Intelligent identification method and system for slope failure instability mode

Abstract

The invention belongs to the technical field of disaster identification and prediction, and provides a slope failure instability mode intelligent identification method and system, and the technical scheme is that the method comprises the steps: taking the engineering feature information subjected to parameterized coding processing in a slope failure instability database as an input parameter, taking the slope failure instability mode as an output parameter, and constructing a plurality of parameter combination schemes; each individual of the genetic algorithm is expressed as an input feature subset, performance evaluation is performed on the slope failure instability mode intelligent identification model trained under different parameter combination schemes, a fitness function of each individual is determined according to a performance evaluation result, and a feature subset corresponding to the optimal individual is found through iterative optimization; and predicting the to-be-identified scheme based on the obtained feature subset corresponding to the optimal individual to obtain a slope failure instability mode prediction result. And the efficiency and the accuracy of the slope failure instability mode identification means are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of disaster identification and prediction, and particularly relates to an intelligent identification method and system for slope failure and instability mode. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] As a frequently-occurring geological disaster, landslide seriously threatens human production and development and easily causes huge loss of life and property. The instability and failure of a slope is a complex geomechanical process, and understanding the instability and failure mode and the incubation and evolution characteristics thereof is an important basis for subsequent slope stability evaluation, monitoring and early warning, and prevention and control. The slope generally needs to undergo a long time course and the combined action of numerous factors from the incubation of a geological disaster to the final instability and failure. A large number of slope instability cases show that the slope will lead to significant differences in the failure and instability mode due to different topography and geomorphology and geological structure, and the deformation and evolution characteristics will also be different due to different failure and instability modes.

[0004] The prediction of the traditional slope failure and instability mode is obtained by relevant researchers through on-site geological investigation, however, the classification of the slope failure and instability mode is not uniform in the industry, the subjective experience of the researchers has a great influence, time and effort are consumed, the efficiency is low, and there is a certain misjudgment. In recent years, machine learning has been gradually applied to the identification of the slope instability mode, trying to replace the traditional manual identification to improve the efficiency. However, the existing machine learning scheme has a core technical bottleneck, that is, the geological information of the slope is not subjected to systematic parameterization and standardization processing, the geological characteristics of the slope are mostly qualitative and textual description, the non-numerical geological characteristics have no uniform coding rules, the model cannot identify the correlation between the "textual description" and the "instability mode", and the hierarchical relationship of the geological characteristics is ignored, the influence weight of different characteristic information on the instability mode is different, resulting in poor implementability, low prediction accuracy, and difficulty in meeting the actual engineering requirements. SUMMARY

[0005] In order to solve at least one technical problem in the background art, the present application provides an intelligent identification method and system for slope failure and instability mode, which considers the hierarchical relationship of the geological characteristics of the slope, predicts the failure and instability mode of the slope to be identified according to the topography and geological conditions, improves the identification efficiency and accuracy of the slope failure and instability mode, and meets the actual engineering requirements.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: The first aspect of the present application provides an intelligent identification method for slope failure and instability mode, comprising the following steps: Based on the obtained historical related engineering feature information, a slope failure instability database is established; Based on the set parameterization coding rule, the engineering feature information and the slope failure instability mode in the slope failure instability database are subjected to parameterization coding processing; The engineering feature information subjected to the parameterization coding processing in the slope failure instability database is taken as an input parameter, and the slope failure instability mode is taken as an output parameter, and a plurality of parameter combination schemes are constructed; Each individual of the genetic algorithm is represented as an input feature subset, the slope failure instability mode intelligent identification model trained under different parameter combination schemes is subjected to performance evaluation, the fitness function of each individual is determined according to the performance evaluation result, and the optimal individual corresponding feature subset is found out through iterative optimization; Based on the obtained optimal individual corresponding feature subset, the to-be-identified scheme is predicted, the slope failure instability mode prediction result is obtained, and the slope failure instability database is updated according to the slope failure instability mode prediction result.

[0007] Further, the slope failure instability mode is divided into 4 categories according to the slope material type and the sliding surface shape and cause, including circular arc non-structural surface sliding, structural surface sliding, toppling fracture surface sliding and buckling fracture surface sliding, wherein the toppling fracture surface sliding includes two types of toppling fracture surface sliding of the slope with anti-inclination structure and toppling fracture surface sliding of the slope with pro-inclination structure.

[0008] Further, the parameterization processing of the engineering feature information and the slope failure instability mode in the slope failure instability database includes: For numerical data, direct assignment is performed; For non-numerical data, a one-hot encoding form is adopted to convert it into numerical data, in the one-hot encoding, each category is converted into a binary vector, the length of the vector is the same as the number of categories, and only one element is 1 and the rest are 0.

[0009] Further, the parameterized coding processing engineering characteristic information in the slope failure instability database is taken as an input parameter, and the slope failure instability mode is taken as an output parameter, a plurality of parameter combination schemes are constructed, including: single parameter combination, double parameter combination, three parameter combination and four parameter combination; wherein the single parameter combination includes slope terrain, slope material, slope internal rock structure and slope rock layer inclination; the double parameter combination includes slope terrain and slope material, slope terrain and slope internal rock structure, slope terrain and slope rock layer inclination, slope material and slope internal rock structure, slope material and slope rock layer inclination, slope internal rock structure and slope rock layer inclination; the three parameter combination includes slope terrain, slope material and slope internal rock structure; slope terrain, slope material and slope rock layer inclination; slope terrain, slope internal rock structure and slope rock layer inclination; slope material, slope internal rock structure and slope rock layer inclination; and the four parameter combination includes slope terrain, slope material, slope internal rock structure and slope rock layer inclination.

[0010] Further, each individual of the genetic algorithm is represented as an input feature subset, each individual adopts binary coding, represents a specific input feature subset; each individual corresponds to an n-bit binary string, each bit represents whether a feature is selected or not, if 1 represents selected, if 0 represents not selected.

[0011] Further, the slope failure instability mode intelligent identification model adopts a generalized regression neural network.

[0012] Further, the fitness value function of each individual is represented as: fitness function = alpha x classification accuracy + beta x geological consistency score + gamma x feature redundancy, wherein alpha, beta and gamma represent weight coefficients.

[0013] The second aspect of the present application provides a slope failure instability mode intelligent identification system, comprising: A database establishment module is configured to establish a slope failure instability database based on the obtained historical related engineering characteristic information; A parameterized coding module is configured to perform parameterized coding processing on the engineering characteristic information and the slope failure instability mode in the slope failure instability database based on a set parameterized coding rule; An iterative optimization module is configured to take the parameterized coding processing engineering characteristic information in the slope failure instability database as an input parameter, take the slope failure instability mode as an output parameter, construct a plurality of parameter combination schemes, represent each individual of a genetic algorithm as an input feature subset, evaluate the performance of the slope failure instability mode intelligent identification model trained under different parameter combination schemes, determine the fitness function of each individual according to the performance evaluation result, and find out the feature subset corresponding to the optimal individual through iterative optimization; An identification module is configured to predict the to-be-identified scheme based on the obtained feature subset corresponding to the optimal individual, obtain a slope failure instability mode prediction result, and update the slope failure instability database according to the slope failure instability mode prediction result.

[0014] A computer readable storage medium having stored thereon a computer program, which, when executed by a processor, implements the steps of the slope failure instability mode intelligent identification method.

[0015] The fourth aspect of the present application provides a computer device.

[0016] A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the slope failure instability mode intelligent identification method when executing the program.

[0017] Compared with the prior art, the present application has the following beneficial effects: The present application encodes the geological features of the slope according to the set coding rules, considers the hierarchical relationship between the geological features based on the coded geological features, and improves the efficiency and accuracy of the high slope failure instability mode identification method, while establishing a standardized processing system of geological information to realize the transformation from "manual qualitative" to "machine recognizable quantitative".

[0018] The advantages of the additional aspects of the present application will be partially given in the following description, partially will become obvious from the following description, or will be understood through the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0019] The drawings accompanying the specification of the present application form a part thereof, serve to provide further understanding of the present application, and together with the description of the exemplary embodiments of the present application and their description serve to explain the present application, and do not constitute an improper limitation of the present application.

[0020] Figure 1 is a slope failure instability mode intelligent identification method flow chart provided by an embodiment of the present application; Figure 2 is a structure of a slope failure instability mode intelligent identification model provided by an embodiment of the present application; Figure 3 is a slope failure instability mode intelligent identification optimal model prediction accuracy result provided by an embodiment of the present application; Figure 4 is a slope failure instability mode intelligent identification optimal model result confusion matrix provided by an embodiment of the present application. DETAILED DESCRIPTION

[0021] The application will be further described below with reference to the drawings and embodiments.

[0022] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0023] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, steps, operations, elements, components, and / or groups thereof, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.

[0024] Embodiment one As Figure 1 shown, the embodiment provides a slope failure instability mode intelligent identification method, comprising the following steps: Step 1: Based on the obtained historical related engineering characteristic information, a slope failure instability database is established; In this embodiment, slope failure cases are collected through field investigation or literature research, and the material type, topography and geological structure and other related engineering characteristic information of all cases are obtained, a slope failure instability database is established, and the specific steps are as follows: Specifically, the obtained historical related engineering characteristic information includes: slope topography, slope material, internal rock structure of slope, slope rock layer inclination and slope failure instability mode; Among them, the slope topography includes the slope angle, and the methods that can be used to measure the slope angle include: the level method, the total station method and the manual measurement method; The slope material is a soil material or a rock material, which can be determined by observing the external morphology, soil color, vegetation, surface water flow and other ground manifestations of the slope.

[0025] The internal rock structure of the slope is a reverse dip structure or a dip structure, and if the slope material is a soil material, this information does not need to be collected; The slope rock layer inclination, if the slope material is a soil material, this information does not need to be collected; the commonly used measurement methods of rock layer inclination include the inclination instrument measurement method, the baffle method, the laser scanning method and the like.

[0026] The slope failure instability mode is initially divided into four categories according to the slope material type and the sliding surface shape and causes, including a circular arc non-structural surface sliding, a structural surface sliding, a toppling rupture surface sliding and a buckling rupture surface sliding. The third category of the damage mode in the classification mode, the toppling rupture surface sliding, is divided into two types according to the type of the slope that occurs the damage, i.e. the toppling rupture surface sliding of the slope with an anti-inclined structure and the toppling rupture surface sliding of the slope with an inclined structure, and the two types of slopes have different internal rock structure and the input data structure is obviously different, so that the information input is clearer and the prediction classification accuracy is higher, and in the application, the two are regarded as two damage modes for discussion. Therefore, there are finally five damage modes in total.

[0027] Step 2: The engineering feature information and the slope failure instability mode in the slope failure instability database are parameterized and coded based on the set parameterized coding rules; The input information is parameterized and processed, the complex and unstructured original information is converted into quantifiable and standardized parameter form, the information is structured and standardized, the ambiguity of information of different sources and formats is eliminated, a unified input format is provided for subsequent model machine learning and numerical input, the compatibility of information processing is improved, the information complexity is simplified by extracting key feature parameters, the core information is retained while the redundancy is removed, and the analysis and calculation efficiency is improved. The specific processing process is as follows: (1) For the numerical data of the slope angle and the rock layer inclination angle, the data can be directly input. The processing mode is as follows: landslide topography (i.e. landslide slope angle β): the collected value must satisfy 0° < β < 90°, which corresponds to the geometric rationality of the natural slope. Float type numerical value (such as 52.3°) is directly used for input; landslide rock layer inclination angle α: the collected value must satisfy 0° < α < 90°. For soil landslide, there is no rock structure, and 0 is forcibly assigned as a distinguishing identifier for rock slope; for inclined rock slope: the measured inclination value (such as α = 35°) is input; for anti-inclined rock slope: two modes are provided for input: mode 1: the measured value is directly input (such as α = 70°); mode 2: the equivalent angle 180°-α (such as 180°-70° = 110°) is converted.

[0028] (2) For the non-numerical data of the slope material and the internal rock structure of the slope, further processing is needed, and the data is converted into numerical data in the form of one-hot coding. In one-hot coding, each category is converted into a binary vector, the length of the vector is the same as the number of categories, and only one element is 1 and the rest are 0.

[0029] Specifically, in the representation method of the landslide material, a1 represents whether the landslide is a rock material, a1 takes "1" to represent that the landslide is a rock material, and a1 takes "0" to represent that the landslide is a non-rock material; in the rock layer structure inside the landslide, a2 represents whether the landslide is a dip structure, a2 takes "1" to represent that the landslide is a dip structure, and a2 takes "0" to represent that the landslide is a non-dip structure.

[0030] (3) For the slope failure instability mode, a parameterization processing method is proposed in the embodiment, the processing object of this scheme is only non-numeric data (i.e. slope material and rock layer structure), which is a coding parameterization processing form of these two types of non-numeric category attributes; and the numeric data (slope angle, rock layer inclination) needs to be processed separately according to the direct assignment principle, and the two together constitute the complete input of the model. Specifically, it includes: In the representation method of the landslide material, m1 represents whether it is a soil material, m1 takes "1" to represent that the landslide is a soil material, and m1 takes "0" to represent that the landslide is a non-soil material; in the representation method of the rock layer structure inside the landslide, m2 is selected to represent whether the landslide is a dip structure, m2 takes "1" to represent that the landslide is a dip structure, and m2 takes "0" to represent that the landslide is a non-dip structure. In summary, in the parameterization processing scheme, the input of the rock dip landslide is (0, 1), the input of the rock anti-dip landslide is (0, 0), and the input of the soil landslide is (1, 0).

[0031] The parameterization processing scheme is shown in Table 1: Table 1 Parameterization processing scheme

[0032] In the traditional landslide failure mode recognition method, human judgment and expert determination are mostly used for failure mode classification. The present application parameterizes the landslide failure mode to ensure that the landslide failure mode can be automatically determined, rather than relying on an expert system, thereby reducing processing costs and reducing subjective errors in human interpretation, ensuring the stability and reliability of the information processing results.

[0033] The third failure mode in the classification method, i.e. the sliding of the collapse fracture surface, is divided into two types according to the structural characteristics of the slope that occurs, i.e. the sliding of the collapse fracture surface of the slope with an anti-dip structure and the sliding of the collapse fracture surface of the slope with a dip structure. The two types of slopes have different internal rock layer structures, and the input data structures are obviously different. In order to make the information input clearer and the classification accuracy higher, the two are discussed as two types of failure modes. The parameterization processing of the output parameters is as follows: "1" represents structural surface sliding, "2" represents circular arc structural surface sliding, "3" represents buckling fracture surface sliding, "4" represents dip collapse fracture surface sliding, and "5" represents anti-dip collapse fracture surface sliding.

[0034] Step 3: The parameterized and coded engineering feature information in the slope failure instability database is taken as the input parameter, and the slope failure instability mode is taken as the output parameter to construct multiple parameter combination schemes. Each individual of the genetic algorithm is represented as a subset of input features. The performance of the trained slope failure instability mode intelligent identification model under different parameter combination schemes is evaluated. The fitness function of each individual is determined according to the performance evaluation results. The optimal individual corresponding to the feature subset is found through iterative optimization. In this embodiment, the generalized regression neural network is used as the identification algorithm for the slope failure instability mode, and the genetic algorithm is used as the optimization algorithm for the input feature combination. The genetic algorithm performs global search in the solution space (all possible feature subsets). The generalized regression neural network evaluates each candidate feature combination provided by the genetic algorithm and feeds back its identification performance (fitness). Through this iterative optimization, the optimal input feature subset that maximizes the identification accuracy and generalization ability of the generalized regression neural network model is automatically selected. This realizes the automation and intelligent optimization of the model input end, overcomes the subjectivity and limitations of traditional manual feature selection, and efficiently searches the high-dimensional slope geological feature space. Thus, the most critical feature input is locked for the generalized regression neural network model, and the reliability of slope stability analysis is improved from the source.

[0035] Specifically, the following steps are included: Step 301: The parameterized and coded engineering feature information in the slope failure instability database is taken as the input parameter, and the slope failure instability mode is taken as the output parameter to construct multiple parameter combination schemes. In this embodiment, the learning sample input parameters of the slope failure instability mode intelligent identification model include four types, which are: slope terrain (i.e., slope angle), slope material (i.e., soil or rock material), internal rock structure of slope (i.e., reverse or inclined structure), and rock layer inclination. The output parameter of the model is the slope failure instability mode.

[0036] Based on the above four types of input parameters, the following parameter combination schemes are determined, which specifically include: ① Single parameter combination (4 groups): slope terrain (slope angle); slope material (soil / rock); internal rock structure of slope (reverse / inclined); slope rock layer inclination; ② Double parameter combination (6 groups): slope terrain and slope material, slope terrain and internal rock structure of slope, slope terrain and slope rock layer inclination, slope material and internal rock structure of slope, slope material and slope rock layer inclination, internal rock structure of slope and slope rock layer inclination; ③ Three-parameter combination (4 groups): slope terrain, slope material and internal rock structure of slope; slope terrain, slope material and slope rock dip angle; slope terrain, internal rock structure of slope and slope rock dip angle; slope material, internal rock structure of slope and slope rock dip angle; ④ Four-parameter combination (1 group): slope terrain, slope material, internal rock structure of slope and slope rock dip angle.

[0037] Step 302, each parameter combination scheme corresponds to an individual in the genetic algorithm, and each individual is encoded in binary, representing a specific input feature subset; In this embodiment, n types of core input features are represented by binary encoding, and each "individual" corresponds to an n-bit binary string, each bit representing whether a feature is selected or not. If it is 1, it means selected, and if it is 0, it means not selected. For example: slope angle = A, material = B, rock structure = C, rock dip angle = D, binary string "1000" represents "single parameter combination (only A)", and "1111" represents "four parameter combination (A+B+C+D)".

[0038] Step 303, for each individual in the population of each generation, decode its corresponding feature combination and use it as the input of the slope failure instability mode intelligent identification model to obtain the performance evaluation result. The performance evaluation result of the slope failure instability mode intelligent identification model is defined as the fitness function, and the optimal individual and the corresponding optimal input feature subset are selected based on the fitness function. In this embodiment, the slope failure instability mode intelligent identification model uses a generalized regression neural network, and the structure is as shown in Figure 2 The generalized neural network consists of four layers: input layer, pattern layer, summation layer and output layer.

[0039] Input layer: the number of input layer neurons is 5, each neuron corresponds to a different sample, and the neuron directly transmits the information received by the input layer to the pattern layer.

[0040] Pattern layer: the number of pattern layer neurons is 80, each neuron of the pattern layer corresponds to a different learning sample, and the transfer function formula (1) of the pattern layer neuron is as follows: (1), Wherein, represents the network input variable; is the learning sample corresponding to the th neuron, .

[0041] Summation layer: two types of neurons are used for summation in the summation layer.

[0042] The first type of neuron calculation formula (2) is an arithmetic sum of all pattern layer neurons, the connection weight value between the pattern layer and each neuron is 1, and the transfer function is formula (3).

[0043] (2), (3), The second type of neuron calculation formula is (4), which is a weighted sum of all pattern layer neurons, the connection weight value between the first th neuron in the pattern layer and the molecular sum neuron in the sum layer is the first element in the first th output sample , and the transfer function is formula (5).

[0044] (4), (5), Output layer: the number of output layer neurons is 1, each neuron divides the output of the sum layer, the neuron output corresponds to the estimated result, and formula (6) is as follows: (6), The fitness function is a core component in genetic algorithms, used to quantitatively evaluate the goodness of individuals (i.e., candidate solutions) in the population. Its role is similar to the "survival of the fittest" principle in nature, guiding the algorithm to evolve towards better solutions by assigning fitness values. Traditional fitness only relies on classification accuracy; In this embodiment, the performance evaluation result of the slope failure instability mode intelligent identification model is defined as the fitness function, prior knowledge constraints of slope instability are added in the present application, and the fitness value function of each individual is represented as: Fitness function = alpha x classification accuracy + beta x geological consistency score + gamma x feature redundancy; Wherein, the confusion matrix, accuracy, precision, recall and F1-Score and other classification performance evaluation indexes are used to evaluate the performance of all models that have been successfully trained, specifically, the accuracy (Accuracy): the proportion of correctly classified samples to the total number of samples, that is, the proportion of the number of correctly identified slope failure instability mode types in the total number of samples, formula (7) is as follows: (7), Wherein, TP represents the number of samples correctly predicted by the model as a certain class, FP represents the number of samples incorrectly predicted by the model as the class, and FN represents the number of samples incorrectly predicted by the model as other classes.

[0045] Precision: The ratio of samples that are actually of a certain type of failure mode among all samples that are predicted to be of the type of failure mode, formula (8) is as follows: (8), Recall: The ratio of samples that are predicted to be of a certain type of failure mode among all samples that are actually of the type of failure mode, formula (9) is as follows: (9), F1-Score: The harmonic mean of precision and recall, used to balance the two indicators, formula (10) is as follows: (10).

[0046] In this embodiment, the classification accuracy is finally used as the classification accuracy for calculating the fitness.

[0047] The geological consistency score and feature redundancy can be selected to be calculated; the geological consistency score is used to punish the feature combination that does not conform to the known mechanism; the feature redundancy is used to preferentially select the minimal combination with clear physical meaning. Weight coefficients (a, b, g): all are non-negative real numbers, satisfying a+b+g=1 (to ensure that the weighted value is within a reasonable range). a must be greater than 0 (classification accuracy is a mandatory item), and b and g can be set to 0 according to actual needs (i.e., not to enable the corresponding indicators).

[0048] The calculation process of the geological consistency score is as follows: ① Set configurable scoring parameters. These parameters at least include: (a) Configurable base score (S b ): Define a full score benchmark value, representing the ideal score when fully complying with the rules. The value is not limited to 100 points, and can be set according to actual needs. (b) Configurable constraint weight (W major , W minor ): Define the deduction weight for different importance levels of rule violation behaviors. For example, configure the deduction weight W major for the key constraint, and the deduction weight W minor for the secondary constraint. The weight value is determined by the domain experts according to the actual impact of the constraint item on the slope stability.

[0049] ② Set the initial score Score initial of the candidate feature combination to the configured base score S b .

[0050] ③ Call the pre-defined slope geological rule library to compare the input candidate feature combination with each rule in the rule library. Each rule in the rule library is pre-defined with its constraint importance (such as "key" or "secondary").

[0051] (iv) In the checking process, identify all instances of candidate feature combinations violating rules. For each violation instance: (a) If the violated rule is a key constraint, deduct W major points from the current score; (b) If the violated rule is a secondary constraint, deduct W minor points from the current score. Accumulate all deductions to obtain the total penalty Penalty total .

[0052] (v) Calculate the final consistency score Score final , using the formula: To ensure the reasonableness of the score, set the lower limit of the score to 0, i.e. when Score final is less than 0, set it to 0.

[0053] The feature redundancy calculation process is as follows: (i) Determine the redundancy of each feature in the candidate feature combination.

[0054] The criteria include but are not limited to: (1) Statistical redundancy: based on historical data, there is a high statistical correlation with other features in the combination. (2) Physical redundancy: there is a fundamental geological contradiction with the known basic type of slope. For example, in the "soil slope" feature combination, the "rock structure" related feature is determined to be physically redundant.

[0055] (ii) Redundant feature quantity statistics: according to the determination results of step (i), count the following quantities: (1) N redundant : the number of features in the candidate feature combination that are determined to be "statistically redundant". (2) N physical : the number of features in the candidate feature combination that are determined to be "physically redundant". (3) N total : the total number of features in the candidate feature combination.

[0056] (iii) Configure weight parameters: set a weight coefficient a, usually in the range of 0.3 to 0.5, to balance the proportion of general redundancy and physical redundancy in the total redundancy calculation. A higher weight (1-a) is assigned to the physical redundancy item to reflect a stricter penalty. The specific value of a can be adjusted according to engineering experience.

[0057] (iv) Feature redundancy calculation: based on the above parameters, calculate the feature redundancy R according to the following formula: , The design of this formula makes the final redundancy R value range between 0 and 1. The higher the R value, the more serious the redundancy of the feature combination.

[0058] Step 5: Based on the obtained optimal individual corresponding feature subset, the to-be-identified scheme is predicted to obtain the slope failure instability mode prediction result.

[0059] To verify the technical effect of the present application, 80 groups of instability slope cases in the database are selected, including 16 groups of each of 5 slope failure instability modes (circular-arc non-structural surface sliding, structural surface sliding, toppling rupture surface sliding-tilt, toppling rupture surface sliding-anti-tilt, and buckling rupture surface sliding). Each data group contains 4 kinds of related engineering feature information of the instability slope: slope terrain (i.e. slope angle), slope material (i.e. soil material or rock material), internal rock structure of the slope (i.e. anti-tilt structure or tilt structure, rock layer inclination), and slope failure instability mode. The selected database details are shown in Table 2: Table 2 Initial information of related engineering features of slope failure instability database

[0060] According to the data preprocessing process of step 2, the final processing result is shown in Table 3: Table 3 Result after data preprocessing

[0061] The classification performance evaluation indexes such as confusion matrix, accuracy, precision, recall and F1-Score are used to evaluate the performance of all trained models, and the optimal slope failure instability mode intelligent identification model is found out. The optimal model performance evaluation indexes and the number of input parameters are shown in Table 4: Table 4 Optimal model performance evaluation indexes and number of input parameters

[0062] Figure 3 The optimal model shows the prediction accuracy of the five kinds of slope failure instability modes on the test set, and intuitively compares the identification ability of the model for different categories. From Figure 3 It can be clearly seen that the prediction accuracy of the model for categories 2, 4 and 5 reaches 100%, indicating that the model is extremely accurate in identifying the three modes of "circular-arc non-structural surface sliding", "tilt toppling rupture surface sliding" and "anti-tilt toppling rupture surface sliding". In contrast, the identification accuracy of the model for categories 1 and 3 is different, which is an acceptable error. Figure 4is a more detailed confusion matrix chart in model performance evaluation. It specifically shows the correspondence between the model prediction results and the true labels through the matrix form, and can clearly reveal the classification error types of the model. It can be observed that the samples of class 1 are mainly misjudged as samples of class 3.

[0063] Embodiment two The embodiment provides a slope failure instability mode intelligent identification system, which comprises: A database establishment module is configured to establish a slope failure instability database based on the obtained historical related engineering feature information; A parameterization coding module is configured to perform parameterization coding processing on the engineering feature information and the slope failure instability mode in the slope failure instability database based on a set parameterization coding rule; An iterative optimization module is configured to take the parameterization coding processed engineering feature information in the slope failure instability database as an input parameter, take the slope failure instability mode as an output parameter, construct a plurality of parameter combination schemes, represent each individual of a genetic algorithm as an input feature subset, perform performance evaluation on the trained slope failure instability mode intelligent identification model under different parameter combination schemes, determine the fitness function of each individual according to the performance evaluation result, and find out the feature subset corresponding to the optimal individual through iterative optimization; An identification module is configured to predict a to-be-identified scheme based on the obtained feature subset corresponding to the optimal individual, obtain a slope failure instability mode prediction result, and update the slope failure instability database according to the slope failure instability mode prediction result.

[0064] It should be noted that the specific implementation mode of the slope failure instability mode intelligent identification system in the embodiment of the present application is similar to that of the slope failure instability mode intelligent identification method in the embodiment of the present application. For details, please refer to the description in the method part. In order to reduce redundancy, this part will not be repeated here.

[0065] Embodiment three The embodiment provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the steps of the slope failure instability mode intelligent identification method.

[0066] Embodiment four The embodiment provides a computer device, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor realizes the steps of the slope failure instability mode intelligent identification method when executing the program.

[0067] Those skilled in the art will appreciate that embodiments of the application can be readily used as software, hardware, or a combination of software and hardware. In a software embodiment, the methods can be tangibly embodied in a computer-readable storage medium having stored

[0068] The present application is described in reference to the flowchart and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart and / or block diagrams, and combinations of blocks in the flowchart and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0069] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0070] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0071] It is to be understood that the above-described embodiments can be implemented by hardware, software, or a combination of software and hardware. In a software embodiment, the methods can be tangibly embodied in a computer-readable storage medium having stored computer-readable instructions. The computer-readable instructions, when executed by a computer, can cause the computer to perform the methods described above. The computer-readable storage medium can be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.

[0072] The above descriptions are only the preferred embodiments of the present application, not intended to limit the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall fall within the scope of the present application.

Claims

1. A method for intelligent identification of slope failure and instability modes, characterized in that, Includes the following steps: Based on the acquired historical engineering feature information, a slope failure and instability database was established; Based on the established parametric coding rules, the engineering feature information and slope failure and instability modes in the slope failure and instability database are parametrically coded. Using the parametrically encoded engineering feature information from the slope failure and instability database as input parameters and the slope failure and instability mode as output parameters, multiple parameter combination schemes are constructed. Each individual in the genetic algorithm is represented as a subset of input features. The performance of the intelligent identification model for slope failure and instability modes trained under different parameter combinations is evaluated. The fitness function of each individual is determined based on the performance evaluation results. The feature subset corresponding to the optimal individual is found through iterative optimization. Based on the feature subset corresponding to the obtained optimal individual, the identification scheme is predicted to obtain the slope failure and instability mode prediction result, and the slope failure and instability database is updated according to the slope failure and instability mode prediction result.

2. The intelligent identification method for slope failure and instability modes as described in claim 1, characterized in that, Based on the slope material type, sliding surface shape, and cause, slope failure and instability modes are classified into four categories, including circular arc non-structural surface sliding, structural surface sliding, toppling rupture surface sliding, and buckling rupture surface sliding. Among them, toppling rupture surface sliding includes two types of toppling rupture surface sliding occurring on slopes with anti-dip structures and toppling rupture surface sliding occurring on slopes with dip-to-slope structures.

3. The intelligent identification method for slope failure and instability modes as described in claim 1, characterized in that, The parameterization processing of engineering characteristic information and slope failure and instability modes in the slope failure and instability database includes: For numerical data, assign values ​​directly; For non-numerical data, one-hot encoding is used to convert it into numerical data. In one-hot encoding, each category is converted into a binary vector. The length of this vector is the same as the number of categories, and only one element is 1, while the rest are 0.

4. The intelligent identification method for slope failure and instability modes as described in claim 1, characterized in that, Using parametrically coded engineering feature information from the slope failure and instability database as input parameters and slope failure and instability modes as output parameters, various parameter combination schemes are constructed, including: single-parameter combination, two-parameter combination, three-parameter combination, and four-parameter combination. The single-parameter combination includes slope topography, slope material, internal rock strata structure, and slope rock strata dip angle; the two-parameter combination includes slope topography and slope material, slope topography and internal rock strata structure, and slope topography and slope rock strata. The three-parameter combination includes slope topography, slope material and internal rock strata structure; slope topography, slope material and slope rock strata dip angle; slope topography, slope internal rock strata structure and slope rock strata dip angle; slope material, slope internal rock strata structure and slope rock strata dip angle; and the four-parameter combination includes slope topography, slope material, slope internal rock strata structure and slope rock strata dip angle.

5. The intelligent identification method for slope failure and instability modes as described in claim 1, characterized in that, Representing each individual of the genetic algorithm as a subset of input features includes using binary encoding for each individual to represent a specific subset of input features; each individual corresponds to an n-bit binary string, where each bit represents whether a feature is selected, with 1 indicating selection and 0 indicating non-selection.

6. The intelligent identification method for slope failure and instability modes as described in claim 1, characterized in that, The intelligent identification model for slope failure and instability modes adopts a generalized regression neural network.

7. The intelligent identification method for slope failure and instability modes as described in claim 1, characterized in that, The fitness value function for each individual is expressed as: Fitness function = α × Classification accuracy + β × Geological consistency score + γ × Feature redundancy, where α, β, and γ represent weight coefficients.

8. An intelligent identification system for slope failure and instability modes, characterized in that, include: The database creation module is used to establish a slope failure and instability database based on the acquired historical relevant engineering feature information; The parametric coding module is used to perform parametric coding processing on the engineering feature information and slope failure and instability modes in the slope failure and instability database based on the set parametric coding rules. The iterative optimization module takes the parametrically encoded engineering feature information from the slope failure and instability database as input parameters and the slope failure and instability mode as output parameters to construct multiple parameter combination schemes. Each individual of the genetic algorithm is represented as a subset of input features. The performance of the intelligent identification model of slope failure and instability mode trained under different parameter combination schemes is evaluated. The fitness function of each individual is determined based on the performance evaluation results. The optimal feature subset corresponding to the best individual is found through iterative optimization. The identification module is used to predict the identification scheme based on the feature subset corresponding to the obtained optimal individual, obtain the slope failure and instability mode prediction result, and update the slope failure and instability database according to the slope failure and instability mode prediction result.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the intelligent identification method for slope failure and instability modes as described in any one of claims 1-7.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the intelligent identification method for slope failure and instability modes as described in any one of claims 1-7.

Images