Concrete faced rockfill dam multi-measuring-point deformation safety monitoring method and system based on seepage flow

Abstract

The invention discloses a concrete faced rockfill dam multi-measuring-point deformation safety monitoring method and system based on seepage flow. The method comprises the following steps: preprocessing seepage flow, related deformation measuring points and environmental quantity monitoring data; dividing a training set and a test set, and constructing a random forest model taking related deformation measurement point measurement values and the like as input and seepage flow as output; analyzing the model by using an SHAP interpretation method, and screening important factors; modeling the important factors by adopting a regression or machine learning method, constructing a multi-effect-quantity combined-driven seepage flow calculation model, and evaluating the precision of the multi-effect-quantity combined-driven seepage flow calculation model; and determining a seepage flow control threshold value, inputting multi-measurement-point deformation and environmental factors into the model, and comparing a calculated value with the threshold value to realize multi-measurement-point deformation safety fusion monitoring. The accuracy of safety monitoring of the concrete faced rockfill dam structure is improved, the physical concept is clear, implementation is easy, high reliability and high adaptability are achieved, the safety state of the dam can be mastered in time, and safe operation of the dam is guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of dam safety monitoring, and is mainly applicable to monitoring the structural safety status of concrete face rockfill dams. Specifically, it relates to a multi-point deformation safety monitoring method and system for concrete face rockfill dams based on seepage flow rate. Background Art

[0002] The working state of a concrete face rockfill dam is related to the life and property safety of people downstream and the full play of project benefits. Deformation and seepage are important characteristics reflecting the operation behavior and safety status of a concrete face rockfill dam. At present, the monitoring methods for the deformation of concrete face rockfill dams mainly monitor through single / multi-point deformation modeling and deformation monitoring indexes of measuring points. The traditional methods for formulating dam deformation monitoring indexes (such as the typical small probability method, confidence interval method, etc.) are mainly based on mathematical statistics principles, statistically analyze historical monitoring data, and calculate static or dynamic monitoring indexes for the deformation of concrete face rockfill dams. However, there are many problems in the actual application of the above methods. For example, the effectiveness of the typical small probability method depends on the probability distribution of monitoring samples and existing adverse load combinations, etc.; the confidence interval method is easily affected by the length of monitoring data and the model fitting accuracy. At the same time, the above methods for formulating deformation monitoring indexes have unclear physical concepts, only monitor based on the measured values of single measuring points, do not consider the relationship between deformation and seepage flow rate changes, and it is difficult to accurately achieve the safety monitoring of the concrete face rockfill dam structure. Summary of the Invention

[0003] Object of the Invention: In order to improve the deficiencies of existing safety monitoring methods for concrete face rockfill dams and enhance the reliability and accuracy of rockfill dam deformation monitoring, the present invention proposes a multi-point deformation safety monitoring method and system for concrete face rockfill dams based on seepage flow rate, realizes monitoring the safety state of dam deformation with a seepage flow rate control threshold, has the advantages of clear physical concept, easy to implement, high reliability, strong adaptability, etc., and can provide a new method and approach for the safety monitoring of concrete face rockfill dam structures.

[0004] Technical Solution: A multi-point deformation safety monitoring method for a concrete face rockfill dam based on seepage flow rate includes the following steps:

[0005] Step 1: Perform data preprocessing on the seepage flow rate monitoring data, relevant deformation measuring point monitoring data, and environmental quantity monitoring data of the concrete face rockfill dam. The relevant deformation measuring points include the rockfill body, foundation, slab joint, and peripheral joint;

[0006] Step 2: Divide the data preprocessed in Step 1 into a training set and a test set. In the training set, select the measured values of relevant deformation measuring points, water level factor, temperature factor, rainfall factor, and aging factor as model inputs, and use the seepage flow rate of the rockfill dam as the model output to construct a random forest model;

[0007] Step 3: Analyze the random forest model constructed in Step 2 based on the SHAP interpretation method. By comparing the influence degrees of various input factors on the seepage flow of the dam, screen out the important factors that meet the specified conditions;

[0008] Step 4: Use the regression analysis method or machine learning method to conduct modeling analysis on the important factors screened out in Step 3. Combine the multi-point deformations of the rockfill body, foundation, slab joints, and peripheral joints, as well as water level, rainfall, temperature, and aging factors, to construct a seepage flow calculation model driven by multiple effect quantities, and use the test set divided in Step 2 to evaluate the model accuracy;

[0009] Step 5: Determine the seepage flow control threshold according to the historical working conditions of the concrete face rockfill dam and the historical change of the seepage flow. Input the multi-point deformations of the rockfill body, foundation, slab joints, and peripheral joints, as well as water level, rainfall, temperature, and aging factors into the seepage flow calculation model driven by multiple effect quantities constructed in Step 4. By comparing the calculated value of the model with the seepage flow control threshold, realize the integrated monitoring of the safety of multi-point deformations.

[0010] Further, in Step 1, the data preprocessing includes gross error elimination, data imputation, and benchmarking processing.

[0011] Further, in Step 2, the water level factor is selected as H1 - H0,

[0012] H represents the upstream water level, H1 is the water level on the monitoring day, H0 is the water level on the initial day of modeling, and H2, H3, H4, H5, H6 are the average water levels on the 1st day before the monitoring day, the 2nd day before, the 3rd - 4th days before, the 5th - 15th days before, and the 16th - 30th days before respectively, are the average water levels on the 1st day before the initial day of modeling, the 2nd day before, the 3rd - 4th days before, the 5th - 15th days before, and the 16th - 30th days before respectively. h represents the downstream water level, and the meanings of each factor are the same as those of the upstream water level; the temperature factor is selected as t is the cumulative number of days from the monitoring day to the initial day of modeling, and t0 is 1; the rainfall factor is selected as U1, U2, U3, U4, U5 are the rainfall amounts on the monitoring day, the 1st day before, the 2nd day before, the 3rd - 4th days before, and the average rainfall amounts on the 5th - 15th days before respectively, are the rainfall amounts on the initial day of modeling, the 1st day before, the 2nd day before, the 3rd - 4th days before, and the average rainfall amounts on the 5th - 15th days before respectively; the aging factor is selected as θ - θ0, lnθ - lnθ0, where θ is t / 100 and θ0 is t0 / 100.

[0013] Further, step 3 specifically includes: calculating the mean of the absolute values of the SHAP values of each factor under all samples, i.e., the mean absolute SHAP value. Taking the mean absolute SHAP value as the standard, evaluating the influence degree of each factor on the seepage flow of the dam through sorting and plotting analysis, selecting the factors with the mean absolute SHAP value within the specified range as important factors, and using the screened factors as the input variables of the seepage flow calculation model jointly driven by multiple effect quantities.

[0014] Further, in step 4, the seepage flow calculation model jointly driven by multiple effect quantities is expressed as:

[0015] Q = a0 + f(W) + f(T) + f(D) + f(C) + f(S)

[0016] Where: W, T, D, and C respectively represent the water level factor, temperature factor, rainfall factor, and aging factor screened in step 3; S represents the measured values of important deformation measurement points screened in step 3; a0 is a constant term; f() represents the function of seepage flow with respect to the corresponding factor.

[0017] Further, in step 4, when evaluating the model accuracy, the correlation coefficient R 2 and the mean absolute percentage error MAPE are used as evaluation indicators.

[0018] Further, in step 5, through the comparison between the model calculated value and the seepage flow control threshold, the integrated monitoring of the deformation safety of multiple measurement points is realized, including: selecting the design control value of the seepage flow of the concrete face rockfill dam or the historical extreme value of the seepage flow as the seepage safety monitoring threshold Q max If the model predicted value Q ≤ Q max in step 4, it indicates that the deformation of each measurement point of the concrete face rockfill dam is in a normal state and the structure is safe; if Q > Q max in step 4, it indicates that the deformation of some measurement points of the concrete face rockfill dam is in an abnormal state.

[0019] A multi-measurement point deformation safety monitoring system for a concrete face rockfill dam based on seepage flow includes:

[0020] A data preprocessing module for preprocessing the seepage flow monitoring data, relevant deformation measurement point monitoring data, and environmental quantity monitoring data of the concrete face rockfill dam. The relevant deformation measurement points include the rockfill body, foundation, slab joints, and peripheral joints;

[0021] A random forest model construction module for dividing the preprocessed data into a training set and a test set. In the training set, selecting the measured values of relevant deformation measurement points, water level factor, temperature factor, rainfall factor, and aging factor as the model inputs, and taking the seepage flow of the rockfill dam as the model output to construct a random forest model;

[0022] A feature screening module, which is used to analyze the constructed random forest model based on the SHAP interpretation method, and screen out important factors that meet the specified conditions by comparing the influence degrees of various input factors on the seepage flow of the dam;

[0023] A seepage flow calculation model construction module, which is used to perform modeling analysis on the screened important factors by using regression analysis methods or machine learning methods, combine the deformations of multiple measuring points of the rockfill dam, foundation, slab joints and peripheral joints, as well as water level, rainfall, temperature and aging factors, construct a seepage flow calculation model jointly driven by multiple effect quantities, and evaluate the model accuracy by using the divided test set;

[0024] A multi-measuring point deformation monitoring result output module, which is used to determine the seepage flow control threshold according to the historical working conditions of the concrete face rockfill dam and the historical change of the seepage flow, input the deformations of multiple measuring points of the rockfill dam, foundation, slab joints and peripheral joints, as well as water level, rainfall, temperature and aging factors into the seepage flow calculation model jointly driven by multiple effect quantities, and realize the integrated monitoring of the safety of multi-measuring point deformation by comparing the calculated value of the model with the seepage flow control threshold.

[0025] The present invention also provides a computer device, including: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of a method for multi-measuring point deformation safety monitoring of a concrete face rockfill dam based on seepage flow as described above are implemented.

[0026] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of a method for multi-measuring point deformation safety monitoring of a concrete face rockfill dam based on seepage flow as described above are implemented.

[0027] Beneficial effects: On the basis of monitoring data preprocessing, the present invention quantitatively analyzes the random forest model by the SHAP interpretation method, screens out important deformation and environmental quantity factors with strong correlation with the change of the seepage flow of the concrete face rockfill dam, constructs a seepage flow calculation model jointly driven by multiple effect quantities by regression analysis methods or machine learning methods, determines the mapping relationship between seepage and deformation, and monitors the deformation status of multiple measuring points of the concrete face rockfill dam through the seepage flow control threshold. This method can realize the real-time monitoring of the deformation safety of the concrete face rockfill dam based on the seepage flow, effectively avoids the limitations of traditional deformation monitoring methods, fully considers the connection between the deformation of key parts of the concrete face rockfill dam and the change of the seepage flow, further improves the accuracy of the structural safety monitoring of the concrete face rockfill dam, has clear physical concepts, is easy to implement, has high reliability and strong adaptability, and can provide useful reference for the operation management and safety status assessment of the dam. Description of the Drawings

[0028] Figure 1 It is the flow chart of the multi - measuring - point deformation safety monitoring method for concrete - faced rock - fill dams based on seepage flow in the present invention;

[0029] Figure 2 It is the average absolute SHAP value (the first 35%) of each factor after the calculation by the SHAP interpretation method in the embodiment of the present invention;

[0030] Figure 3 It is the training and prediction results of the seepage - flow calculation model jointly driven by multiple effect quantities in the embodiment of the present invention. Specific implementation manners

[0031] The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0032] The concrete - faced rock - fill dam has become one of the mainstream dam types. With the development of pumped - storage power stations, it is relatively common to use a concrete - faced rock - fill dam as the water - retaining structure for the upper reservoir. The strain soft - sensing method for the concrete slab of the rock - fill dam based on the deformation of the rock - fill body has broad application prospects. The existing monitoring methods for the deformation of concrete - faced rock - fill dams have problems of insufficient reliability, and the existing methods for formulating deformation monitoring indexes have unclear physical concepts. They only monitor based on the measured values of a single measuring point and do not consider the connection between deformation and the change of seepage flow, making it difficult to accurately achieve the safety monitoring of the structure of the concrete - faced rock - fill dam. In fact, seepage is a comprehensive reflection of the working state of the concrete - faced rock - fill dam. The change of seepage flow is related not only to environmental quantities such as water level and rainfall, but also the working states of the panel stress, slab joints, peripheral joints, and dam - foundation anti - seepage structure are important influencing factors. Among them, the panel stress and the dam - foundation anti - seepage structure, in addition to the influence of loads, are mainly related to the deformation of the dam foundation and the rock - fill body. The sudden increase in seepage flow, in addition to the change of environmental quantities such as water level, rainfall, and temperature, often means the combined effect of excessive deformation of the rock - fill body and the foundation, that is, the panel cracking or extrusion failure and the abnormal deformation of slab joints and peripheral joints caused thereby. Therefore, based on the seepage flow, the integrated monitoring of the deformation safety of the rock - fill body, foundation, slab joints, and peripheral joints can be realized. Thus, the present invention proposes a multi - measuring - point deformation safety integrated monitoring method for concrete - faced rock - fill dams based on seepage flow.

[0033] Based on the preprocessing of monitoring data, the method of the present invention uses the random forest algorithm and the SHAP (SHapley Additive exPlanations) interpretability method to screen important deformation and environmental quantity factors with strong correlation with the change of seepage flow rate. Through regression analysis or machine learning methods, a seepage flow rate calculation model jointly driven by the deformation of multiple measuring points such as the rockfill body, foundation, slab joints, and peripheral joints is constructed to determine the mapping relationship between seepage and deformation. Then, based on the seepage flow rate control threshold (design control value or historical maximum value), real-time fusion monitoring of the deformation of multiple measuring points such as the rockfill body, foundation, slab joints, and peripheral joints of the concrete face rockfill dam is realized to timely grasp the safety status of the dam and ensure the safe operation of the dam.

[0034] Referring to Figure 1 , the multi-measurement point deformation safety monitoring method for concrete face rockfill dams based on seepage flow rate proposed by the present invention includes the following steps:

[0035] Step 1: Perform data preprocessing work such as gross error rejection, data interpolation, and benchmarking on the seepage flow rate monitoring data of the concrete face rockfill dam, the monitoring data of relevant deformation measuring points, and the monitoring data of environmental quantities such as corresponding water levels and rainfall to construct model input samples. Among them, relevant deformation measuring points can consider the settlement of the internal rockfill body and measuring points with strong correlation with seepage flow rate such as the opening and closing degrees of slab joints, construction joints, and peripheral joints. For measuring points with large measured values, benchmarking can be performed.

[0036] Step 2: Divide the data preprocessed in Step 1 into a training set and a test set. Take the first 90% as the training set and the last 10% as the test set. In the training set, select the measured values of relevant deformation measuring points such as the rockfill body, foundation, slab joints, and peripheral joints, water level factors, temperature factors, rainfall factors, and time effect factors as model inputs, and take the seepage flow rate of the rockfill dam as the model output to construct a random forest model.

[0037] Among them, the upstream and downstream water levels are the direct driving factors of seepage, directly affecting the hydraulic gradient and seepage path, and causing deformation of the dam joints or cracks by changing the water pressure, thereby affecting the seepage characteristics. At the same time, due to the hysteresis of the seepage process, the influence of water level changes on seepage usually has a time delay. Therefore, considering the upstream and downstream water levels and their previous terms, the water level factors are selected as H1 - H0, h1 - h0, H represents the upstream water level, H1 is the water level on the monitoring day, H0 is the water level on the initial day of modeling, and H i-2 (i = 4 - 8) are the average water levels on the 1st day before the monitoring day, the 2nd day before, the 3rd - 4th days before, the 5th - 15th days before, and the 16th - 30th days before respectively, They are the water levels on the day before the initial modeling day, the water levels on the second day before, the average water levels from the 3rd to the 4th day before, the 5th to the 15th day before, and the 16th to the 30th day before respectively. h represents the downstream water level, and the meanings of each factor are the same as those of the upstream water level. The temperature changes of the panel concrete and the dam foundation cause the deformation of joints, cracks, and joint fissures in the dam foundation through the thermal expansion and contraction effect, thereby causing changes in seepage. For dams that have been in operation for many years, the temperatures of the dam body and the dam foundation usually show a quasi-steady state, and the temperature factor can be represented by a periodic term. Therefore, the temperature factor is selected t is the cumulative number of days from the monitoring day to the initial modeling day, and t0 is 1. During rainfall, some rainwater infiltrates into the dam body at the crest and downstream surface of the rockfill dam, causing the redistribution of the internal seepage field. At the same time, there is also a lag process in the seepage change caused by rainfall infiltration. Therefore, the rainfall factor is selected U i (i = 1 to 5) are the average rainfall amounts on the monitoring day, the day before, the second day before, the 3rd to the 4th day before, and the 5th to the 15th day before respectively. U i 0 (i = 1 to 5) are the average rainfall amounts on the initial modeling day, the day before, the second day before, the 3rd to the 4th day before, and the 5th to the 15th day before respectively. After the rockfill dam is completed and impounded, it causes changes in the structural particles of the rockfill body and the settlement of the dam body, affecting the seepage channels in the dam body. At the same time, natural blankets gradually form due to sedimentation in front of the dam. The influence of these factors on seepage has an aging process, generally with a sharp change in the initial stage and a gradually stable change in the later stage. Therefore, the aging factors θ - θ0 and lnθ - lnθ0 are selected, where θ is t / 100 and θ0 is t0 / 100. In addition, the settlement of the rockfill body will cause changes in the contact state between the concrete and the rockfill body, and affect the pore structure and crack deformation of the dam body, thereby changing the seepage path and the overall seepage field. The inter-slab joints, construction joints, and peripheral joints usually serve as the main channels for water flow in the dam, and their opening and closing states directly determine the seepage capacity and seepage path of the water flow. Therefore, the measured values of relevant deformation measurement points such as the rockfill body, foundation, inter-slab joints, and peripheral joints are selected as the model inputs

[0038] Step 3: Analyze the random forest model constructed in Step 2 based on the SHAP interpretation method. By comparing the influence degrees of each input factor on the seepage flow of the dam, screen out important factors and eliminate factors with low correlation

[0039] Specifically: Calculate the mean of the absolute values of the SHAP values of each factor under all samples, that is, the mean absolute SHAP value. Taking the mean absolute SHAP value as the standard, evaluate the influence degree of each factor on the seepage flow of the dam through sorting and plotting analysis. According to the actual engineering situation, select the factors with the mean absolute SHAP value in the top 30% - 50% as important factors, and use the screened factors as the input variables of the seepage flow calculation model driven by multiple effect quantities

[0040] The SHAP interpretability method is mainly based on game theory ideas and can effectively interpret complex machine learning black-box models such as random forests, XGBoost, and neural networks. Its core idea is to calculate the marginal contribution of input individual features to the model output. By assigning a SHAP importance value to each feature, it represents the degree of its influence on the output result. Since the SHAP value only represents the influence of individual features on the output result under a single sample, the method of the present invention uses the average absolute SHAP value of individual features under all samples as the evaluation criterion to comprehensively reflect the influence degree of each input feature on the dam seepage flow rate. For the SHAP value of a specific feature i, the calculation formula is usually expressed as:

[0041]

[0042] In the formula: N represents the set of all features; S represents any subset in the feature set N except feature i; |S| is the number of features in subset S; f S (x S ) is the predicted output of the model considering only the features in subset S; f S∪{i} (x S∪{i} ) is the predicted output of the model after adding feature i to subset S.

[0043] Then the average absolute SHAP value of feature i under all samples can be expressed as:

[0044]

[0045] In the formula: M is the total number of samples; represents the SHAP value of feature i corresponding to the mth sample.

[0046] Step 4: Use the regression analysis method or machine learning method to conduct modeling analysis on the important factors screened in step 3. Combine the deformations of multiple measuring points such as the rockfill body, foundation, slab joints, and peripheral joints, as well as influencing factors such as water level, rainfall, temperature, and aging, to construct a seepage flow rate calculation model driven by multiple effect quantities, and use the test set divided in step 2 to evaluate the model accuracy.

[0047] Among them, the seepage flow rate calculation model driven by multiple effect quantities can be expressed as:

[0048] Q = a0 + f(W) + f(T) + f(D) + f(C) + f(S) (3)

[0049] Where: W, T, D, and C respectively represent the water pressure factor, temperature factor, rainfall factor, and aging factor after screening in Step 3; S represents the measured values of important deformation measuring points after screening in Step 3; a0 is the constant term. f() represents the function of seepage flow with respect to a certain factor. For example, the water pressure factor component of seepage flow can be expressed as:

[0050]

[0051] Taking the correlation coefficient R 2 and the mean absolute percentage error MAPE as evaluation indicators, the closer the correlation coefficient R 2 is to 1 or the smaller the mean absolute percentage error MAPE, the higher the model accuracy. When the model evaluation indicators satisfy the correlation coefficient R 2 ≥0.9 or the mean absolute percentage error MAPE ≤ 5%, it is considered that the model accuracy is relatively high and can be used as the basis for the deformation safety monitoring of multiple measuring points of the concrete face rockfill dam.

[0052] The specific expressions of the correlation coefficient R 2 and the mean absolute percentage error MAPE are as follows:

[0053]

[0054] Where: n is the number of samples, y i , are respectively the measured value and predicted value of the i-th sample, is the mean value of the measured values of n samples.

[0055] Step 5: According to the historical working conditions of the concrete face rockfill dam and the historical change of seepage flow, determine the seepage flow control threshold, such as the design control value of seepage flow or the historical maximum value. Input the deformations of multiple measuring points such as the rockfill body, foundation, slab joints, and peripheral joints, as well as the influencing factors such as water level, rainfall, temperature, and aging, into the seepage flow calculation model jointly driven by multiple effect quantities constructed in Step 4. By comparing the model calculation value with the seepage flow control threshold, realize the integrated monitoring of the deformation safety of multiple measuring points.

[0056] Specifically: Select the design control value of the concrete face rockfill dam or the historical extreme value of seepage flow as the seepage safety threshold Q max . If the model calculation value Q ≤ Q max in Step 4, it indicates that the deformations of each measuring point of the concrete face rockfill dam are in a normal state and the structure is safe; if Q > Q max , it indicates that the deformations of some measuring points of the concrete face rockfill dam are in an abnormal state, and it is necessary to strengthen monitoring, inspection, and analysis, and take measures if necessary.

[0057] In order to verify the performance of the proposed method, in one embodiment, the multi-measurement point deformation safety fusion monitoring method of the concrete face rockfill dam based on seepage flow of the present invention is used to conduct a structural safety monitoring experiment on a certain concrete face rockfill dam, which specifically includes the following steps:

[0058] Step S1: Preprocess the total seepage flow monitoring data, relevant deformation monitoring data and environmental quantity monitoring data of a certain concrete face rockfill dam from January 1, 2017 to September 1, 2024. Considering the strong correlation with seepage, select the settlement of the internal rockfill body (70 pipe settlement gauges), the opening of the slab joints, the opening and closing of the construction joints (28 unidirectional joint meters, 9 crack meters and 5 bidirectional joint meters), and the opening and closing of the peripheral joints (5 three-way joint meters) as the relevant deformation monitoring data. There are 117 measuring points in total. During the preprocessing process, benchmarking processing is carried out on the measured values of some measuring points of settlement and joint meters.

[0059] Step S2: Divide the data preprocessed in Step S1 into a training set and a test set. Take the first 90% as the training set, with a total of 2520 groups of data, and the last 10% as the test set, with a total of 280 groups of data. Calculate the water level factor, temperature factor, rainfall factor and time effect factor corresponding to the monitoring date in the training set, use them and the measured values of relevant deformation measuring points as the model inputs, and use the total seepage flow monitoring value as the model output to construct a random forest model.

[0060] Step S3: Analyze the random forest model constructed in Step S2 based on the SHAP interpretation method, and calculate the contribution degree of each input factor to the seepage flow. Take the average absolute SHAP value as the evaluation standard for the influence degree of each factor on the seepage flow of the dam, and conduct sorting. The results are as Figure 2 shown. In the figure, W_1 represents H1 - H0, W_2 represents W_3 represents W_4 represents W_5 represents W_6 represents W_7 represents W_12 represents W_14 represents W_15 represents T_3 represents D_5 represents C_1 represents θ - θ0, C_2 represents lnθ - lnθ0. In this embodiment, the factors with the top 35% of the average absolute SHAP values are taken as important factors, a total of 40, which are used as the input variables of the seepage flow calculation model driven by multi-effect quantities jointly.

[0061] From the feature screening results, it can be seen that the seepage flow rate of the concrete face rockfill dam is mainly strongly correlated with the upstream water level and its previous values, the measured values of some settlement measuring points inside the rockfill body (SV01_1_37, SV01_1_25, SV01_3_13, SV01_3_3, etc.), the measured values of some crack gauges measuring points (RJ01_19_X, SJ01_3_Y, J_3, etc.), and characteristic factors such as aging.

[0062] Step S4: Use the stepwise regression method to perform modeling analysis on the important factors screened in step S3, construct a seepage flow rate calculation model jointly driven by multiple effect quantities, and input the test set divided in step S2 to evaluate the prediction accuracy of the model. The results are as Figure 3 shown. In the test section, the mean absolute percentage error (MAPE) of the model prediction results is 6.41%, and the correlation coefficient R 2 is 0.93, indicating that the model prediction accuracy is relatively high and can be used as the basis for the multi-point deformation safety monitoring of the concrete face rockfill dam.

[0063] Arrange the above model into an expression form, specifically as follows:

[0064]

[0065] In the formula: a 1i 、a 2i 、b 12 、d5, c1, c2, e i are all stepwise regression model coefficients; S i is the measured value of the important deformation measuring point screened in step S3.

[0066] Step S5: According to the previous inspection conditions of the concrete face and the change of the seepage flow rate, it can be known that the annual extreme value of the seepage flow rate during the normal operation period of the dam (that is, excluding the period when the concrete face has obvious damage) is 66.6 L / s, which is used as the seepage safety monitoring threshold. That is, the output value Q of the seepage flow rate calculation model jointly driven by multiple effect quantities constructed in step S4 needs to satisfy:

[0067]

[0068] After that, the multi-point deformation safety monitoring of the concrete face rockfill dam can be carried out according to formula (5), and the structural safety status of the concrete face rockfill dam can be monitored according to whether it meets formula (5). For example, input the important deformation measuring point and environmental quantity monitoring data of the concrete face rockfill dam on September 2, 2024 into the seepage flow rate calculation model jointly driven by multiple effect quantities constructed in step S4, and the calculated seepage flow rate value Q = 58.86 ≤ Q max , indicating that the deformation of each measuring point of the concrete face rockfill dam is in a normal state and the structure is safe.

[0069] The method of the present invention mines the deep connection between the seepage and deformation of the dam by combining the random forest and SHAP interpretation method, then establishes the mapping relationship between the seepage and deformation by using the regression analysis method or machine learning method, and finally monitors the safety state of the dam deformation with the seepage flow control threshold. It has the advantages of clear physical concept, easy implementation, high reliability, strong adaptability, etc., and can provide new methods and ways for the structural safety monitoring of concrete face rockfill dams.

[0070] Based on the same technical concept as the method embodiment, the present invention also provides a multi-point deformation safety monitoring system for a concrete face rockfill dam based on seepage flow, including:

[0071] A data preprocessing module for preprocessing the seepage flow monitoring data, relevant deformation measurement point monitoring data and environmental quantity monitoring data of the concrete face rockfill dam, where the relevant deformation measurement points include the rockfill body, foundation, slab joints and peripheral joints;

[0072] A random forest model construction module for dividing the preprocessed data into a training set and a test set. In the training set, the measured values of relevant deformation measurement points, water level factors, temperature factors, rainfall factors and aging factors are selected as model inputs, and the seepage flow of the rockfill dam is used as the model output to construct a random forest model;

[0073] A feature screening module for analyzing the constructed random forest model based on the SHAP interpretation method, and screening out important factors that meet the specified conditions by comparing the influence degrees of each input factor on the seepage flow of the dam;

[0074] A seepage flow calculation model construction module for modeling and analyzing the selected important factors by using the regression analysis method or machine learning method, combining the multi-point deformation of the rockfill body, foundation, slab joints and peripheral joints, as well as water level, rainfall, temperature and aging factors, to construct a seepage flow calculation model driven by multiple effect quantities, and evaluating the model accuracy by using the divided test set;

[0075] A multi-point deformation monitoring result output module for determining the seepage flow control threshold according to the historical working conditions of the concrete face rockfill dam and the historical change of the seepage flow, inputting the multi-point deformation of the rockfill body, foundation, slab joints and peripheral joints, as well as water level, rainfall, temperature and aging factors into the seepage flow calculation model driven by multiple effect quantities, and realizing the integrated monitoring of multi-point deformation safety by comparing the model calculation value with the seepage flow control threshold.

[0076] It should be understood that the multi-point deformation safety monitoring system for a concrete face rockfill dam based on seepage flow in the embodiment of the present invention can implement all the technical solutions in the above method embodiment, and the functions of its respective functional modules can be specifically implemented according to the method in the above method embodiment. The specific implementation process can refer to the relevant description in the above method embodiment and will not be elaborated here.

[0077] The present invention also provides a computer device, including: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of the multi-point deformation safety monitoring method for a concrete face rockfill dam based on seepage flow as described above are implemented.

[0078] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the multi-point deformation safety monitoring method for a concrete face rockfill dam based on seepage flow as described above are implemented.

[0079] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a device (system), a computer device, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0080] The present invention is described with reference to the flowchart of the method according to the embodiments of the present invention. It should be understood that each process in the flowchart and the combination of processes in the flowchart can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in one Figure 1 process or multiple processes.

[0081] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the functions specified in one Figure 1 process or multiple processes.

[0082] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable devices provide steps for implementing the functions specified in one Figure 1 process or multiple processes.

Claims

1. A multi-point deformation safety monitoring method for concrete face rockfill dams based on seepage flow rate, characterized in that, It includes the following steps: Step 1: Perform data preprocessing on the seepage flow monitoring data of the concrete face rockfill dam, the monitoring data of relevant deformation measuring points, and the monitoring data of environmental quantities. The relevant deformation measuring points include the rockfill body, the foundation, the joints between slabs, and the peripheral joints; Step 2: Divide the data preprocessed in Step 1 into a training set and a test set. In the training set, select the measured values of relevant deformation measuring points, water level factors, temperature factors, rainfall factors, and time effect factors as model inputs, and use the seepage flow of the rockfill dam as the model output to construct a random forest model; Step 3: Analyze the random forest model constructed in Step 2 based on the SHAP interpretation method. By comparing the influence degrees of each input factor on the seepage flow of the dam, screen out the important factors that meet the specified conditions; Step 4: Use the regression analysis method or the machine learning method to perform modeling analysis on the important factors screened in Step 3. Combine the deformations of multiple measuring points of the rockfill body, the foundation, the joints between slabs, and the peripheral joints, as well as the water level, rainfall, temperature, and time effect factors to construct a seepage flow calculation model driven by multiple effect quantities, and use the test set divided in Step 2 to evaluate the model accuracy; Step 5: Determine the seepage flow control threshold according to the historical working conditions of the concrete face rockfill dam and the historical change of the seepage flow. Input the deformations of multiple measuring points of the rockfill body, the foundation, the joints between slabs, and the peripheral joints, as well as the water level, rainfall, temperature, and time effect factors into the seepage flow calculation model driven by multiple effect quantities constructed in Step 4. By comparing the model calculated value with the seepage flow control threshold, realize the integrated monitoring of the safety of multiple measuring point deformations.

2. The method according to claim 1, characterized in that, In Step 1, the data preprocessing includes gross error rejection, data interpolation, and benchmarking processing.

3. The method according to claim 1, characterized in that, In step 2, the water level factors selected are H1 - H0, h1 - h0, H represents the upstream water level, H1 is the water level on the monitoring day, H0 is the water level on the initial day of modeling, and H2, H3, H4, H5, H6 are the average water levels on the day before the monitoring day, the second day before, the third to fourth days before, the fifth to fifteenth days before, and the sixteenth to thirtieth days before, respectively. They are the average water levels on the day before the initial day of modeling, the second day before, the third to fourth days before, the fifth to fifteenth days before, and the sixteenth to thirtieth days before, respectively. h represents the downstream water level, and the meanings of each factor are the same as those of the upstream water level; the temperature factor selected is t is the cumulative number of days from the monitoring day to the initial day of modeling, and t0 is 1; the rainfall factor selected is U1, U2, U3, U4, U5 are the average rainfall amounts on the monitoring day, the day before, the second day before, the third to fourth days before, and the fifth to fifteenth days before, respectively. They are the average rainfall amounts on the initial day of modeling, the day before, the second day before, the third to fourth days before, and the fifth to fifteenth days before, respectively; the time effect factor selected is θ - θ0, lnθ - lnθ0, where θ is t / 100 and θ0 is t0 / 100.

4. The method according to claim 1, wherein Step 3 specifically includes: calculating the mean of the absolute values of the SHAP values of each factor under all samples, that is, the mean absolute SHAP value. Taking the mean absolute SHAP value as the standard, evaluate the influence degree of each factor on the seepage flow of the dam through sorting and plotting analysis. Select the factors with the mean absolute SHAP value within the specified range as important factors, and use the screened factors as the input variables of the seepage flow calculation model driven by multiple effect quantities.

5. The method according to claim 1, characterized in that, The seepage flow calculation model driven by multiple effect quantities in Step 4 is expressed as: Q = a0 + f(W) + f(T) + f(D) + f(C) + f(S) In the formula: W, T, D, and C respectively represent the water level factor, temperature factor, rainfall factor, and time effect factor screened in Step 3; S represents the measured values of important deformation measuring points screened in Step 3; a0 is a constant term; f() represents the function of the seepage flow with respect to the corresponding factor.

6. The method according to claim 1, characterized in that, In step 4, when evaluating the model accuracy, the correlation coefficient R 2 and the mean absolute percentage error MAPE are used as evaluation indicators.

7. The method according to claim 1, characterized in that, In the above step 5, through the comparison between the model calculated value and the seepage flow control threshold, the integrated monitoring of the deformation safety of multiple measuring points is realized, including: selecting the designed control value of the seepage flow of the concrete face rockfill dam or the historical extreme value of the seepage flow as the seepage safety monitoring threshold Q max , if the model predicted value Q≤Q in step 4 max , it indicates that the deformation of each measuring point of the concrete face rockfill dam is in a normal state and the structure is safe; if Q>Q max , it indicates that the deformation of some measuring points of the concrete face rockfill dam is in an abnormal state.

8. A multi-point deformation safety monitoring system for a concrete face rockfill dam based on seepage flow rate, characterized in that, It includes: A data preprocessing module for performing data preprocessing on the seepage flow monitoring data of the concrete face rockfill dam, the monitoring data of relevant deformation measuring points, and the monitoring data of environmental quantities. The relevant deformation measuring points include the rockfill body, the foundation, the joints between slabs, and the peripheral joints; A random forest model construction module for dividing the preprocessed data into a training set and a test set. In the training set, select the measured values of relevant deformation measuring points, water level factors, temperature factors, rainfall factors, and time effect factors as model inputs, and use the seepage flow of the rockfill dam as the model output to construct a random forest model; A feature screening module, which is used to analyze the constructed random forest model based on the SHAP interpretation method, and screen out important factors that meet the specified conditions by comparing the influence degrees of various input factors on the seepage flow of the dam; A seepage flow calculation model construction module, which is used to perform modeling analysis on the screened important factors by using regression analysis methods or machine learning methods, and construct a seepage flow calculation model jointly driven by multiple effect quantities by combining the deformations of multiple measuring points of the rockfill body, foundation, slab joints and peripheral joints, as well as water level, rainfall, temperature and aging factors, and use the divided test set to evaluate the model accuracy; A multi-measurement point deformation monitoring result output module, which is used to determine the seepage flow control threshold according to the historical working conditions of the concrete face rockfill dam and the historical change of the seepage flow, input the deformations of multiple measuring points of the rockfill body, foundation, slab joints and peripheral joints, as well as water level, rainfall, temperature and aging factors into the seepage flow calculation model jointly driven by multiple effect quantities, and realize the integrated monitoring of the safety of multi-measurement point deformation by comparing the calculated value of the model with the seepage flow control threshold.

9. A computer device, characterized in that, Comprising: One or more processors; A memory; And one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the program is executed by the processor, the steps of the multi-measurement point deformation safety monitoring method of the concrete face rockfill dam based on seepage flow as described in any one of claims 1-7 are realized.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of a multi-measurement point deformation safety monitoring method of a concrete face rockfill dam based on seepage flow as described in any one of claims 1-7 are realized.

Images