Method for determining phreatic line of stepped bank slope group, program product and electronic device

Abstract

The present disclosure provides a method for determining the phreatic line of a stepped reservoir bank slope group, a program product and an electronic device, and relates to the technical field of engineering data processing. The method comprises: obtaining first monitoring data of a target slope and second monitoring data of a related associated slope; fusing the first monitoring data and the second monitoring data according to a time-varying weight of hydraulic association to obtain a hydraulic association fusion feature; processing the hydraulic association fusion feature of the target slope and the hydraulic association fusion feature of a plurality of sample data by using a kernel extreme learning machine to obtain a phreatic line elevation inversion value; constructing a hydraulic association strength field based on the distance between slopes and the fluctuation rate of the reservoir water level; fusing the phreatic line elevation inversion value and the key parameters in the first monitoring data to obtain a group perception fusion feature; and obtaining a phreatic line elevation optimization value according to the group perception fusion feature, the hydraulic association strength field and the phreatic line elevation inversion value. The present disclosure improves the accuracy and stability of determining the phreatic line of the stepped reservoir bank slope group.

Description

Technical Field

[0001] This disclosure relates to the field of engineering data processing technology, and more specifically, to methods, program products, and electronic equipment for determining the seepage line of a cascade reservoir bank slope group. Background Technology

[0002] With the transformation of hydropower development towards a basin-wide cascade model, the large-scale construction of pumped storage power stations has fundamentally changed the hydrological conditions of reservoir areas. Reservoirs periodically impound and release water according to the scheduling plan, and reservoir water level fluctuations gradually change from low-frequency, gradual variations to high-frequency, rapid changes. The soil and rock masses in the slope drawdown zone repeatedly undergo saturation-drying cycles. During this process, the dynamic changes in the phreatic line are a core indicator reflecting the evolution of the seepage field within the slope and controlling slope stability; its accurate determination is crucial for project safety.

[0003] In related technologies, most methods for determining the phreatic line of a slope rely on monitoring data of a single slope, without considering the hydraulic correlation between slopes, and cannot reflect the coupling effect between slope groups. This results in insufficient accuracy and low reliability of the phreatic line determination results, making it difficult to adapt to the complex operating conditions of cascade reservoir bank slope groups. Summary of the Invention

[0004] This disclosure provides a method, program product, and electronic equipment for determining the phreatic line of a cascade reservoir bank slope group, so as to improve the accuracy of the phreatic line determination results to at least a certain extent.

[0005] According to a first aspect of this disclosure, a method for determining the phreatic line of a cascade reservoir bank slope group is provided. The method includes: acquiring first monitoring data of a target slope and second monitoring data of associated slopes related to the target slope for a target slope in the cascade reservoir bank slope group; fusing the first monitoring data and the second monitoring data according to the hydraulic correlation time-varying weights of the target slope and the associated slopes to obtain hydraulic correlation fusion features of the target slope; and processing the hydraulic correlation fusion features of the target slope and the hydraulic correlation fusion features corresponding to multiple sets of sample data using a kernel limit learning machine to obtain the phreatic line elevation inversion value of the target slope. The multiple sets of sample data include: real sample data from actual monitoring of the cascade reservoir bank slope group; non-real sample data generated by inputting the real sample data into the generator; a hydraulic correlation intensity field of the cascade reservoir bank slope group is constructed based on the distance between the target slope and the associated slope and the reservoir water level rise and fall rate; a swarm intelligent perception model is used to fuse the inverted value of the seepage line elevation and the key parameters in the first monitoring data to obtain the swarm perception fusion characteristics of the target slope; and the optimized value of the seepage line elevation of the target slope at the target time is obtained based on the swarm perception fusion characteristics, the hydraulic correlation intensity field, and the inverted value of the seepage line elevation.

[0006] According to a second aspect of this disclosure, a device for determining the phreatic line of a cascade reservoir bank slope group is provided. The device includes: a monitoring data acquisition module configured to acquire first monitoring data of a target slope and second monitoring data of associated slopes related to the target slope for a target slope in the cascade reservoir bank slope group; a hydraulic correlation fusion feature determination module configured to fuse the first monitoring data and the second monitoring data according to the hydraulic correlation time-varying weights of the target slope and the associated slopes to obtain hydraulic correlation fusion features of the target slope; and a phreatic line inversion module configured to process the hydraulic correlation fusion features of the target slope and the hydraulic correlation fusion features corresponding to multiple sets of sample data using a kernel extreme learning machine to obtain the phreatic line elevation inversion value of the target slope; the multiple... The sample data includes: real sample data from actual monitoring of the cascade reservoir bank slope group, and non-real sample data generated by inputting the real sample data into the generator; a hydraulic correlation intensity field construction module, configured to construct the hydraulic correlation intensity field of the cascade reservoir bank slope group based on the distance between the target slope and the associated slope and the reservoir water level rise and fall rate; a swarm perception fusion feature determination module, configured to fuse the inverted elevation value of the seepage line and the key parameters in the first monitoring data using a swarm intelligent perception model to obtain the swarm perception fusion feature of the target slope; and a seepage line optimization module, configured to obtain the optimized seepage line elevation value of the target slope at the target time based on the swarm perception fusion feature, the hydraulic correlation intensity field, and the inverted elevation value of the seepage line.

[0007] According to a third aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the method of the first aspect described above and possible implementations thereof.

[0008] According to a fourth aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the method of the first aspect and possible implementations thereof by executing the executable instructions.

[0009] The technical solution disclosed herein has the following beneficial effects: For target slopes in a cascade reservoir bank slope group, based on single slope monitoring, a second monitoring data from associated slopes is introduced, and hydraulic correlation time-varying weights are combined to fuse the monitoring data and obtain hydraulic correlation fusion characteristics. A kernel limit learning machine and a swarm intelligence perception model are used to invert the seepage line elevation and fuse key parameters to obtain swarm perception fusion characteristics. A hydraulic correlation intensity field is constructed, and the optimized seepage line elevation value is obtained by combining the swarm perception fusion characteristics with the seepage line elevation inversion value. This overcomes the limitations of single slope analysis, realizes the quantitative characterization of the spatial correlation and dynamic water level influence of the slope group, and upgrades from single-point monitoring to swarm collaborative analysis. This improves the accuracy, stability, and engineering applicability of seepage line determination for the cascade reservoir bank slope group, providing reliable data support for slope stability assessment and safety control. Attached Figure Description

[0010] Figure 1 A flowchart illustrating a method for determining the seepage line of a cascade reservoir bank slope group according to one embodiment of this disclosure is shown. Figure 2 This diagram illustrates a flowchart of obtaining first parameter evolution information in one embodiment of the present disclosure; Figure 3 This diagram illustrates the overall architecture of a method according to one embodiment of the present disclosure. Figure 4 This diagram shows a comparison of inversion results in one embodiment of the present disclosure; Figure 5 This diagram shows a comparison between the predicted dynamic seepage line value and the dynamic early warning threshold of a slope in one embodiment of the present disclosure. Figure 6 A schematic diagram showing the slope distribution and single slope early warning level in one embodiment of this disclosure is provided. Figure 7 This diagram shows a device for determining the seepage line of a cascade reservoir bank slope group according to one embodiment of the present disclosure; Figure 8 A schematic diagram of an electronic device according to one embodiment of the present disclosure is shown. Detailed Implementation

[0011] Exemplary embodiments of this disclosure will be described more fully below with reference to the accompanying drawings.

[0012] The accompanying drawings are schematic illustrations of this disclosure and are not necessarily drawn to scale. Some block diagrams shown in the drawings may be functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in hardware modules or integrated circuits, or in networks, processors, or microcontrollers. Implementations can be carried out in various forms and should not be construed as limited to the examples set forth herein. The features, structures, or characteristics described in this disclosure can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a full description of the technical solutions of this disclosure. However, those skilled in the art will recognize that one or more specific details may be omitted when implementing the technical solutions provided in this disclosure, or other methods, components, apparatuses, steps, etc., may be used to replace one or more specific details.

[0013] With the transformation of hydropower development towards a basin-wide cascade model, the large-scale construction of pumped storage power stations has fundamentally changed the hydrological conditions of reservoir areas. Reservoirs periodically impound and release water according to the scheduling plan, and reservoir water level fluctuations gradually change from low-frequency, gradual variations to high-frequency, rapid changes. The soil and rock masses in the slope drawdown zone repeatedly undergo saturation-drying cycles. During this process, the dynamic changes in the phreatic line are a core indicator reflecting the evolution of the seepage field within the slope and controlling slope stability; its accurate determination is crucial for project safety.

[0014] The fundamental difference between a cascade reservoir bank slope group and a single slope lies in the hydraulic connection and risk transmission mechanism between upstream and downstream slopes in a cascade reservoir bank slope group. Landslide surges generated after upstream slope instability propagate downstream, altering the hydrological boundary conditions of the downstream slope and potentially inducing its failure. Changes in downstream reservoir water level affect the phreatic line elevation at the toe of the upstream slope through the backwater backwater effect, forming a reverse coupling. This bidirectional mechanism of forward linkage and reverse coupling makes the phreatic line analysis of a cascade reservoir bank slope group far more complex than that of a single slope.

[0015] In related technologies, most methods for determining the phreatic line of slopes rely on monitoring data of a single slope, analyzing each slope independently and ignoring the coupling effect of upstream and downstream hydraulic connections on the phreatic line. Risk transmission is simplified to a series or parallel model. This approach fails to reflect the coupling effects between slope groups, resulting in insufficient accuracy and low reliability in the determination of the phreatic line, making it difficult to adapt to the complex operating conditions of cascade reservoir bank slope groups. Furthermore, these technologies also suffer from the following problems: Traditional methods for monitoring seepage lines primarily rely on point sensors such as piezometers and permeameters, which lack sufficient spatial resolution to accurately depict the continuous distribution of the seepage line. Furthermore, methods that determine the seepage line through discrete point interpolation suffer from limited interpolation accuracy when spatial variability exists in the soil and rock mass, leading to significant accumulation of indirect measurement errors. Moreover, the insufficient spatial resolution of the monitoring data results in a small-sample learning problem; machine learning models used for seepage line analysis are prone to overfitting in small sample conditions, negatively impacting the analysis results.

[0016] In view of one or more of the above-mentioned problems, this disclosure provides a method for determining the seepage line of a cascade reservoir bank slope group. Figure 1 An exemplary flowchart for determining the phreatic line of a cascade reservoir bank slope group is shown, including the following steps: S110: For the target slope in the cascade reservoir bank slope group, acquire the first monitoring data of the target slope and the second monitoring data of the associated slopes related to the target slope. S120, the first monitoring data and the second monitoring data are fused according to the hydraulic correlation time-varying weight of the target slope and the associated slope to obtain the hydraulic correlation fusion characteristics of the target slope; S130 uses a kernel limit learning machine to process the hydraulic correlation fusion characteristics of the target slope and the hydraulic correlation fusion characteristics corresponding to multiple sets of sample data to obtain the inversion value of the seepage line elevation of the target slope; the multiple sets of sample data include: real sample data from actual monitoring of the cascade reservoir bank slope group, and non-real sample data generated by inputting the real sample data into the generator. S140, based on the spacing between the target slope and the associated slopes and the rate of rise and fall of the reservoir water level, constructs the hydraulic correlation intensity field of the cascade reservoir bank slope group; S150 uses a swarm intelligence perception model to fuse the inverted elevation values ​​of the seepage line and key parameters in the first monitoring data to obtain the swarm perception fusion characteristics of the target slope. S160, based on the ensemble perception fusion characteristics, hydraulic correlation intensity field, and inversion value of the seepage line elevation, the optimized value of the seepage line elevation of the target slope at the target time is obtained.

[0017] Based on the above method, for target slopes in a cascade reservoir bank slope group, in addition to single-slope monitoring, second monitoring data from associated slopes are introduced, and hydraulic correlation time-varying weights are combined to fuse the monitoring data and obtain hydraulic correlation fusion characteristics. Kernel Limit Learning Machine and swarm intelligence sensing model are used to invert the seepage line elevation and fuse key parameters to obtain swarm perception fusion characteristics. A hydraulic correlation intensity field is constructed, and the optimized seepage line elevation value is obtained by combining the swarm perception fusion characteristics with the seepage line elevation inversion value. This overcomes the limitations of single-slope analysis, realizes the quantitative characterization of the spatial correlation and dynamic water level influence of slope groups, and upgrades from single-point monitoring to swarm collaborative analysis. This improves the accuracy, stability, and engineering applicability of seepage line determination for cascade reservoir bank slope groups, providing reliable data support for slope stability assessment and safety control.

[0018] The following describes, in conjunction with one or more embodiments and related accompanying drawings, Figure 1 Each step in the process will be explained in detail.

[0019] refer to Figure 1 In step S110, for the target slope in the cascade reservoir bank slope group, the first monitoring data of the target slope and the second monitoring data of the associated slopes related to the target slope are obtained.

[0020] The target slope is the slope in the cascade reservoir bank slope group that requires determination of its phreatic line, and can be any slope in the cascade reservoir bank slope group. For example, if it is necessary to determine the phreatic line for each slope in the cascade reservoir bank slope group, each slope can be taken as the target slope, and the optimized value of the phreatic line elevation of the target slope can be determined using the method of this disclosure. The associated slope is the slope in the cascade reservoir bank slope group that has a hydraulic connection with the target slope. The target slope and the associated slope can influence each other through wave propagation, backwater backwater, etc. The first monitoring data is the monitoring data collected by monitoring equipment deployed at the target slope site, and the second monitoring data is the monitoring data collected by monitoring equipment deployed at the associated slope site. This disclosure does not limit the specific parameter types of the first monitoring data and the second monitoring data. For example, the first monitoring data and the second monitoring data can both include measured values ​​of key parameters such as phreatic line elevation, surface displacement, pore water pressure, and reservoir water level.

[0021] In one implementation, a hydraulic correlation mapping model for upstream and downstream slopes can be established to quantify the coupling effect of surge propagation and backwater backwater. The formula for the hydraulic correlation mapping model is as follows: (1) Among them, h j ( t )for t Downstream of the moment j Elevation of slope seepage line (in meters), hi ( t -Δ t )for t -Δ t upstream of time i The elevation of the slope seepage line, Δ t It could be the propagation time of the surge; w ij ( t ) is a slope i With slope j The hydraulic correlation time-varying weight is a parameter that quantifies the strength and degree of hydraulic coupling between two slopes; h w,j ( t )for t Time slope j Reservoir water level (in meters) in the relevant section; Δh ij The rise in the phreatic line caused by backwater (in meters) can be determined by the reservoir water level fluctuation rate and the slope spacing. The hydraulic time-varying weights of any two slopes can be calculated using formula (1). For the target slope, the hydraulic time-varying weights of each of the other slopes relative to the target slope are calculated, and slopes with hydraulic time-varying weights reaching a specific value (such as...) are selected. w ij ( t The slope with a value of ≥0.2 is used as the associated slope of the target slope.

[0022] In one implementation, monitoring data is collected for the target slope within a cascade reservoir bank slope group. Monitoring equipment such as piezometers, displacement gauges, and water level gauges are deployed at key locations on the target slope, including the slope body and toe. Information such as the phreatic line elevation, surface displacement, pore water pressure, and reservoir water level is continuously collected at a fixed time frequency to form the first monitoring data. Simultaneously, information such as the phreatic line elevation, surface displacement, pore water pressure, and reservoir water level are collected on related slopes that are hydraulically connected to the target slope to form the second monitoring data. This ensures the temporal continuity and spatial coverage of the data collection process, guaranteeing that the data accurately reflects the slope condition and related effects.

[0023] Continue to refer to Figure 1 In step S120, the first monitoring data and the second monitoring data are fused according to the hydraulic correlation time-varying weight of the target slope and the associated slope to obtain the hydraulic correlation fusion characteristics of the target slope.

[0024] Among them, the hydraulic correlation fusion feature considers the slope characteristics under the hydraulic correlation effect of the associated slopes. In one embodiment, based on the time-varying weight of the hydraulic correlation between the target slope and the associated slopes, the first monitoring data and the second monitoring data of each associated slope are weighted and fused. During the fusion process, the dominant role of the target slope's own monitoring data is retained, while the state information of the associated slopes is reasonably integrated according to the strength of the correlation. Finally, a hydraulic correlation fusion feature with unified dimensions and comprehensive information is formed. This feature includes both the state information of the target slope itself and the hydraulic coupling influence information of the associated slopes, which can comprehensively reflect the overall seepage state of the slope group.

[0025] In one implementation, the above-mentioned fusion of the first monitoring data and the second monitoring data based on the hydraulic correlation time-varying weights of the target slope and the associated slope to obtain the hydraulic correlation fusion characteristics of the target slope includes the following process: The hydraulic correlation time-varying weights are normalized using the following formula to obtain the hydraulic correlation fusion weights: (2) in, i Indicates the target slope as the first i A slope, j Indicates the first j A related slope; ω ij For the first i The slope and the first j Hydraulic correlation fusion weights for each associated slope; M The number of associated slopes. In formula (2), for w ij ( t The weights are normalized to obtain the hydraulic correlation fusion weights.

[0026] Then, the first and second monitoring data are fused using the following formula to obtain the hydraulic correlation fusion characteristics: (3) in, X i,fusion ( t )for t Time of the first i Hydraulic correlation and fusion characteristics of individual slopes; ω 0 is the first i The weight of the self-monitoring data of each slope can be determined based on experience or specific needs, such as taking 0.6; X i,self ( t )for t Time of the first i The first monitoring data for each slope; Xj,rel for t Time of the first j The second monitoring data for the associated slope.

[0027] In one implementation, some parameters from the first and second monitoring data can be fused, while retaining the unfused parameters from the first monitoring data, to obtain hydraulic correlation fusion characteristics. For example, formula (3) can be used to analyze the surface displacement in the first monitoring data. u i ( t ), pore water pressure p i ( t ), surface displacement in the second monitoring data u j ( t ), pore water pressure p j ( t The data will be merged, and the reservoir water level h from the first monitoring data will be retained. w,i ( t ), thus obtaining hydraulic correlation and fusion characteristics. X i,fusion ( t ).

[0028] In one implementation, a Hydraulic Correlation Attention (HCA) module can be constructed. This module can perform the calculation process of the above formulas (2) and (3), calculate the hydraulic correlation fusion weight, and obtain the hydraulic correlation fusion feature by fusing the first monitoring data and the second monitoring data.

[0029] Continue to refer to Figure 1 In step S130, a kernel limit learning machine is used to process the hydraulic correlation fusion features of the target slope and the hydraulic correlation fusion features corresponding to multiple sets of sample data to obtain the inversion value of the seepage line elevation of the target slope. The multiple sets of sample data include: real sample data from actual monitoring of the cascade reservoir bank slope group, and non-real sample data generated by the generator.

[0030] The Kernel Extreme Learning Machine (KELM) is a fitting model constructed using kernel functions, used for the inversion calculation of the seepage line elevation. Real sample data consists of slope monitoring data collected from on-site monitoring equipment. Non-real sample data is synthetic monitoring data generated by a generator based on the real sample data. For example, for each set of multiple sample data, the hydraulic correlation fusion characteristics corresponding to each set of sample data are obtained by fusing the monitoring data of the slope itself and related slopes. The hydraulic correlation fusion characteristics of the target slope and the hydraulic correlation fusion characteristics corresponding to each set of sample data are input into the KELM to output the inverted seepage line elevation value of the target slope.

[0031] The process of generating non-real sample data is illustrated below.

[0032] In one implementation, reference Figure 2 As shown, the method also includes the following steps: S210: Input real sample data into the generator to be trained to generate sample data to be evaluated; S220, determine the generated loss based on the time-series mean square error between the real sample data and the sample data to be evaluated; S230, determine the physical constraint loss based on the degree to which the sample data to be evaluated conforms to the boundary constraints of the wetting line elevation, the consistency constraints of seepage, and the constraints of hydraulic correlation propagation; S240, determine the value of the first loss function based on the generation loss and the physical constraint loss; S250, the generator is updated based on the first loss function value.

[0033] For example, a Hydrological ConditionsConstraining Generation Module (HCGM) can be constructed. This module, combined with hydrophysical constraints, controls the quality of generated non-realistic sample data, achieving high-quality enhancement of soil and rock time-series data (such as seepage line elevation, surface displacement, and pore water pressure) under small sample conditions. HCGM can employ a structure of temporal convolution and hydrological constraint gating, ensuring that the generated samples strictly adhere to seepage physics and hydraulic correlation characteristics, avoiding the problem of virtual samples from traditional generation models being disconnected from engineering reality. HCGM can include a generator, a hydrological constraint module, and a quality control module. These will be described separately below.

[0034] (a) Generator

[0035] The generator takes real sample data as input, which can be a small-sample real time-series monitoring data sequence. It selects three types of parameters—wetting line elevation, surface displacement, and pore water pressure—to construct the real sample data, as shown in the following formula: (4) in, X real For real sample data, its dimension can be 3×T; h real,i ( x , y , t )for t Time of the first i A slope ( x , y ) Measured elevation of the phreatic line at the location; u real,i ( t )for t Time of the first i Measured surface displacement values ​​of each slope; p real,i ( t )for t Time of the first i Measured values ​​of pore water pressure on each slope; T The length of the time series data is in hours (h), which can be determined based on the monitoring frequency and monitoring period. The generator can output a synthesized time series data vector through time series convolutional layers, batch normalization layers, and activation functions, combined with hydrological constraint gating, as shown in the following formula: (5) in, Represents the tensor product. X gen For synthesized sample data (dimension can be 3×) T ), such as the sample data to be evaluated or non-real sample data; and X real correspond, X gen It can include the synthetic wetting line elevation h gen,i ( x , y , t ), Synthetic surface displacement u gen,i ( t Synthetic pore water pressure p gen,i ( t ); G ( ) is the generator mapping function; θ G This is the set of generator parameters, including temporal convolution kernel size, number of convolutional layers, etc. This is a temporal convolution operation used to capture temporal correlation features in data; BN( ) is a batch normalization operation used to accelerate model training and avoid gradient vanishing; AF( ) is the activation function, which can be the ReLU (Rectified Linear Unit) activation function, and the formula is AF( x )=max(0, x ); g ( t ) is the hydrological constraint gating function constructed in the embodiments of this disclosure.

[0036] (II) Hydrological Constraint Module

[0037] Hydrological constraint gating function g ( t The following formula is used to dynamically constrain the physical rationality of generated samples: (6) in, W g The gated weight matrix can be 3×3 in dimension. b g This is a bias term and can be set to 0.1. X ( t ) represents the input monitoring data, such as X real or X gen ;AF( ) is the activation function, which can be the sigmoid activation function, used to map the gate value to the interval [0,1] to realize dynamic constraints on the generated samples. The closer the gate value is to 1, the more the generated samples conform to the hydrophysical laws.

[0038] To ensure that the synthetic sample data conforms to engineering physical laws, three core hydrophysical constraints can be set to replace the traditional Maximum Mean Discrepancy (MMD) quality control. The quantification formula is as follows: ① Boundary constraints on the elevation of the seepage line: The composite elevation of the seepage line shall not exceed the reservoir water level at the corresponding time, and shall not be less than 0 (based on the slope toe). The formula is as follows: (7) Among them, h gen,i ( x , y , t () represents the elevation of the synthetic wetting line, h w,j ( t )for t Time of the firsti Synthetic samples that do not meet the constraint regarding the reservoir water level of the slope section are directly eliminated.

[0039] ② Seepage Consistency Constraint: The synthetic phreatic line data must satisfy the unsaturated-saturated seepage control equation for a single slope, which is as follows: (8) Where h is the elevation of the seepage line (in meters). k x , k y , k z They are respectively x、y、z The permeability coefficient (in m / d) in the direction of entrapment is related to the elevation h of the entrapment line; S S Water storage ratio (unit: 1 / m) reflects the water storage capacity of the rock and soil mass; t For time. The seepage control equation is used to characterize the fundamental laws governing the spatiotemporal evolution of the wetting line (referred to as the seepage law). The degree to which the synthesized sample data satisfies the seepage law is quantified using the residual method, as shown in the following formula: (9) Where Res is the residual of the seepage control equation. The smaller the residual, the more consistent the synthetic sample data is with the seepage law. For example, a residual threshold (such as 0.02) can be set according to experience or specific needs. When Res≤0.02, the synthetic sample data is determined to meet the seepage consistency constraint.

[0040] ③ Hydraulic correlation propagation constraints: The synthetic sample data must satisfy the hydraulic correlation law between upstream and downstream slopes to ensure spatial correlation, as shown in the following formula: (10) Where, Δh threshold The allowable deviation for the propagation of the immersion line (which can be determined based on experience, historical statistical results, etc., for example, 0.05m after calibration), h gen,j ( t )for t Downstream of the moment j The composite phreatic line elevation of each slope, h gen,i ( t -Δ t )for t -Δ t upstream of time i The composite phreatic line elevation of each slope. w ij ( t ) is a slope i With slope jThe hydraulic correlation time-varying weights. Synthetic sample data that do not satisfy formula (10) can be removed.

[0041] (III) Quality Control Module

[0042] The quality control module ensures the authenticity and physical plausibility of the synthesized sample data through a first loss function, and can update the generator based on the first loss function to achieve model training. In one implementation, the value of the first loss function is calculated using the following formula: (11) (12) (13) in, L 1 represents the value of the first loss function. L gen To generate loss, L phy Loss due to physical constraints; ρ This is a weighting coefficient, which can be set based on experience or specific needs, such as 0.6, to prioritize physical rationality. T The timing length can be compared with... X real The timing lengths are consistent; D Based on the dimensions of real sample data, it can correspond to three types of key monitoring parameters (wetting line elevation, surface displacement, and pore water pressure). X gen,d ( t )for t Time of the first d The sample data to be evaluated corresponding to the class parameters; X real,d ( t )for t Time of the first Real sample data for class parameters; Res is the residual of the seepage control equation; h gen,i ( x , y , t )for t Time of the first i A slope ( x , y The sample data to be evaluated corresponding to the elevation of the phreatic line at the location; h w,j ( t )for t Time of the first i The water level of the reservoir section where the slope is located; h gen,j ( t )for t Time of the first jThe sample data to be evaluated corresponding to the phreatic line elevation of each slope, h gen,i ( t -Δ t )for t -Δ t Time of the first i The sample data to be evaluated corresponding to the phreatic line elevation of each slope; Δh threshold Allowable deviation for immersion line propagation.

[0043] For example, during the model training phase, the generated synthetic sample data is used as the sample data to be evaluated. The first loss function value can be calculated based on the sample data to be evaluated, and the generator can be updated accordingly. During the model application phase after the generator training is completed, the generated synthetic sample data is used as non-real sample data among the aforementioned multiple sets of sample data. Alternatively, during the model application phase, the generator first generates the sample data to be evaluated. The hydrological constraint module and the quality control module evaluate the sample data to be evaluated. If the evaluation result is satisfactory (e.g., the hydrological constraints are met, and the first loss function value is lower than the loss threshold), then the sample data to be evaluated is used as non-real sample data among the aforementioned multiple sets of sample data. If the evaluation result is unsatisfactory, the sample data to be evaluated can be regenerated, or the generator can be updated.

[0044] Based on the above HCGM, high-quality synthetic sample data can be generated, providing a sufficient data foundation for the inversion calculation of the seepage line elevation, which is conducive to obtaining accurate optimized seepage line elevation results in the end. In one implementation, a kernel limit learning machine can be used to calculate the inverted value of the seepage line elevation of the target slope using the following formula: (14) in, i Indicates the target slope as the first i One slope; h inv,i ( x , y , t )for t Time of the first i A slope ( x,y The inverted elevation value of the phreatic line at the location; N The number of multiple sets of sample data; β k The weights for the kernel-based extreme learning machine; kernel ( , Let ) be the Gaussian kernel function, and its formula is: σ is the kernel width, which can be set according to experience and specific needs, such as σ=0.8; X i,fusion ( t ) is the firsti Hydraulic correlation and fusion characteristics of individual slopes; X k,fusion ( t ) is the first k Hydraulic correlation fusion features corresponding to the group of sample data.

[0045] Since the hydraulic correlation fusion feature inherently contains spatial location information, the Gaussian kernel function can realize the mapping from location features to spatial field. Therefore, formula (14) can output the first... i Inverted values ​​of the seepage line elevation at different locations on the slope.

[0046] In one implementation, the inversion objective function is constructed as follows: (15) in, j Indicates the first j One associated slope; h inv,i,k For the first k The first group of sample data corresponding to the i The inversion value of the seepage line elevation of each slope, h real,i,k For the first k The first group of sample data corresponding to the i The elevation monitoring value of the seepage line of each slope, h inv,j,k For the first k The first group of sample data corresponding to the j Inverted values ​​of the seepage line elevation of each associated slope; β HC These are the weighting coefficients; w ij ( t )for i The slope and the first j Hydraulic time-varying weights of associated slopes; M This represents the number of associated slopes. In this inversion objective function, the first term is the inversion relative error term, ensuring consistency between the inverted values ​​and the measured values. The second term is the hydraulic correlation consistency constraint term. β HC To balance inversion accuracy and correlation rationality, its value can be set based on experience or specific needs, such as 0.3. It can be minimized... ε The kernel limit learning machine is optimized. Thus, the numerical accuracy of the inversion of the wetting line is ensured by the relative error term, and the physical rationality of the inversion results is ensured by the hydraulic correlation consistency constraint term. The model weights are optimized with the minimization of the inversion objective function as the guide, achieving a synergistic improvement in accuracy and rationality, avoiding high-precision but physically incompatible inversion results, and enhancing the reliability of the model engineering.

[0047] In one implementation, the constraint terms in the inversion objective function can be examined. If the value is ≤0.02, it indicates that the inversion results conform to the hydraulic correlation law; otherwise, the fusion weights should be readjusted. ω ij Then, the inversion process is performed again to obtain the inverted elevation value of the seepage line.

[0048] Continue to refer to Figure 1 In step S140, a hydraulic correlation intensity field of the cascade reservoir bank slope group is constructed based on the distance between the target slope and the associated slope and the reservoir water level rise and fall rate.

[0049] The hydraulic correlation intensity field is a parameter field characterizing the spatial distribution and intensity of hydraulic correlations among slope groups. It can quantify the degree of hydraulic influence of one slope on another by a single point value. For example, using the spatial distance between the target slope and the associated slopes and the reservoir water level fluctuation rate as core parameters, and combining the time-varying weights of hydraulic correlations for comprehensive calculation, the weakening effect of spatial distance on hydraulic correlations is represented by an exponential decay term, and the influence of dynamic water level on slope correlations is represented by a water level fluctuation rate term. Finally, a hydraulic correlation intensity field that can cover the entire cascade reservoir bank slope group is formed to clearly characterize the spatial correlation and dynamic coupling characteristics between slopes.

[0050] In one implementation, the hydraulic correlation intensity field of the cascade reservoir bank slope group can be constructed using the following formula: (16) in, i Indicates the target slope as the first i A slope, j Indicates the first j One associated slope; HIF ij ( x , y , t )for t Time of the first i A slope ( x,y The position is affected by the first j Hydraulic correlation intensity field values ​​of the influence of each associated slope; μ The attenuation coefficient of the hydraulic correlation intensity field refers to a fixed coefficient used to characterize the attenuation of hydraulic correlation intensity as the slope spacing increases. It reflects the weakening effect of spatial distance on hydraulic correlation. For example, it can be taken as... μ =0.001; L ij For the first i The slope and the first j The spacing of the associated slopes can be specifically the first... i A slope ( x,y ) position and number j The spacing of the associated slopes, by substituting ( x,y) Position coordinates or corresponding spacing, calculate ( x,y Hydraulic correlation intensity field value at location ) V w ( t )for t The rate of rise and fall of the reservoir water level at any given moment; V w,max This represents the maximum rate of rise and fall of the reservoir water level. w ij ( t )for i The slope and the first j The hydraulic correlation time-varying weights of the associated slopes ensure that the hydraulic correlation intensity field can reflect the comprehensive impact of spatial distance and water level fluctuations on the slope group correlation; exp( ) is an exponential function. Formula (16) comprehensively considers three core factors: the time-varying characteristics of hydraulic correlation, the spatial distance attenuation effect, and the dynamic influence of reservoir water level. The constructed hydraulic correlation intensity field can truly and comprehensively reflect the hydraulic coupling law of the slope group in the cascade reservoir area. The field value is clearly quantified and has a clear physical meaning. It can be directly used for collaborative prediction and risk analysis, and provides accurate correlation basis for the optimization of the seepage line of the slope group.

[0051] Continue to refer to Figure 1 In step S150, a swarm intelligent perception model is used to fuse the inverted elevation value of the seepage line and the key parameters in the first monitoring data to obtain the swarm perception fusion characteristics of the target slope.

[0052] Key parameters are those parameters in the first monitoring data that have a significant controlling effect on the dynamic changes of the seepage line, such as surface displacement, pore water pressure, and reservoir water level fluctuation rate. The swarm intelligent perception model is an intelligent fusion model used to weightedly fuse multi-source monitoring parameters and inversion results to form predictive input features. The swarm perception fusion feature is obtained by fusing the seepage line elevation inversion value and key parameters through the swarm intelligent perception model, and is used for seepage line optimization prediction. For example, the swarm intelligent perception model is used to weightedly fuse the seepage line elevation inversion value and key parameters in the first monitoring data, assigning fusion weights according to the importance of each parameter's influence on the seepage line. Information such as the seepage line elevation inversion value, surface displacement, pore water pressure, and reservoir water level fluctuation rate are integrated into a unified swarm perception fusion feature. This feature integrates the inversion results and real-time monitoring information, providing high-quality input for subsequent optimization calculations.

[0053] In one implementation, key parameters include surface displacement, pore water pressure, and reservoir water level fluctuation rate. A swarm intelligence sensing model can be used, and the swarm sensing fusion features can be obtained through the following formula: (17) in, iIndicates the target slope as the first i A slope; X i,pred ( t )for t Time of the first i The collective perception fusion characteristics of individual slopes; the 1. the 2. the 3. the 4 represents the weighting for swarm intelligence sensing data fusion; h inv,i ( x , y , t )for t Time of the first i A slope ( x , y The inverted elevation value of the phreatic line at the location; p i ( t )for t Time of the first i Pore ​​water pressure on a slope; u i ( t )for t Time of the first i Surface displacement of the slope; V w ( t )for t The rate of rise and fall of reservoir water level at any given time. Formula (17) ensures that the fused information is highly correlated with the infiltration line by selecting key parameters in a targeted manner; it highlights the influence of core parameters and improves the efficiency of feature expression by reasonably allocating the fusion weight of the group perception; it achieves purification and integration of multi-source information through weighted fusion, eliminates redundant noise, and obtains complete and representative second feature information, which significantly improves the accuracy and stability of subsequent infiltration line optimization calculation.

[0054] Continue to refer to Figure 1 In step S160, the optimized value of the seepage line elevation of the target slope at the target time is obtained based on the swarm perception fusion characteristics, the hydraulic correlation intensity field, and the inversion value of the seepage line elevation.

[0055] The optimized seepage line elevation is a precise result calculated collaboratively using swarm perception fusion features, hydraulic correlation intensity field, and seepage line elevation inversion value. The target time can be any time, such as a future time obtained by overlaying the prediction step size onto the current time. The optimized seepage line elevation at the target time can be a relatively accurate predicted value. For example, using swarm perception fusion features as the basic prediction term, combined with the slope group correlation effect reflected by the hydraulic correlation intensity field, and introducing the seepage line elevation inversion value as an accuracy constraint, the optimized seepage line elevation of the target slope at the target time is obtained through collaborative calculation. The calculation process fully considers the influence of related slopes, achieving a combination of single slope monitoring and swarm collaborative analysis, ultimately obtaining a precise and reliable seepage line determination result.

[0056] In one implementation, the optimized value of the seepage line elevation of the target slope at the target time can be obtained by the following formula: (18) Among them, with t+ Δ t The time indicates the target time; for t+ Δ t Time of the first i The optimized value of the phreatic line elevation of each slope, and may include t+ Δ t Time of the first i A slope ( x,y Optimized value of the phreatic line elevation at the location; α 1. α 2 represents the balancing weights, which respectively control the correction intensity of the collective intelligence collaborative fusion characteristics and the hydraulic correlation transmission. δ The weights are adjusted to control the proportion of the prediction error correction term. For example, α 1. α 2. δ The values ​​are 0.5, 0.4, and 0.1 respectively. for t+ Δ t Time of the first j Optimized values ​​for the seepage line elevation of each associated slope; for t Time of the first i Optimized values ​​for the seepage line elevation of each slope; M The number of associated slopes; h inv,i ( x , y , t )for t Time of the first i A slope ( x,yThe elevation inversion value of the seepage line at the location. Formula (18) ensures the dominant role of the target slope's own state through the basic prediction term, incorporates the group coupling effect through the hydraulic correlation synergy term, and eliminates the prediction error through the dynamic correction term. The triple constraint synergy overcomes the defect of ignoring the group correlation in single slope prediction. The calculation results are accurate, stable, and time-series continuous, which greatly improves the engineering applicability of seepage line prediction and provides a reliable basis for safety control.

[0057] In one implementation, a TG-HGCP (Targeted Gradient Hydraulic Correlation Group Intelligence Collaborative Prediction Method) model can be constructed, and the optimized value of the saturation line elevation can be output by executing formula (18).

[0058] In one implementation, the prediction accuracy of the TG-HGCP model can be evaluated using the mean absolute error (MAE) and root mean square error (RMSE), as shown in the following formula: (20) ;(twenty one) in, For the first k Sample t+ Δ t Optimized value of the immersion line elevation at any given time; h real,i,k ( t+ Δ t ) is the first k Sample t+ Δ t Measured elevation of the seepage line at any given time; N To predict the total number of samples, smaller MAE and RMSE indicate higher prediction accuracy. Based on actual engineering needs, a target of MAE ≤ 0.1m and RMSE ≤ 0.15m can be established. If this target is not achieved, one or more of the above models can be further updated, optimized, or adjusted. ω ij After obtaining the parameters, the method is re-executed. This further ensures that the output of high-quality wetting line determination results is obtained.

[0059] In one implementation, the method further includes: The maximum elevation of the seepage line of each slope in the cascade reservoir bank slope group is obtained, and the dynamic warning threshold of each slope is determined based on the maximum elevation of the seepage line and the warning threshold correction coefficient of each slope. Using the current time plus the predicted step size as the target time, obtain the optimized value of the seepage line elevation for each slope at the target time; The warning level for each slope is determined based on the relationship between the optimized value of the seepage line elevation at the target time and the dynamic warning threshold.

[0060] For example, the historical maximum value h of the phreatic line elevation for each slope is collected. max,i Based on the characteristics of the slope's soil and rock mass, a correction coefficient for the warning threshold is determined for each slope. λ i (Can be within the range of 0.85~0.95); Calculate the dynamic early warning threshold h for each slope. th,i ( t )= λ i h max,i Extract each slope from the output of the TG-HGCP method. t +Δ t Optimized value of immersion line elevation at any time Based on the single slope warning level classification standard, the warning level of each slope is determined, and targeted prevention and control suggestions are given for different warning levels, as shown in Table 1 below.

[0061] Table 1 Classification of Early Warning Levels for Single Slopes

[0062] In one implementation, the method further includes: The percentage of slopes with preset warning levels is determined based on the warning level of each slope. Based on the proportion and the average instability correlation of the cascade reservoir bank slope group, the comprehensive index of the group coordinated early warning of the cascade reservoir bank slope group is determined, and the group coordinated early warning level of the cascade reservoir bank slope group is determined based on the comprehensive index of the group coordinated early warning of the cascade reservoir bank slope group.

[0063] For example, by combining the proportion of single-slope warning levels with the correlation with instability, the group coordinated warning level can be quantitatively determined, as shown in the following formula: ;(twenty two) ;(twenty three) in, γ ij The slope group instability correlation degree is used to quantitatively describe the degree of cascading instability correlation among slopes; p j ( t )for t Time slope j Pore ​​water pressure (unit: kPa). p max h represents the maximum pore water pressure (determined by statistical analysis of monitoring data). w,j ( t -Δt )for t -Δ t upstream of time i The water level (in meters) of each reservoir section, in h i ( t -Δ t )for t -Δ t upstream of time i The elevation of the slope seepage line. γ ij The value range is [0,1]; The average instability correlation of the slope group. W warn The comprehensive index for group collaborative early warning can be set to a value within [0,1] based on experience or specific needs. ω rate The weighting of the early warning level for a single slope can be set based on experience or specific needs, such as 0.6. R warn The percentage of slopes with preset warning levels (such as yellow and red warning levels) among all slopes, with a value of [0,1]; ω γ The weight for the degree of instability can be set based on experience or specific needs, such as 0.4.

[0064] according to W warn The values ​​are used to determine the collaborative early warning level for the entire reservoir area's slope group, and targeted prevention and control suggestions are given for different early warning levels, as shown in the table below: Table 2 Group Collaborative Early Warning Level Determination Table

[0065] In one implementation, Figure 3A schematic diagram of the overall architecture of the method is shown. This architecture includes a physical mechanism layer, a small sample data augmentation layer, a hydraulic correlation inversion layer, a swarm intelligence collaborative optimization layer, and a hierarchical early warning decision layer. Different layers are used to execute different stages of the method. In the physical mechanism layer, the unsaturated-saturated seepage control equation for a single slope is established, a hydraulic correlation mapping model for upstream and downstream slopes is established, and the instability correlation degree of the slope group is established. In the small sample data augmentation layer, qualified synthetic sample data is generated through a generator, a hydrological constraint module, and a quality control module. In the hydraulic correlation inversion layer, the inverted value of the seepage line elevation is output through a hydraulic correlation feature fusion module, KELM, and an inversion objective function. In the swarm intelligence collaborative optimization layer, a hydraulic correlation intensity field is constructed, multi-source data fusion is performed using swarm intelligence perception, a TG-HGCP model is constructed, and evaluation indicators are designed. The optimized value of the seepage line elevation is output through the TG-HGCP model. In the hierarchical early warning decision layer, the early warning level of a single slope and the early warning level of the group collaborative slopes are determined, and finally, the collaborative early warning results and prevention and control suggestions for the entire reservoir bank slope group are output.

[0066] The embodiments of this disclosure will be further illustrated and verified through examples below.

[0067] A pumped storage power station has 13 main slopes (numbered S1) in its cascade reservoir area. S13), slope height 18 42m. The reservoir water level exhibits typical high-frequency fluctuations, with an average of 3 fluctuations per day. 4 times, maximum fluctuation rate 0.5 m / h, during the flood season (6 In September, the reservoir water level fluctuated significantly, and fissure water infiltrated rapidly in the plateau area, making the spatiotemporal evolution of the slope phreatic line more complex and easily triggering a chain of landslide risks.

[0068] The monitoring system for the cascade reservoir bank slopes was deployed as follows: one piezometer was installed at the top, middle, and toe of each slope (a total of 36 piezometers) to collect measured values ​​of the seepage line elevation; surface displacement monitors and pore water pressure sensors were deployed, with a monitoring frequency of 1 hour per measurement; reservoir water level and slope spatial coordinate data were collected simultaneously to form a complete monitoring dataset. Due to the complex geographical environment of the reservoir area and the high monitoring cost, only 42 sets of real sample data were obtained during the actual monitoring period (the actual sample size was less than 50 sets, which meets the small sample scenario). The sample time span was 30 days, covering both normal water level fluctuations and heavy rainfall conditions, for model training, validation, and performance comparison. The 42 sets of real sample data were divided into a training set (containing 29 sets of real sample data) and a test set (containing 13 sets of real sample data) in a 7:3 ratio.

[0069] Based on 29 sets of real sample data from the training set, 50 sets of initial synthetic sample data were generated using the HCGM data augmentation method. After screening with hydrophysical constraints, 40 sets of qualified synthetic sample data were obtained, with a qualified screening rate of 80%, meeting the requirements for small sample data augmentation. Two schemes were used to train the HCA-KELM (Hydraulic Correlation Attention-Kernel Extreme Learning Machine) model: one using "original real sample data (29 sets)" and the other using "original real sample data + HCGM synthetic sample data (29 + 40 = 69 sets)". For both schemes, 13 sets of real sample data from the test set were used to verify the inversion accuracy. The comparison results are as follows: Table 3 Comparison of Inversion Accuracy Results

[0070] As shown in the table above, after HCGM data augmentation, the average inversion error of the HCA-KELM model decreased from 5.8% to 4.0%, and the accuracy improved by 31.03%, successfully meeting the preset target of inversion error ≤ 5.5%. This verifies the effectiveness of the HCGM small sample data augmentation method in scenarios with no preprocessing of the original small sample. It can effectively supplement the sample size, improve the model's generalization ability, and solve the problem of low inversion accuracy caused by insufficient original small sample data.

[0071] Based on 69 sets of sample data augmented by HCGM, the HCA-KELM inversion method was implemented, with kernel width σ=0.8, regularization parameter C=10, and fusion weights. ω 0=0.6, number of associated slopes M Depend on w ij ( t The selection criteria were ≥0.2 (each slope was associated with 3-4 adjacent slopes). The inversion effect was verified using 13 sets of real sample data from the test set. The results are as follows: Figure 4 As shown, the average relative error of the 13 sets of real sample data is only 1.75%, and the maximum relative error is 2.98% (corresponding to sample number S10), both of which are far below the target of ≤5.5% for the inversion error preset in this invention. This shows that the HCA-KELM inversion method can achieve high-precision inversion of the wetting line and is suitable for scenarios where raw data from engineering sites are directly applied.

[0072] Based on the HCA-KELM inversion values ​​and 42 sets of real sample data, the TG-HGCP collaborative prediction method was implemented, and the model parameters were set as follows: μ =0.001、 α 1 = 0.5 α 2 = 0.4 δ =0.2, prediction step size Δ t =1h, MAE and RMSE were used to evaluate the prediction accuracy. The TG-HGCP method of this disclosure, the traditional KELM method, and the independent prediction method (which predicts each slope in the cascade reservoir area separately without considering the hydraulic correlation and synergistic effect between slope groups) were compared. The results are shown in the table below: Table 4 Comparison of Collaborative Prediction Results

[0073] The prediction results show that the TG-HGCP method has an MAE of 0.087 and an RMSE of 0.132, both of which meet the preset accuracy targets. The collaborative prediction accuracy is 22.3% higher than that of independent prediction, reaching the improvement requirement of ≥20%. However, when the traditional KELM method and the independent prediction method use the same real sample data, the MAE or RMSE exceeds the preset threshold, resulting in insufficient prediction accuracy.

[0074] Figure 5 The table shows a comparison between the predicted dynamic seepage line values ​​(i.e., the optimized seepage line elevation values ​​output by the TG-HGCP method) and the dynamic early warning thresholds for 13 slopes. Combined with Table 1, the system can accurately identify slope S7 as an early warning level (core risk source), slopes S3 and S4 as levels of concern, and slopes S1, S2, S5, S6, and S8-S13 as normal levels. Figure 6 A schematic diagram showing the slope distribution and single-slope early warning levels is presented. This result is consistent with on-site seepage anomaly monitoring records, fully demonstrating the accuracy and reliability of the single-slope early warning determination.

[0075] Based on the individual slope warning levels and other relevant parameters of 13 slopes along the cascade reservoir banks, a comprehensive evaluation of the overall risk status of the slope group was conducted. The results showed that among the 13 monitored slopes, only S7 reached the warning level (yellow), S3 and S4 were at the concern level (blue), and the remaining slopes were at the normal level (green). The proportion of warning-level slopes was 7.69%, and the average instability correlation of the slope group was 0.63. Based on this, the group warning index was calculated to be 0.402, and according to the judgment criteria, it was comprehensively judged as the group concern level (blue warning). This result indicates that although the overall project is currently still within a controllable range, S7, as the core risk source, has formed a hydraulic correlation effect on the surrounding slopes. Through the chain-like risk transmission path of S7→S4→S3, the risk status of secondary slopes has gradually increased. The comprehensive judgment of the group collaborative warning not only objectively reflects the existence of individual point risks but also effectively characterizes the overall risk situation of the cascade reservoir area slope group, providing a scientific basis for key source control, strengthening related slopes, and coordinating overall safety in project implementation.

[0076] As can be seen from the above, the method of this disclosure is suitable for the operation and maintenance scenario of cascade reservoir bank engineering, and can effectively solve technical bottlenecks such as monitoring small sample raw data, complex hydraulic correlations, and difficulty in preventing and controlling chain risks. It has good engineering practicality and promotion value.

[0077] In summary, the embodiments disclosed herein have the following beneficial effects: By using the small-sample data augmentation method of the Hydrological Conditions Constraint Generation Module (HCGM), the amount of monitoring data was effectively expanded and the quality was improved, resulting in a 31.03% increase in inversion accuracy. It has the advantage of strong physical rationality of the synthesized samples and solves the problem of the disconnect between the samples generated by traditional data augmentation methods and engineering reality.

[0078] By using the combined approach of the upstream and downstream slope hydraulic correlation mapping model and the hydraulic correlation feature fusion module (HCA), quantitative modeling and collaborative inversion of the spatial correlation of slope groups are realized, reducing inversion errors and having the advantages of making full use of the spatial redundancy of monitoring information and high inversion accuracy.

[0079] By constructing the TG-HGCP method for the cascade reservoir bank slope group, high-precision joint calculation of the seepage line of the slope group was achieved. The collaborative optimization accuracy was improved by 22.3% compared with the traditional independent prediction method. It has the advantages of novel model structure, sufficient fusion of multi-source data and high optimization accuracy.

[0080] By combining the dynamic early warning threshold of a single slope with the comprehensive index of group collaborative early warning, a hierarchical early warning mechanism has been established, which has achieved a leap from single-point monitoring and independent early warning to group collaboration and chain prevention and control. It has the advantages of a complete early warning system, comprehensive risk identification, and highly targeted prevention and control recommendations.

[0081] By deeply integrating physical mechanisms with data-driven approaches, the physical interpretability and engineering credibility of the model output results are improved, effectively solving the black box problem of pure data-driven models. It has the advantages of clear theoretical basis and high engineering acceptability.

[0082] This disclosure also provides a device for determining the seepage line of a tiered reservoir bank slope group. (Reference) Figure 7 As shown, the seepage line determination device 700 for the cascade reservoir bank slope group includes: The monitoring data acquisition module 710 is configured to acquire first monitoring data of the target slope and second monitoring data of the associated slopes related to the target slope in the cascade reservoir bank slope group. The hydraulic correlation fusion feature determination module 720 is configured to fuse the first monitoring data and the second monitoring data according to the hydraulic correlation time-varying weight of the target slope and the associated slope to obtain the hydraulic correlation fusion feature of the target slope; The seepage line inversion module 730 is configured to use a kernel limit learning machine to process the hydraulic correlation fusion features of the target slope and the hydraulic correlation fusion features corresponding to multiple sets of sample data to obtain the seepage line elevation inversion value of the target slope; the multiple sets of sample data include: real sample data from actual monitoring of the cascade reservoir bank slope group, and non-real sample data generated by the generator from the input of the real sample data. The hydraulic correlation intensity field construction module 740 is configured to construct the hydraulic correlation intensity field of the cascade reservoir bank slope group based on the distance between the target slope and the associated slope and the reservoir water level rise and fall rate; The swarm perception fusion feature determination module 750 is configured to use a swarm intelligent perception model to fuse the inverted elevation value of the seepage line and the key parameters in the first monitoring data to obtain the swarm perception fusion features of the target slope. The seepage line optimization module 760 is configured to obtain the optimized seepage line elevation value of the target slope at the target time based on the swarm perception fusion features, the hydraulic correlation intensity field, and the seepage line elevation inversion value.

[0083] In one embodiment, the device is further configured to: The real sample data is input into the generator to be trained to generate sample data to be evaluated. The generation loss is determined based on the time-series mean square error between the real sample data and the sample data to be evaluated; The physical constraint loss is determined based on the degree to which the sample data to be evaluated conforms to the wetting line elevation boundary constraints, seepage consistency constraints, and hydraulic correlation propagation constraints. The value of the first loss function is determined based on the generation loss and the physical constraint loss; The generator is updated based on the first loss function value.

[0084] In one implementation, the first loss function value is calculated using the following formula: ; ; ; in, L 1 represents the value of the first loss function. L gen To generate loss, L phy Loss due to physical constraints; ρ These are the weighting coefficients; T The timing length; D (the dimension of the real sample data). X gen,d (t )for t Time of the first d The sample data to be evaluated corresponding to the class parameters; X real,d ( t )for t Time of the first Real sample data for class parameters; Res is the residual of the seepage control equation; h gen,i ( x , y , t )for t Time of the first i A slope ( x , y The sample data to be evaluated corresponding to the elevation of the phreatic line at the location; h w,j ( t )for t Time of the first i The water level of the reservoir section where the slope is located; h gen,j ( t )for t Time of the first j The sample data to be evaluated corresponding to the phreatic line elevation of each slope, h gen,i ( t -Δ t )for t -Δ t Time of the first i The sample data to be evaluated corresponding to the phreatic line elevation of each slope; Δh threshold Allowable deviation for immersion line propagation.

[0085] In one embodiment, the step of fusing the first monitoring data and the second monitoring data according to the hydraulic correlation time-varying weights of the target slope and the associated slope to obtain the hydraulic correlation fusion characteristics of the target slope includes: The hydraulic correlation time-varying weights are normalized using the following formula to obtain the hydraulic correlation fusion weights: ; in, i The target slope is indicated as the first... i A slope, j Indicates the first j A related slope; ω ij For the first i The slope and the first j Hydraulic correlation fusion weights for each associated slope; M The number of the associated slopes; The hydraulic correlation fusion feature is obtained by fusing the first monitoring data and the second monitoring data using the following formula: ; in, X i,fusion ( t )for t Time of the first i The hydraulic correlation and fusion characteristics of the slope; ω 0 is the first i Weighting of the self-monitoring data of each slope; X i,self ( t )for t Time of the first i The first monitoring data for each slope; X j,rel for t Time of the first j The second monitoring data for each associated slope.

[0086] In one embodiment, the step of using a kernel limit learning machine to process the hydraulic correlation fusion features of the target slope and the hydraulic correlation fusion features corresponding to multiple sets of sample data to obtain the inversion value of the seepage line elevation of the target slope includes: Using the aforementioned kernel limit learning machine, the inverted value of the seepage line elevation of the target slope is calculated using the following formula: ; in, i The target slope is indicated as the first... i One slope; h inv,i ( x , y , t )for t Time of the first i A slope ( x,y The inverted elevation value of the phreatic line at the location; N The number of the multiple sets of sample data; β k The weights of the kernel extreme learning machine; kernel ( , ) is the Gaussian kernel function; X i,fusion ( t ) is the first i Hydraulic correlation and fusion characteristics of individual slopes; X k,fusion ( t ) is the first k Hydraulic correlation fusion features corresponding to the group of sample data.

[0087] In one embodiment, the device is further configured to: The inversion objective function is constructed as follows: ; in, j Indicates the first j One associated slope; h inv,i,k For the first k The first group of sample data corresponding to the i The inversion value of the seepage line elevation of each slope, h real,i,k For the first k The first group of sample data corresponding to the i The elevation monitoring value of the seepage line of each slope, h inv,j,k For the first k The first group of sample data corresponding to the j Inverted values ​​of the seepage line elevation of each associated slope; β HC These are the weighting coefficients; w ij ( t )for i The slope and the first j Hydraulic time-varying weights of associated slopes; M The number of the associated slopes; By minimizing ε Optimize the kernel extreme learning machine.

[0088] In one embodiment, constructing the hydraulic correlation intensity field of the cascade reservoir bank slope group based on the distance between the target slope and the associated slope and the reservoir water level fluctuation rate includes: The hydraulic correlation intensity field of the cascade reservoir bank slope group is constructed using the following formula: ; in, i The target slope is indicated as the first... i A slope, j Indicates the first j One associated slope; HIF ij ( x , y , t )for t Time of the first i A slope ( x,y The position is affected by the first j Hydraulic correlation intensity field values ​​of the influence of each associated slope; μ The attenuation coefficient of the hydraulic correlation intensity field; L ij For the first i The slope and the first j The spacing between the associated slopes; V w ( t )fort The rate of rise and fall of the reservoir water level at any given moment; V w,max This represents the maximum rate of rise and fall of the reservoir water level. w ij ( t )for i The slope and the first j The hydraulic time-varying weights of the associated slopes; exp( ) is an exponential function.

[0089] In one embodiment, the key parameters include surface displacement, pore water pressure, and reservoir water level fluctuation rate; the fusion of the inverted elevation value of the seepage line and the key parameters in the first monitoring data using a swarm intelligent perception model to obtain the swarm perception fusion characteristics of the target slope includes: Using the aforementioned swarm intelligence perception model, the swarm perception fusion features are obtained through the following formula: ; in, i The target slope is indicated as the first... i A slope; X i,pred ( t )for t Time of the first i The group perception fusion features of the individual slopes; the 1. the 2. the 3. the 4 represents the weighting for swarm intelligence sensing data fusion; h inv,i ( x , y , t )for t Time of the first i A slope ( x , y The inverted elevation value of the phreatic line at the location; p i ( t )for t Time of the first i Pore ​​water pressure on a slope; u i ( t )for t Time of the first i Surface displacement of the slope; V w ( t )for t The rate of rise and fall of the reservoir water level at any given time.

[0090] In one implementation, obtaining the optimized value of the seepage line elevation of the target slope at the target time based on the swarm perception fusion features, the hydraulic correlation intensity field, and the seepage line elevation inversion value includes: The optimized value of the seepage line elevation of the target slope at the target time is obtained by the following formula: ; Among them, with t+ Δ t The time represents the target time; for t+ Δ t Time of the first i Optimized values ​​for the seepage line elevation of each slope; α 1. α 2 represents the balancing weight. δ To adjust the weights; for t+ Δ t Time of the first j Optimized values ​​for the seepage line elevation of each associated slope; for t Time of the first i Optimized values ​​for the seepage line elevation of each slope; M h is the number of associated slopes. inv,i ( x , y , t )for t Time of the first i A slope ( x,y The elevation inversion value of the phreatic line at the location.

[0091] In one embodiment, the device is further configured to: The maximum elevation of the seepage line of each slope in the cascade reservoir bank slope group is obtained, and the dynamic early warning threshold of each slope is determined based on the maximum elevation of the seepage line and the early warning threshold correction coefficient of each slope. Using the current time plus the prediction step size as the target time, obtain the optimized value of the seepage line elevation for each slope at the target time; The warning level for each slope is determined based on the relationship between the optimized value of the seepage line elevation of each slope at the target time and the dynamic warning threshold.

[0092] In one embodiment, the device is further configured to: The percentage of slopes with preset warning levels is determined based on the warning level of each slope. Based on the stated proportion and the average instability correlation of the cascade reservoir bank slope group, a comprehensive index for group-wide coordinated early warning of the cascade reservoir bank slope group is determined, and the group-wide coordinated early warning level of the cascade reservoir bank slope group is determined based on the comprehensive index for group-wide coordinated early warning of the cascade reservoir bank slope group.

[0093] The specific details of each part of the above-mentioned device have been described in detail in the method section of the implementation plan. For any undisclosed details, please refer to the implementation plan of the method section, and therefore will not be repeated here.

[0094] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to exemplary embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0095] This disclosure also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the method steps of various exemplary embodiments of this disclosure.

[0096] In one implementation, the computer program product can be a tangible product, such as a computer-readable storage medium storing a computer program. The readable storage medium can be based on electrical, magnetic, optical, electromagnetic, infrared, or other signals, and includes, but is not limited to: Random Access Memory (RAM), Read-Only Memory (ROM), magnetic tape, floppy disk, flash memory, Hard Disk Drive (HDD), Solid State Disk (SSD), etc. For example, the computer program product can be a non-volatile storage medium storing a computer program, such as read-only memory, NAND flash memory, etc.

[0097] In one implementation, the computer program product can be an intangible product. For example, the computer program product can be a virtual digital product, such as an executable file or installation package containing a computer program.

[0098] Computer program code can be written in one or more programming languages. Examples of programming languages ​​include C, Java, and C++. Program code can execute entirely on the user's computing device, partially on the user's computing device, or as a standalone software package. It can also execute partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, such as a Local Area Network (LAN) or a Wide Area Network (WAN), or it can be connected to an external computing device (e.g., via an internet connection provided by a mobile network operator).

[0099] Computer programs can be carried or transmitted via signals such as electrical, magnetic, optical, electromagnetic, and infrared rays. Electronic devices can convert the signals carrying computer programs into digital signals, thereby running the computer programs. When a computer program runs on an electronic device, its code is used to cause the electronic device to execute (more specifically, to be executed by the processor of the electronic device) the method steps of various embodiments of this disclosure, such as... Figure 1 The method and steps.

[0100] Implementing the above method steps through a computer program achieves the following technical effects: For target slopes in a cascade reservoir bank slope group, based on single-slope monitoring, second monitoring data from associated slopes is introduced, and hydraulic correlation time-varying weights are combined to fuse the monitoring data and obtain hydraulic correlation fusion characteristics. A kernel limit learning machine and a swarm intelligence perception model are used to invert the seepage line elevation and fuse key parameters to obtain swarm perception fusion characteristics. A hydraulic correlation intensity field is constructed, and the optimized seepage line elevation value is obtained by combining the swarm perception fusion characteristics with the seepage line elevation inversion value. This overcomes the limitations of single-slope analysis, realizes the quantitative characterization of the spatial correlation and dynamic water level influence of the slope group, and represents an upgrade from single-point monitoring to swarm collaborative analysis. It improves the accuracy, stability, and engineering applicability of seepage line determination for cascade reservoir bank slope groups, providing reliable data support for slope stability assessment and safety control.

[0101] This disclosure also provides an electronic device. The electronic device includes a processor and a memory. The memory stores executable instructions for the processor, such as computer programs. The processor executes the executable instructions to perform the method steps of various exemplary embodiments of this disclosure.

[0102] The following is for reference. Figure 8 The electronic device is illustrated by way of a general-purpose computing device. It should be understood that... Figure 8 The electronic device 800 shown is merely an example and should not be construed as limiting the functionality or scope of this disclosure.

[0103] like Figure 8 As shown, the electronic device 800 may include: a processor 810, a memory 820, a bus 830, an I / O (input / output) interface 840, and a network adapter 850.

[0104] The memory 820 may include volatile memory, such as RAM 821 and cache unit 822, and may also include non-volatile memory, such as ROM 823. The memory 820 may also include one or more program modules 824, including but not limited to: an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. For example, program module 824 may include the modules described above.

[0105] The processor 810 may include one or more processing units, such as an AP (Application Processor), a modem processor, a GPU (Graphics Processing Unit), an ISP (Image Signal Processor), a controller, an encoder, a decoder, a DSP (Digital Signal Processor), a baseband processor, and / or an NPU (Neural-Network Processing Unit).

[0106] The processor 810 can be used to execute executable instructions stored in the memory 820 to perform method steps of various embodiments of this disclosure, such as... Figure 1 The method and steps.

[0107] By executing the above method steps through processor 810, the following technical effects are achieved: For target slopes in a cascade reservoir bank slope group, based on single-slope monitoring, second monitoring data from associated slopes is introduced, and hydraulic correlation time-varying weights are combined to fuse the monitoring data and obtain hydraulic correlation fusion characteristics. A kernel limit learning machine and a swarm intelligence perception model are used to invert the seepage line elevation and fuse key parameters to obtain swarm perception fusion characteristics. A hydraulic correlation intensity field is constructed, and the optimized seepage line elevation value is obtained by combining the swarm perception fusion characteristics with the seepage line elevation inversion value. This overcomes the limitations of single-slope analysis, realizes the quantitative characterization of the spatial correlation and dynamic water level influence of the slope group, and represents an upgrade from single-point monitoring to swarm collaborative analysis. This improves the accuracy, stability, and engineering applicability of the seepage line determination for the cascade reservoir bank slope group, providing reliable data support for slope stability assessment and safety control.

[0108] Bus 830 is used to connect different components of electronic device 800 and may include data bus, address bus and control bus.

[0109] Electronic device 800 can communicate with one or more external devices 900 (such as keyboard, mouse, external controller, etc.) through I / O interface 840.

[0110] Electronic device 800 can communicate with one or more networks via network adapter 850. For example, network adapter 850 can provide mobile communication solutions such as 3G / 4G / 5G, or wireless communication solutions such as wireless LAN, Bluetooth, and near-field communication. Network adapter 850 can communicate with other modules of electronic device 800 via bus 830.

[0111] In one embodiment, the electronic device 800 further includes a display for displaying a graphical user interface.

[0112] although Figure 8 As not shown in the diagram, other hardware and / or software modules may also be configured in the electronic device 800, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (Redundant Arrays of Independent Disks) systems, tape drives, and data backup storage systems.

[0113] As can be seen from the above, the technical solutions disclosed herein can be implemented as methods, apparatus, systems, computer program products, storage media, electronic devices, etc. Those skilled in the art will understand that various aspects of this disclosure can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software aspects. Exemplarily, these three forms can be referred to as "circuit," "module," and "system," respectively.

[0114] It should be understood that this disclosure is not limited to the specific methods, steps, or structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. Those skilled in the art will readily conceive of other embodiments based on the specific implementations provided in this disclosure. Therefore, the specific implementations provided in this disclosure are merely exemplary, and the scope and spirit of this disclosure are indicated by the claims, and should cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary technical means in the art not disclosed in this disclosure.

Claims

1. A method for determining the seepage line of a tiered reservoir bank slope group, characterized in that, The method includes: For a target slope in the cascade reservoir bank slope group, acquire the first monitoring data of the target slope and the second monitoring data of the associated slopes related to the target slope; The first monitoring data and the second monitoring data are fused based on the hydraulic correlation time-varying weights of the target slope and the associated slope to obtain the hydraulic correlation fusion characteristics of the target slope; A kernel limit learning machine is used to process the hydraulic correlation fusion features of the target slope and the hydraulic correlation fusion features corresponding to multiple sets of sample data to obtain the inversion value of the seepage line elevation of the target slope. The multiple sets of sample data include: real sample data from actual monitoring of the cascade reservoir bank slope group, and non-real sample data generated by the generator by inputting the real sample data. The hydraulic correlation intensity field of the cascade reservoir bank slope group is constructed based on the distance between the target slope and the associated slope and the reservoir water level rise and fall rate. A swarm intelligence perception model is used to fuse the inverted elevation value of the seepage line and the key parameters in the first monitoring data to obtain the swarm perception fusion characteristics of the target slope. Based on the ensemble perception fusion features, the hydraulic correlation intensity field, and the inversion value of the seepage line elevation, the optimized value of the seepage line elevation of the target slope at the target time is obtained.

2. The method according to claim 1, characterized in that, The method further includes: The real sample data is input into the generator to be trained to generate sample data to be evaluated. The generation loss is determined based on the time-series mean square error between the real sample data and the sample data to be evaluated; The physical constraint loss is determined based on the degree to which the sample data to be evaluated conforms to the wetting line elevation boundary constraints, seepage consistency constraints, and hydraulic correlation propagation constraints. The value of the first loss function is determined based on the generation loss and the physical constraint loss; The generator is updated based on the first loss function value.

3. The method according to claim 2, characterized in that, The first loss function value is calculated using the following formula: ； ； ； in, L 1 represents the value of the first loss function. L gen To generate loss, L phy Loss due to physical constraints; ρ These are the weighting coefficients; T The timing length; D The dimension of the real sample data; X gen,d ( t )for t Time of the first d The sample data to be evaluated corresponding to the class parameters; X real,d ( t )for t Time of the first Real sample data for class parameters; Res is the residual of the seepage control equation; h gen,i ( x , y , t )for t Time of the first i A slope ( x , y The sample data to be evaluated corresponding to the elevation of the phreatic line at the location; h w,j ( t )for t Time of the first i The water level of the reservoir section where the slope is located; h gen,j ( t )for t Time of the first j The sample data to be evaluated corresponding to the phreatic line elevation of each slope, h gen,i ( t -Δ t )for t -Δ t Time of the first i The sample data to be evaluated corresponding to the phreatic line elevation of each slope; Δh threshold Allowable deviation for immersion line propagation.

4. The method according to claim 1, characterized in that, The step of fusing the first monitoring data and the second monitoring data according to the hydraulic correlation time-varying weights of the target slope and the associated slope to obtain the hydraulic correlation fusion characteristics of the target slope includes: The hydraulic correlation time-varying weights are normalized using the following formula to obtain the hydraulic correlation fusion weights: ； in, i The target slope is indicated as the first... i A slope, j Indicates the first j A related slope; ω ij For the first i The slope and the first j Hydraulic correlation fusion weights for each associated slope; M The number of the associated slopes; The hydraulic correlation fusion feature is obtained by fusing the first monitoring data and the second monitoring data using the following formula: ； in, X i,fusion ( t )for t Time of the first i The hydraulic correlation and fusion characteristics of the slope; ω 0 is the first i Weighting of the self-monitoring data of each slope; X i,self ( t )for t Time of the first i The first monitoring data for each slope; X j,rel for t Time of the first j The second monitoring data for each associated slope.

5. The method according to claim 1, characterized in that, The process of using a kernel limit learning machine to process the hydraulic correlation fusion features of the target slope and the hydraulic correlation fusion features corresponding to multiple sets of sample data to obtain the inverted elevation value of the target slope includes: Using the aforementioned kernel limit learning machine, the inverted value of the seepage line elevation of the target slope is calculated using the following formula: ； in, i The target slope is indicated as the first... i One slope; h inv,i ( x , y , t )for t Time of the first i A slope ( x,y The inverted elevation value of the phreatic line at the location; N The number of the multiple sets of sample data; β k The weights of the kernel extreme learning machine; kernel ( , ) is the Gaussian kernel function; X i,fusion ( t ) is the first i Hydraulic correlation and fusion characteristics of individual slopes; X k,fusion ( t ) is the first k Hydraulic correlation fusion features corresponding to the group of sample data.

6. The method according to claim 5, characterized in that, The method further includes: The inversion objective function is constructed as follows: ； in, j Indicates the first j One associated slope; h inv,i,k For the first k The first group of sample data corresponding to the i The inversion value of the seepage line elevation of each slope, h real,i,k For the first k The first group of sample data corresponding to the i The elevation monitoring value of the seepage line of each slope, h inv,j,k For the first k The first group of sample data corresponding to the j Inverted values ​​of the seepage line elevation of each associated slope; β HC These are the weighting coefficients; w ij ( t )for i The slope and the first j Hydraulic time-varying weights of associated slopes; M The number of the associated slopes; By minimizing ε Optimize the kernel extreme learning machine.

7. The method according to claim 1, characterized in that, The construction of the hydraulic correlation intensity field of the cascade reservoir bank slope group based on the distance between the target slope and the associated slope and the reservoir water level rise and fall rate includes: The hydraulic correlation intensity field of the cascade reservoir bank slope group is constructed using the following formula: ； in, i The target slope is indicated as the first... i A slope, j Indicates the first j A related slope; HIF ij ( x , y , t )for t Time of the first i A slope ( x,y The position is affected by the first j Hydraulic correlation intensity field values ​​of the influence of each associated slope; μ The attenuation coefficient of the hydraulic correlation intensity field; L ij For the first i The slope and the first j The spacing between the associated slopes; V w ( t )for t The rate of rise and fall of the reservoir water level at any given moment; V w,max This represents the maximum rate of rise and fall of the reservoir water level. w ij ( t )for i The slope and the first j The hydraulic time-varying weights of the associated slopes; exp( ) is an exponential function.

8. The method according to claim 7, characterized in that, The key parameters include surface displacement, pore water pressure, and reservoir water level fluctuation rate; the swarm intelligence sensing model is used to fuse the inverted elevation value of the seepage line and the key parameters in the first monitoring data to obtain the swarm perception fusion characteristics of the target slope, including: Using the aforementioned swarm intelligence perception model, the swarm perception fusion features are obtained through the following formula: ； in, i The target slope is indicated as the first... i A slope; X i,pred ( t )for t Time of the first i The group perception fusion features of the individual slopes; η 1. η 2. η 3. η 4 represents the weighting for swarm intelligence sensing data fusion; h inv,i ( x , y , t )for t Time of the first i A slope ( x , y The inverted elevation value of the phreatic line at the location; p i ( t )for t Time of the first i Pore ​​water pressure on a slope; u i ( t )for t Time of the first i Surface displacement of the slope; V w ( t )for t The rate of rise and fall of the reservoir water level at any given time.

9. The method according to claim 8, characterized in that, The step of obtaining the optimized value of the seepage line elevation of the target slope at the target time based on the ensemble perception fusion features, the hydraulic correlation intensity field, and the inverted seepage line elevation value includes: The optimized value of the seepage line elevation of the target slope at the target time is obtained by the following formula: ； Among them, with t+ Δ t The time represents the target time; for t+ Δ t Time of the first i Optimized values ​​for the seepage line elevation of each slope; α 1. α 2 represents the balancing weight. δ To adjust the weights; for t+ Δ t Time of the first j Optimized values ​​for the seepage line elevation of each associated slope; for t Time of the first i Optimized values ​​for the seepage line elevation of each slope; M h is the number of associated slopes. inv,i ( x , y , t )for t Time of the first i A slope ( x,y The elevation inversion value of the phreatic line at the location.

10. The method according to any one of claims 1 to 9, characterized in that, The method further includes: The maximum elevation of the seepage line of each slope in the cascade reservoir bank slope group is obtained, and the dynamic early warning threshold of each slope is determined based on the maximum elevation of the seepage line and the early warning threshold correction coefficient of each slope. Using the current time plus the prediction step size as the target time, obtain the optimized value of the seepage line elevation for each slope at the target time; The warning level for each slope is determined based on the relationship between the optimized value of the seepage line elevation of each slope at the target time and the dynamic warning threshold.

11. The method according to claim 10, characterized in that, The method further includes: The percentage of slopes with preset warning levels is determined based on the warning level of each slope. Based on the stated proportion and the average instability correlation of the cascade reservoir bank slope group, a comprehensive index for group-wide coordinated early warning of the cascade reservoir bank slope group is determined, and the group-wide coordinated early warning level of the cascade reservoir bank slope group is determined based on the comprehensive index for group-wide coordinated early warning of the cascade reservoir bank slope group.

12. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 11.

13. An electronic device, characterized in that, include: Processor and memory; The memory is used to store executable instructions of the processor; the processor is configured to implement the method of any one of claims 1 to 11 by executing the executable instructions.

Images