminarily screened to narrow the scope of Granger causality test and improve the analysis efficiency. Grey relational analysis is a multi-factor statistical analysis method. It judges the degree of association between factors by calculating the similarity of the geometric shapes of data sequence curves. In this embodiment, each parameter in the original parameter set is regarded as a factor, and the historical data of slope stability is used as the reference sequence to calculate the grey relational degree between each parameter sequence and the reference sequence. The larger the relational degree value is, the more similar the parameter is to the change trend of slope stability, and the greater its potential impact on slope stability. After sorting the parameters from largest to smallest according to the grey relational degree, the parameter pairs with high relational degree are preferentially selected for Granger causality test. This is because the parameters with high relational degree are more likely to have a causal relationship with slope stability. Prioritizing the test on them can focus on key factors, reduce unnecessary computational effort, and significantly improve the analysis efficiency while ensuring the accuracy of the analysis.

[0037] When conducting the Granger causality test, the Granger causality test is optimized according to the dynamic change characteristics of time series data. The change point detection algorithm can identify the points where the data characteristics in the time series change significantly, that is, the structural mutation points. In the study of slope stability, due to the influence of factors such as geological activities and meteorological conditions, there may be multiple mutation points in the time series of parameters, and the change laws and mutual relationships of data in different intervals may be different. After determining these mutation points based on the change point detection algorithm, the time series is divided into multiple mutation intervals. For each interval, appropriate Granger causality test lag orders are set respectively according to the fluctuation characteristics and change laws of the data. The lag order reflects the influence duration of the historical information of one variable on the current value of another variable. Different lag orders are required for the data in different mutation intervals to accurately capture the causal relationship. In this way, the Granger causality test can better adapt to the dynamic changes of the data, improve the accuracy of causal relationship judgment, and avoid misjudgment or missed judgment caused by a fixed lag order.

[0038] Furthermore, by constructing a causal relationship strength evaluation model to quantify the causal relationship strength between each pair of parameters, significant causal relationships are screened out to construct a parameter causal network. The causal relationship strength evaluation model comprehensively considers factors such as the statistics obtained from Granger causality tests (such as F-statistics, P-values), the fluctuation range of the parameters themselves, and the sample size. The causal relationship strength of each pair of parameters is scored through a specific calculation formula. The F-statistics and P-values reflect the significance level of the causal relationship, the fluctuation range reflects the degree of influence of parameter changes on other parameters, and the sample size affects the reliability of the results. The higher the score, the stronger the causal relationship between the parameter pairs. A reasonable threshold is set, and the parameter pairs with scores higher than the threshold are retained. These parameter pairs form the edges of the parameter causal network. With each parameter as a node and the retained parameter pairs as edges, a parameter causal network is constructed. This network intuitively shows the causal associations between the parameters. By removing the weaker causal relationships, the network becomes more concise and clear, highlighting the key causal connections.

[0039] Finally, the Louvain algorithm is used to deeply analyze the parameter causal network to screen out the key parameters that truly affect slope stability. The Louvain algorithm is an efficient graph community partitioning algorithm that can partition the parameter causal network into multiple closely connected communities, and the parameters within each community have strong internal correlations. Calculate the cohesion coefficient of each community, which measures the tightness of the connections between the nodes within the community; at the same time, combined with the historical data of slope stability and the network structure, evaluate the contribution of each community to slope stability. The higher the contribution, the greater the influence of the parameters within the community on slope stability. Select the core parameters from the communities with contributions higher than the set value. These core parameters play a key role within the community and have strong causal associations with other parameters; in addition, select the hub parameters that connect across communities. These parameters are the key nodes for information transmission between different communities and are crucial for the connectivity of the entire network and the influence on slope stability. In this way, the local role and global influence of the parameters in the network are comprehensively considered, and the key parameters screened out can more accurately reflect the core factors affecting slope stability.

[0040] In one embodiment, based on the Granger causality test algorithm, the original parameter set is analyzed to determine the causal influence relationships between the parameters, construct a parameter causal network, and screen out the key parameters that affect slope stability, including: Perform three-dimensional convolution on the original parameter set through a three-dimensional convolutional kernel to obtain a spatio-temporal parameter set with enhanced features; Adopt a dynamic time window partitioning method to adaptively adjust the time window length according to the parameter fluctuation frequency, segment the spatio-temporal parameter set, and obtain multiple parameter subsequences; Performing Granger causality test on each of the parameter subsequences respectively, and using attention mechanism to calculate the weight of the test results of each parameter subsequence, and weighted fusion to obtain a comprehensive causal relationship matrix; Based on the recursive neural network, the evolution law of the comprehensive causal relationship matrix over time is learned to predict the change trend of the causal influence relationship of each parameter in the future period; Based on the comprehensive causal relationship matrix and the changing trend, a parameter causal network with a time dimension is constructed, the temporal importance of each node in the parameter causal network is calculated, and parameters with a temporal importance higher than a preset value are selected as key parameters affecting slope stability.

[0041] In this embodiment, first, the original parameter set is feature extracted and enhanced using a three-dimensional convolution operation. The three-dimensional convolution kernel can process data in both time and space dimensions and parameter channel dimensions at the same time. Compared with the traditional two-dimensional convolution, it can better capture the local characteristics of the parameters in the time and space dimensions and the correlation characteristics between different parameters. When processing the original parameter set, the parameters are arranged in three dimensions according to the time series, spatial position and parameter type, and the three-dimensional convolution kernel slides on the data body, and the characteristic information of the parameters in the local area of ​​time and space is extracted through the convolution operation. For example, in the slope monitoring scene, the coordinated change characteristics between different types of parameters (such as displacement, humidity, and pressure parameters) in a certain time period and a certain local spatial area can be effectively extracted. After the three-dimensional convolution, the feature-enhanced spatiotemporal parameter set obtained not only retains the basic information of the original parameters, but also strengthens the spatiotemporal feature expression of the parameters, providing a more representative data basis for the subsequent causal relationship analysis, which helps to more accurately explore the causal relationship between parameters.

[0042] Furthermore, a dynamic time window division strategy is adopted to address the dynamic change characteristics of parameter time series. The fluctuation frequencies of different parameters in different time periods are different. For example, meteorological parameters such as rainfall have a high fluctuation frequency in the rainy season, while geological structure-related parameters have a low fluctuation frequency under normal circumstances. The traditional fixed time window division method is difficult to adapt to such changes, which will lead to insufficient data feature extraction or information redundancy. The dynamic time window division method adjusts the time window length adaptively in real time by analyzing the fluctuation frequency of the parameters. In specific implementation, the sliding window can be used to calculate the fluctuation statistics (such as standard deviation and variance) of the parameters within a certain time range. When the fluctuation statistics exceed the set threshold, the time window length is shortened to capture the rapidly changing characteristics; when the fluctuation is stable, the time window length is appropriately extended to obtain more comprehensive trend information. In this way, the feature-enhanced spatiotemporal parameter set is divided into multiple parameter subsequences, and the parameter data in each subsequence has similar fluctuation characteristics, so that when Granger causality test is performed on each subsequence in the future, the causal relationship between parameters under a specific fluctuation state can be analyzed more accurately, thereby improving the accuracy of causal relationship judgment.

[0043] Then, perform Granger causality tests on the parameter subsequences and introduce an attention mechanism to optimize the result fusion. Performing Granger causality tests independently on each parameter subsequence can analyze the causal influence relationships among parameters under specific fluctuation states. Compared with testing the entire time series, it can better capture the dynamic changes in the causal relationships among parameters. However, due to differences in data characteristics, fluctuation conditions, etc. among different parameter subsequences, the importance of their test results for judging the overall causal relationship also varies. The attention mechanism assigns corresponding weights to the test results of each subsequence by learning the characteristics of different parameter subsequences. For example, for the subsequence containing data on the precursors of slope instability, its test result is more crucial for judging the causal relationship of slope stability and will be assigned a higher weight; while for the subsequence containing normal fluctuation data, the weight is relatively low. According to the calculated weights, the Granger causality test results of each parameter subsequence are weighted and fused to integrate them into a comprehensive causal relationship matrix. This matrix comprehensively considers the causal relationships among parameters under different fluctuation states, more comprehensively reflects the causal association characteristics among parameters, and provides richer and more accurate information for subsequent analysis.

[0044] Furthermore, use a Recurrent Neural Network (RNN) to model and predict the temporal dynamic characteristics of the comprehensive causal relationship matrix. RNN has a memory function and can process data with time series characteristics. Through the recurrent connections in the hidden layer, it can learn the dependencies of data in the time dimension. Input the comprehensive causal relationship matrix into the RNN in chronological order, and the network learns the evolution law of the causal influence relationships among parameters over time by continuously updating the hidden layer state. For example, during the learning process, RNN can capture the changing trend that the causal relationship between the rainfall amount parameter and the slope displacement parameter gradually strengthens as rainfall continues. Based on the learned evolution law, RNN can predict the changing trends of the causal influence relationships among parameters in future time periods, obtaining a causal relationship prediction matrix at different future time points. This predictive ability enables the system to not only analyze the current causal relationships among parameters but also anticipate the changes in future causal relationships in advance, providing strong support for the prospective analysis and early warning of slope stability.

[0045] Finally, by constructing a parameter causal network and analyzing the importance of nodes, key parameters are screened out. Combining the comprehensive causal relationship matrix and the predicted trend of causal relationship changes, a parameter causal network with a time dimension is constructed with parameters as nodes and causal relationships as edges. This network not only shows the current causal relationships between parameters but also reflects the changes in causal relationships over time, being able to more realistically reflect the dynamic associations between parameters in the slope system. In the parameter causal network, the temporal importance of each node (i.e., parameter) is calculated. The temporal importance comprehensively considers factors such as the strength of causal relationships of parameters at different time points, the degree of influence on other parameters, and the key degree in the evolution of causal relationships. For example, during the slope instability process, those parameters with continuously increasing causal relationship strength and having an important impact on other key parameters will have a higher temporal importance. A preset value is set, and the parameters with temporal importance higher than this preset value are selected as the key parameters affecting slope stability. These key parameters play a dominant role in the process of slope stability change. By focusing on these parameters, slope stability can be analyzed and predicted more accurately, providing a core basis for slope monitoring and early warning.

[0046] In one embodiment, through the dynamic time warping algorithm, the similarity of each of the key parameters in the time and space dimensions is calculated to generate a spatio-temporal correlation matrix, including: A convolutional neural network is used to extract the local features of the key parameters in the space dimension and splice them into spatial sequence features; A long short-term memory network is used to capture the long-term dependence features of the key parameters in the time dimension and splice them into time sequence features; The time sequence features are mapped to quantum states. Through the dynamic time warping algorithm, multiple time warping paths are combined with quantum parallel computing to evaluate the similarity of each time warping path, and the optimal matching path is screened out to obtain a time dimension similarity matrix; Based on a pre-constructed spatial similarity evaluation model, the spatial sequence features are analyzed to generate a space dimension similarity matrix; Through tensor product operation, the time dimension similarity matrix and the space dimension similarity matrix are fused to obtain the spatio-temporal correlation matrix.

[0047] In this embodiment, first, in the prediction of road slope stability, key parameters are collected from different sensors, and their original data differ in aspects such as dimension, numerical range, and sampling frequency. For example, the unit of the displacement parameter may be millimeters, while the unit of the pressure parameter is Pascal, and the numerical magnitudes of the two are vastly different; at the same time, the sampling frequencies of different sensors may also be inconsistent. For spatio-temporal dimension standardization processing, in the time dimension, through methods such as linear transformation or interpolation, the time series of each key parameter is unified into the same sampling frequency and time reference to ensure the consistency of the time scale; in the space dimension, based on the geographic coordinate system, the parameter data of monitoring points at different positions are converted into a unified spatial coordinate system, and the spatial coordinate values are normalized to eliminate spatial scale differences. After standardization, a standardized spatio-temporal parameter set is formed, making each key parameter comparable in the spatio-temporal dimension and avoiding interference with the similarity calculation results due to dimension and scale issues.

[0048] Furthermore, by leveraging the powerful local feature extraction ability of the convolutional neural network (CNN), the feature information of key parameters in the space dimension is mined. The convolutional layer of the CNN performs convolutional operations by sliding the convolutional kernel over the data and can automatically extract the local features of the data. When processing the spatial data of key parameters, the spatial dimension data (such as the parameter values of different monitoring points) in the standardized spatio-temporal parameter set is organized into a two-dimensional or three-dimensional data structure as the input of the CNN. The convolutional kernel slides over this data structure to extract the local correlation features between the parameters of adjacent monitoring points, such as the co-variation features of parameters such as displacement and humidity in a certain area of the slope. After operations such as multiple layers of convolution and pooling, the data dimension is further compressed and the feature expression is strengthened. Finally, the extracted local features are concatenated in the order of spatial positions to form spatial sequence features. Such spatial sequence features not only retain the original information of the parameters of each monitoring point but also highlight the local correlation characteristics of the parameters in space, providing more representative feature data for subsequent calculation of spatial dimension similarity.

[0049] Furthermore, the long short-term memory network (LSTM) is used to solve the long-term dependence problem in time series data and obtain the dynamic change characteristics of key parameters in the time dimension. LSTM is a special recurrent neural network that can effectively process and remember information in long time series through a gating mechanism (input gate, forget gate, and output gate), avoiding the problem of gradient vanishing or explosion. The time series data in the standardized spatio-temporal parameter set is input into LSTM. When processing the data at each time step, LSTM selectively retains or forgets historical information according to the gating mechanism, thereby capturing the long-term dependence relationship of parameters in the time dimension. For example, when analyzing the change of slope displacement over time, LSTM can remember the change trend of the previous displacement and combine the current data to judge whether there are long-term trends such as acceleration or deceleration in the displacement change. As the time step progresses, LSTM outputs the hidden state corresponding to each time step, and these hidden states are concatenated in chronological order to form time series features. This time series feature contains the long-term evolution law and dynamic change information of key parameters in the time dimension, providing a key feature representation for subsequent time dimension similarity calculation.

[0050] Then, quantum computing is combined with the dynamic time warping algorithm to improve the efficiency and accuracy of time dimension similarity calculation. First, the time series features output by LSTM are mapped into quantum states, and the superposition and entanglement characteristics of quantum states are used to achieve parallel computing of multiple time warping paths. When the traditional dynamic time warping algorithm calculates the similarity of two time series, it needs to traverse all possible time alignment paths, and the computational complexity is relatively high. Based on quantum parallel computing, multiple time warping paths can be evaluated simultaneously, greatly improving the computational efficiency. During the calculation process, by designing appropriate quantum gate operations and measurement methods, the quantum states corresponding to each time warping path are calculated to obtain the similarity metric values of the two time series under each path. Then, according to these similarity metric values, the optimal matching path with the highest similarity is selected, and the similarity value corresponding to this path is used as the similarity score of the two time series in the time dimension. The above calculations are performed pairwise on the time series of all key parameters to obtain a time dimension similarity matrix, and each element in the matrix represents the similarity degree of the corresponding two key parameter time series, providing a quantitative basis for comprehensively analyzing the correlation relationship of parameters in the time dimension.

[0051] Furthermore, relying on a pre-constructed spatial similarity evaluation model, in-depth analysis is carried out on the spatial sequence characteristics of key parameters. The spatial similarity evaluation model can be constructed based on machine learning algorithms (such as support vector machines, random forests, etc.) or deep learning algorithms (such as graph neural networks), and its training data comes from slope historical monitoring data and related geological and geographical information. The input of the model is the spatial sequence characteristics extracted and spliced by CNN, and the output is the similarity scores of each key parameter in the spatial dimension. During the training process of the model, the mapping relationship between the parameter characteristics and similarity at different spatial positions is learned. For example, factors such as the geographical distance of monitoring points, the similarity of geological conditions, and the parameter value differences are considered to establish the rules or model parameters for spatial similarity evaluation. When new spatial sequence characteristics are input, the model calculates the spatial similarity scores between each key parameter according to the learned knowledge, organizes these scores into a matrix form according to the parameter correspondence relationship, and generates a spatial dimension similarity matrix. This matrix reflects the degree of association of key parameters in the spatial dimension and provides an important reference for analyzing the mutual relationship between parameters at different positions of the slope.

[0052] Finally, the organic fusion of time and space dimension similarity information is achieved through tensor product operation. Tensor product is a mathematical operation that can combine the information of two matrices to generate a new tensor (in the case of two-dimensional matrices, the fusion result is still a matrix). The time dimension similarity matrix describes the similarity relationship of key parameters in the time series, the spatial dimension similarity matrix describes the similarity relationship of parameters in the spatial position, and the spatio-temporal correlation matrix needs to reflect the comprehensive correlation characteristics of parameters in both spatio-temporal dimensions. Taking the time dimension similarity matrix and the spatial dimension similarity matrix as the operands of the tensor product operation, the tensor product operation is carried out. During the operation process, calculations are made according to the corresponding relationship of matrix elements, so that the elements of the fused matrix contain both time dimension similarity information and spatial dimension similarity information. The finally obtained spatio-temporal correlation matrix comprehensively depicts the mutual correlation relationship of each key parameter in the time and space dimensions, provides core data support for subsequent parameter weighted fusion and slope stability prediction based on spatio-temporal correlation information, and helps to more accurately grasp the spatio-temporal law of slope stability change.

[0053] In one embodiment, dynamically updating the warning threshold of the slope stability prediction model includes: Based on historical prediction data, the warning threshold of the slope stability prediction model is dynamically updated using the sliding window algorithm.

[0054] In this embodiment, by introducing the sliding window algorithm, the dynamic adjustment of the warning threshold of the slope stability prediction model is realized based on historical prediction data to adapt to the complex changes of the slope environment. The core of the sliding window algorithm lies in dividing the historical prediction data in a certain time order to form a set of continuous and overlapping window data. In this solution, the window size can be set according to actual needs. For example, a window containing the most recent N prediction data is selected. As time goes by, the window slides backward with a fixed step length. Each time it slides, the warning threshold is updated based on the data within the current window, enabling the model to continuously adapt to new data features and changes in the slope state.

[0055] During specific operations, first collect and organize the historical prediction data. The above data includes the predicted values of the slope stability in the past and the corresponding actual slope state information (such as whether instability occurs and the degree of instability, etc.). Taking the sliding window as a unit, analyze the data within the window and extract key statistical features, such as the mean, standard deviation, maximum value, minimum value, etc. of the predicted values. These statistical features reflect the distribution and fluctuation of the slope stability predicted values during this time period. For example, if the standard deviation of the predicted values within the window is large, it indicates that the slope stability fluctuates violently during this period, and the original warning threshold may no longer be applicable.

[0056] Based on the statistical features of the window data, combined with the actual working conditions of the slope and historical instability cases, adopt a suitable calculation method to update the warning threshold. A feasible way is to set the warning threshold as the mean plus k times the standard deviation (k is a coefficient determined according to risk preference and historical experience) according to the mean and standard deviation of the predicted values. When the window slides to a new time period, as the data is updated, the statistical features of the data within the window will also change, thereby causing the calculated warning threshold to be adjusted accordingly. For example, during the rainy season, the slope stability fluctuates more due to the influence of rainfall. Through the sliding window algorithm, using the prediction data containing more rainfall influence during this period, the calculated warning threshold will be lowered according to the actual fluctuation to more sensitively capture the slope instability risk. Dynamically updating the warning threshold using the sliding window algorithm has significant advantages. On the one hand, compared with the fixed threshold, this method can reflect the dynamic change trend of the slope stability predicted values in real time, avoiding false alarms or missed alarms caused by environmental changes. For example, in areas with complex geological conditions or frequent climate changes, it is difficult for the fixed threshold to adapt to the changing situations, while the dynamically updated threshold can be flexibly adjusted according to the actual data. On the other hand, through the segmented analysis of historical prediction data, the interference caused by short-term data fluctuations is effectively filtered, making the adjustment of the warning threshold more scientific and reliable, thus providing a strong guarantee for the accurate assessment and timely warning of slope stability and better serving the road safety maintenance and disaster prevention work.

[0057] In one embodiment, after outputting the slope instability warning information, it includes: Compile the slope instability warning information into character stream information; Group the character stream information according to preset rules to obtain a plurality of character combinations; each character combination includes 1 or 2 characters; Obtain a preset graph structure template, and sequentially add each character combination to each node of the graph structure template to generate a character combination graph structure; wherein, each node has a corresponding serial number; Sequentially detect whether the character combinations in each node of the character combination graph structure are repeated with the previous nodes. If repeated, perform deduplication, and use the deduplicated nodes as target nodes; Sequentially connect each target node to generate a curve; based on the positional relationship between the curve and each node, reorder the character combinations on each node to obtain a character combination sequence; Obtain a coding table; the coding table includes a corresponding original data column and a coding data column; Sequentially replace the character combinations in the character combination sequence into the coding data column of the coding table until all character combinations are replaced into the coding table or the coding data column is all replaced, to obtain a replacement coding table; Encode the original parameter set based on the replacement coding table to obtain encoded data, and send it to the management terminal.

[0058] In this embodiment, the slope instability warning information includes various types of data, such as slope displacement values, geological parameter changes, meteorological conditions, etc. These data in different formats are uniformly converted into character stream form for subsequent processing and analysis. A specific coding method (such as UTF-8 coding) is used to convert various data elements in the warning information into character sequences to form continuous character stream information. For example, convert the displacement value "10.5" into characters "1", "0", "5", convert the geological parameter "sandstone" into corresponding character combinations, and integrate the character representations of all data to obtain complete character stream information.

[0059] Furthermore, grouping the character stream information and dividing it into smaller character combinations helps to process and analyze the data more carefully. At the same time, restricting the number of characters in each combination (1 or 2 characters) can make the grouping more standardized and orderly. According to the preset grouping rules, starting from the starting position of the character stream, sequentially select 1 or 2 characters as a character combination. For example, if the character stream is "ABCD", according to the rules, character combinations "A", "BC", "D" can be obtained. Reasonably dividing the character stream information provides basic data units for subsequent graph structure generation and further processing, making the data structure clearer and facilitating various analyses and operations.

[0060] Furthermore, using a preset graph structure template, associate character combinations with the nodes of the graph to construct a character combination graph structure. By node numbers, clarify the positions and orders of character combinations, which helps analyze and process character combinations from the perspective of the graph. First, obtain the preset graph structure template, which defines information such as the number of nodes and connection relationships. Then, in the order of character combinations, add each character combination to the corresponding nodes of the graph structure template. If the number of nodes is more than the number of character combinations, the remaining nodes remain empty or are filled according to specific rules. Each node is assigned a unique number to identify its position in the graph structure. Presenting character combinations in the form of a graph structure facilitates analyzing the associations and orders between character combinations, provides an intuitive structural framework for subsequent deduplication and reordering operations, and at the same time, node numbers provide clear indexes for data processing.

[0061] In the character combination graph structure, detecting and removing duplicate character combinations can reduce data redundancy, make the character combination sequence more concise and accurate, and highlight the uniqueness and effectiveness of the data. Starting from the first node of the graph structure, sequentially compare the character combination of the current node with the character combinations of all previous nodes (nodes determined to be in front according to the arrangement of node numbers). If a duplicate is found (i.e., the contents of the character combinations are exactly the same), then delete the current node and retain the non-duplicate nodes. The deduplicated node is then used as the target node. For example, if the character combinations of the first three nodes in the graph structure are "A", "BC", and "A" respectively, and it is detected that the character combination "A" of the third node is duplicate with the first node, then the third node is deduplicated and used as the target node.

[0062] In this embodiment, removing duplicate character combinations and optimizing the character combination graph structure enable subsequent processing to be based on more refined data, improve the efficiency of data processing and the accuracy of results, and avoid errors or redundant analysis caused by duplicate data.

[0063] Furthermore, by connecting the target nodes to generate a curve and reordering the character combinations using the positional relationship between the curve and the nodes, potential order relationships between character combinations can be explored, making the character combination sequence more conform to a certain logic or rule, and providing a more reasonable order for subsequent data encoding. Specifically, connect the deduplicated target nodes in sequence according to their positions in the graph structure to form a curve. According to the trend of the curve and the positions of each node on the curve, re-determine the order of the character combinations. For example, if the curve passes through the target nodes from left to right, adjust the character combinations on the nodes according to the order of the nodes on the curve to obtain a new character combination sequence. The positional relationship can be determined by calculating the coordinates or distances of the nodes on the curve, etc., and then sorting is performed. Rearranging the character combinations makes the character combination sequence more unique and provides a more suitable data order for encoding operations based on the encoding table.

[0064] The coding table serves as the basis for data coding. Through the one-to-one mapping relationship between the original data column and the coded data column, character combinations are converted into specific coding forms to achieve data encryption or standardized representation, facilitating data transmission and storage. Obtain a pre-established coding table that contains two columns. One column is the original data column for storing the original character combinations or other related data, and the other column is the coded data column for storing the coded values corresponding to the original data column. For example, in the coding table, it is stipulated that the original data "12" corresponds to the coded data "AB", which clarifies the mapping relationship between the original data and the coded data. It provides specific rules and a basis for data coding, enabling character combinations to be converted according to a specific mapping relationship, achieving effective data coding, ensuring the security and standardization of data during transmission, and facilitating the management terminal to decode and analyze the data.

[0065] Finally, match the character combinations in the character combination sequence with the coded data column in the coding table, and replace the corresponding content in the coded data column with the character combinations to generate a unique replacement coding table for subsequent data sending. Starting from the first character combination in the character combination sequence, replace the first coded data in the coded data column with the character combination. Repeat this process, processing each character combination in the character combination sequence in turn until all character combinations are replaced in the coding table or the coded data column is fully replaced, obtaining the replacement coding table.

[0066] Furthermore, use the replacement coding table to code the original parameter set, convert the original parameters into coded data corresponding to the replacement coding table, achieve data standardization and encryption processing, ensure the security and effectiveness of data during transmission, and facilitate the management terminal to receive and analyze.

[0067] According to the mapping relationship between the character combinations and the original data in the replacement coding table, code each parameter in the original parameter set. For example, the original parameter set contains the parameters "12" and "4". According to the mapping rules of the replacement coding table, code "12" into the corresponding coded value and "4" into the corresponding value to obtain the coded data. Then, through a specific communication protocol and channel, send the coded data to the management terminal to ensure the accuracy and reliability of data transmission. Code and send the original parameter set to achieve the secure transmission of data from the sending end to the management terminal. The management terminal can decode the received coded data according to the coding table to restore the original parameter set for further analysis and decision-making. Through the above improved coding table, the original parameter set can have stronger security during transmission, avoiding data leakage. At the same time, the generation of the above replacement coding table is closely related to the slope instability warning information, enhancing the reusability and traceability.

[0068] Refer to Figure 2, in another embodiment of the present invention, there is also provided a device for predicting the stability of a road slope, including: An acquisition unit, configured to collect multi-dimensional environmental parameters of the road slope in real time to form an original parameter set; A screening unit, configured to analyze the original parameter set based on the Granger causality test algorithm, determine the causal influence relationships between the parameters, construct a parameter causal network, and screen out the key parameters affecting the slope stability; An association unit, configured to calculate the similarity of each of the key parameters in the time and space dimensions through the dynamic time warping algorithm to generate a spatio-temporal association matrix; A fusion unit, configured to perform weighted fusion on the key parameters based on the spatio-temporal association matrix to obtain a fusion parameter; A prediction unit, configured to input the fusion parameter into a pre-trained slope stability prediction model to obtain a predicted value of the slope stability; An early warning unit, configured to dynamically update the early warning threshold of the slope stability prediction model, and output a slope instability early warning message when the predicted value exceeds the dynamically updated early warning threshold.

[0069] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to that described in the above method embodiment, and details are not described herein again.

[0070] Refer to Figure 3 , in an embodiment of the present invention, there is also provided an electronic device. This electronic device may be a server, and its internal structure may be as Figure 3 shown. The electronic device includes a processor, a memory, a display screen, an input device, a network interface, and a database connected through a system bus. Among them, the processor of this computer design is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the electronic device is used to store the corresponding data in this embodiment. The network interface of the electronic device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0071] Those skilled in the art can understand that Figure 3 the structure shown in

[0072] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0073] In summary, the present invention provides a method, device, storage medium and electronic device for predicting the stability of a road slope in an embodiment of the present invention, including: collecting multi-dimensional environmental parameters of the road slope in real time to form an original parameter set; based on the Granger causality test algorithm, analyzing the original parameter set to determine the causal influence relationship between parameters, constructing a parameter causal network, and screening out key parameters affecting the slope stability; calculating the similarity of each of the key parameters in the time and space dimensions through the dynamic time warping algorithm to generate a spatio-temporal correlation matrix; based on the spatio-temporal correlation matrix, performing weighted fusion on the key parameters to obtain a fusion parameter; inputting the fusion parameter into a pre-trained slope stability prediction model to obtain a predicted value of the slope stability; dynamically updating the warning threshold of the slope stability prediction model, and when the predicted value exceeds the dynamically updated warning threshold, outputting a slope instability warning message. In the present invention, by analyzing the original parameter set, determining the causal influence relationship between parameters, constructing a parameter causal network, screening out key parameters affecting the slope stability, calculating the similarity of each of the key parameters in the time and space dimensions, generating a spatio-temporal correlation matrix, and dynamically updating the warning threshold of the slope stability prediction model at the same time; fully excavating the potential relationship between multi-dimensional environmental parameters, overcoming the defects of inaccurate prediction and data redundancy in the current road slope stability prediction method.

[0074] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium provided by the present invention and used in the embodiments can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or an external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0075] It should be noted that in this text, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method including a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, apparatus, article, or method including that element.

[0076] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall equally be included in the patent protection scope of the present invention.

Claims

1. A method for predicting the stability of road slopes, characterized in that, It includes the following steps: Collect the multi-dimensional environmental parameters of the road slope in real time to form an original parameter set; Based on the Granger causality test algorithm, analyze the original parameter set, determine the causal influence relationships among the parameters, construct a parameter causal network, and screen out the key parameters affecting the slope stability; Calculate the similarities of the key parameters in the time and space dimensions through the dynamic time warping algorithm to generate a spatio-temporal correlation matrix; Based on the spatio-temporal correlation matrix, perform weighted fusion on the key parameters to obtain fusion parameters; Input the fusion parameters into a pre-trained slope stability prediction model to obtain the predicted value of the slope stability; Dynamically update the warning threshold of the slope stability prediction model. When the predicted value exceeds the dynamically updated warning threshold, output a slope instability warning message.

2. The method for predicting the stability of a road slope according to claim 1, wherein The environmental parameters include geological parameters, mechanical parameters, environmental parameters, and displacement parameters.

3. The road slope stability prediction method according to claim 1, characterized in that, When performing weighted fusion on the key parameters, assign weights based on the correlation degree on the spatio-temporal correlation matrix, and generate fusion parameters through matrix dot product operation.

4. The method for predicting the stability of a road slope according to claim 1, characterized in that, Based on the Granger causality test algorithm, analyze the original parameter set, determine the causal influence relationships among the parameters, construct a parameter causal network, and screen out the key parameters affecting the slope stability, including: Calculate the grey correlation degrees between the parameters in the original parameter set and the historical data of slope stability, perform a preliminary ranking of the parameters according to the grey correlation degrees, and give priority to performing Granger causality tests on the parameter pairs with high correlation degrees; When performing Granger causality tests, based on the change point detection algorithm, identify the structural change points of the parameters in the time series, and set different lag orders of the Granger causality tests for different mutation intervals; Based on the causal relationship strength evaluation model, calculate the causal relationship strength scores between each parameter pair; retain the parameter pairs with causal relationship strength scores higher than the threshold, and construct a parameter causal network; Use the Louvain algorithm to perform community division on the parameter causal network, calculate the cohesion coefficient of each community and its contribution to the slope stability; select the core parameters in the community with a contribution higher than the set value, and the hub parameters connecting different communities as the key parameters affecting the slope stability.

5. The road slope stability prediction method according to claim 1, characterized in that Based on the Granger causality test algorithm, analyze the original parameter set, determine the causal influence relationships among the parameters, construct a parameter causal network, and screen out the key parameters affecting the slope stability, including: Perform three-dimensional convolution on the original parameter set through a three-dimensional convolutional kernel to obtain a spatio-temporal parameter set with enhanced features; Adopt a dynamic time window division method, adaptively adjust the time window length according to the parameter fluctuation frequency, segment the spatio-temporal parameter set to obtain multiple parameter subsequences; Perform Granger causality tests on each parameter subsequence respectively, and use the attention mechanism to calculate the weights of the test results of each parameter subsequence, and perform weighted fusion to obtain a comprehensive causal relationship matrix; Based on the recurrent neural network, learn the evolution law of the comprehensive causal relationship matrix over time, and predict the change trend of the causal influence relationships among the parameters in the future period. Based on the comprehensive causal relationship matrix and the change trend, construct a parameter causal network with a time dimension, calculate the temporal importance of each node in the parameter causal network, and select the parameters with temporal importance higher than the preset value as the key parameters affecting slope stability.

6. The method for predicting the stability of a road slope according to claim 1, characterized in that Calculate the similarity of each of the key parameters in the time and space dimensions through the dynamic time warping algorithm to generate a spatio-temporal correlation matrix, including: Use a convolutional neural network to extract the local features of the key parameters in the space dimension and splice them into spatial sequence features; Use a long short-term memory network to capture the long-term dependence features of the key parameters in the time dimension and splice them into time sequence features; Map the time sequence features to quantum states, and through the dynamic time warping algorithm, combine quantum parallel computing for multiple time warping paths, evaluate the similarity of each time warping path, and screen out the optimal matching path to obtain the time dimension similarity matrix; Based on the pre-constructed spatial similarity evaluation model, analyze the spatial sequence features to generate a spatial dimension similarity matrix; Through tensor product operation, fuse the time dimension similarity matrix and the spatial dimension similarity matrix to obtain the spatio-temporal correlation matrix.

7. The method for predicting the stability of a road slope according to claim 1, characterized in that, Dynamically update the warning threshold of the slope stability prediction model, including: Based on historical prediction data, use the sliding window algorithm to dynamically update the warning threshold of the slope stability prediction model.

8. A device for predicting the stability of a road slope, characterized in that, Including: An acquisition unit for real-time collecting multi-dimensional environmental parameters of the road slope to form an original parameter set; A screening unit for analyzing the original parameter set based on the Granger causality test algorithm, determining the causal influence relationship between parameters, constructing a parameter causal network, and screening out the key parameters affecting slope stability; An association unit for calculating the similarity of each of the key parameters in the time and space dimensions through the dynamic time warping algorithm to generate a spatio-temporal correlation matrix; A fusion unit for performing weighted fusion on the key parameters based on the spatio-temporal correlation matrix to obtain a fusion parameter; A prediction unit for inputting the fusion parameter into a pre-trained slope stability prediction model to obtain a predicted value of slope stability; An early warning unit for dynamically updating the warning threshold of the slope stability prediction model, and when the predicted value exceeds the dynamically updated warning threshold, outputting a slope instability warning message.

9. An electronic device, comprising a memory and a processor, wherein a computer program is stored in the memory, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 7.

Images