Intelligent monitoring and early warning system for cofferdam settlement deformation

Abstract

The invention relates to the technical field of intelligent monitoring and early warning, and particularly discloses an intelligent monitoring and early warning system for cofferdam settlement deformation, which is characterized in that a data acquisition module monitors the settlement gradient, stress distribution and strain value of each monitoring area of a cofferdam in real time to generate original monitoring data; the area identification module identifies an abnormal settlement concentration area based on settlement rate and gradient change analysis; the regional stress analysis module is used for carrying out stress and strain analysis on the abnormal settlement region and evaluating the stress stability; the influence analysis module establishes an influence relation model through machine learning, and analyzes the influence of stress and strain on the settlement rate; the early warning regulation and control module adjusts stress distribution according to the evaluation result, generates graded early warning information and displays the risk distribution situation of the cofferdam, and by accurately calculating the settlement abnormal coefficient and the stress and strain abnormal coefficient and combining with a machine learning optimization model, potential risks can be efficiently recognized, and response can be made.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent monitoring and early warning, and particularly relates to an intelligent monitoring and early warning system for the settlement and deformation of cofferdams. Background Art

[0002] A cofferdam is a commonly used temporary water retaining structure in the fields of water conservancy projects, port construction, and river regulation, etc. Its main function is to provide a dry land environment for construction. Due to the complexity of the cofferdam structure and the variability of the stress conditions, its safety is directly related to the smooth progress of the project construction and the stability of the surrounding environment. However, during the construction process, the cofferdam is prone to be affected by various adverse factors such as settlement, deformation, and stress concentration. Especially in an environment with complex geological conditions or large fluctuations in external loads, the risk of local instability and concentrated settlement is higher. With the development of intelligent monitoring technology, real-time monitoring of the cofferdam structure state, identification of potential risks, and timely early warning have become the key technical directions to ensure the safe operation of the cofferdam.

[0003] The existing technologies have the following deficiencies:

[0004] The existing technologies often rely on manual observation or simple monitoring methods, and cannot effectively monitor the dynamic and precise settlement changes and stress analysis of each area of the cofferdam structure, resulting in difficulty in timely discovering potential risk points, and thus unable to effectively prevent the deformation and damage of the cofferdam. The lack of intelligent monitoring, machine learning analysis, and multi-level data fusion processing leads to the system underestimating the settlement rate or deformation trend and being unable to give timely early warning. When the cofferdam suddenly becomes unstable due to exceeding the bearing capacity, it will cause instantaneous water body breakage. When the cofferdam suddenly becomes unstable, the rescue system may face tasks beyond its bearing capacity, and the resource scheduling gets out of control. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent monitoring and early warning system for the settlement and deformation of cofferdams to solve the problems in the above background.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] An intelligent monitoring and early warning system for the settlement and deformation of cofferdams, comprising:

[0008] A data acquisition module, which divides the cofferdam structure into several identical monitoring areas, and real-time monitors the settlement gradient change, stress distribution, and strain value of each monitoring area of the cofferdam, and generates original monitoring data;

[0009] An area identification module, which analyzes the settlement rate and gradient change in each monitoring area, and identifies the settlement anomaly concentration area according to the degree of settlement rate fluctuation and the settlement gradient change situation;

[0010] Regional stress analysis module, which dynamically analyzes the stress distribution and strain values in the area with concentrated settlement anomalies, and evaluates the stress stability in the corresponding area according to the stress state and deformation characteristics of the materials inside the cofferdam in each monitoring area;

[0011] Impact analysis module, which, based on the abnormal settlement rate, corresponding stress distribution and strain values in the area with concentrated settlement anomalies, uses machine learning algorithms to establish an impact relationship model and analyzes the degree of influence of stress and strain on the settlement rate;

[0012] Early warning and regulation module, which, based on the judgment result, regulates the stress distribution and strain values, generates corresponding hierarchical early warning information for the area with concentrated settlement anomalies, and displays the risk distribution of the cofferdam through a visualization platform.

[0013] As a further solution of the present invention: The identification of the area with concentrated settlement anomalies specifically includes:

[0014] Calculate the settlement rate in each monitoring area, calculate the settlement rate fluctuation anomaly coefficient according to the degree of settlement rate fluctuation during the monitoring period, calculate the settlement gradient anomaly coefficient according to the change of the settlement gradient in each monitoring area during the monitoring period, perform normalization calculation on the settlement rate fluctuation anomaly coefficient and the settlement gradient anomaly coefficient to obtain the regional settlement anomaly coefficient, and judge whether the regional settlement anomaly coefficient is greater than or equal to the preset threshold. If so, it is recorded as the area with concentrated settlement anomalies; if not, it is recorded as the non-area with concentrated settlement anomalies.

[0015] As a further solution of the present invention: The acquisition logic of the settlement rate fluctuation anomaly coefficient is as follows:

[0016] During the monitoring period, obtain the settlement rate in each monitoring area according to the time series to obtain the settlement rate time series data in each monitoring area;

[0017] Construct the feature matrix X of the time series. The settlement rate fluctuation is represented by the feature matrix and the sparse weight vector, and calculate the sparse prior distribution;

[0018] Optimize α and noise variance σ based on the maximization of the marginal likelihood 2 ;

[0019] After iterative update, obtain the posterior distribution mean of the sparse weight, and the calculation expression is:

[0020] Calculate the settlement rate fluctuation anomaly coefficient according to the sparse weight vector and the reconstruction error of the time series, and the calculation expression is:

[0021]

[0022] Among them, A i represents the abnormal coefficient of the settlement rate fluctuation in the i-th monitoring area, and y i represents the time series of the settlement rate in the i-th monitoring area, represents the sparse weight vector, and i represents the i-th monitoring area.

[0023] As a further solution of the present invention: the acquisition logic of the settlement gradient abnormal coefficient is as follows:

[0024] Obtain the settlement data d of each monitoring area within the monitoring period according to the time series a ={d 1 , d 2 , …, d n}, where d a represents the settlement value at the a-th time acquisition point, and n represents the total number of acquisition time points;

[0025] Calculate the difference between the settlement values of adjacent monitoring points and calculate the ratio with the time interval to obtain the settlement gradient of each monitoring point;

[0026] For each monitoring point d a , calculate the reciprocal of the difference measure of the settlement gradients of adjacent points to obtain the local density of each monitoring point. The calculation expression is:

[0027]

[0028] In the formula, ρ(d a ) represents the local density of the a-th time acquisition point, represents the settlement gradient of the d b -th time acquisition point, represents the settlement gradient of the a-th time acquisition point, and N k (a) represents the k-neighborhood of the a-th time acquisition point;

[0029] For each monitoring point, calculate the local outlier factor value;

[0030] Perform a mean calculation on the local outlier factor values of all time acquisition points within the monitoring area to obtain the average local outlier factor value of the current area. Use the standard deviation calculation formula to calculate the standard deviation of the local outlier factor values within the current area, and calculate the ratio of the average local outlier factor value to the standard deviation of the local outlier factor values to obtain the settlement gradient abnormal coefficient of the current area.

[0031] As a further solution of the present invention: evaluating the stress stability in the corresponding area specifically includes:

[0032] Based on the area with concentrated settlement anomalies, the strain values within the area with concentrated settlement anomalies are monitored in real time. According to the degree of fluctuation of the strain values during the monitoring period, the strain value anomaly coefficient is calculated. According to the degree of anomaly of the stress distribution during the monitoring period, the stress distribution anomaly coefficient is calculated. The strain value anomaly coefficient and the stress distribution anomaly coefficient are fused and calculated to obtain the regional stress stability coefficient.

[0033] As a further solution of the present invention: the process of obtaining the strain value anomaly coefficient is as follows:

[0034] Within the area with concentrated settlement anomalies, the stress values during the monitoring period are collected according to the time series to form the time series data s m =[s 1 , s 2 , …, s M , where s m represents the strain value of the mth collection point, m represents the number of collection points, and M represents the total number of collection points;

[0035] The normalized sequence s mz is subjected to discrete Fourier transform to convert the time-domain signal into a frequency-domain signal, and the Fourier coefficients are calculated;

[0036] Calculate the amplitude at frequency f, and calculate the energy of each frequency component;

[0037] Sum up the energies of all frequency components to obtain the total energy of the entire time series;

[0038] According to the preset high-frequency threshold, calculate the energy proportion of the high-frequency components. The calculation expression is:

[0039]

[0040] In the formula, R high represents the energy proportion of the high-frequency components, f thresh represents the preset high-frequency threshold, E total represents the total energy of the entire time series, and E f represents the energy of the frequency component;

[0041] Calculate the average value of the amplitudes of the high-frequency components, and calculate the ratio of the high-frequency energy proportion to the average value of the high-frequency amplitudes to obtain the strain value anomaly coefficient.

[0042] As a further solution of the present invention: the process of obtaining the stress distribution anomaly coefficient is as follows:

[0043] Obtain the time series data of the stress values and construct a stress distribution matrix;

[0044] Normalize the stress distribution matrix to obtain a normalized matrix, and decompose the normalized stress distribution matrix into a low-rank matrix and a sparse matrix, satisfying that the sum of the low-rank matrix and the sparse matrix is equal to the stress distribution matrix;

[0045] Calculate the norm of the sparse matrix, and the calculation expression is:

[0046]

[0047] In the formula, E represents the sparse matrix, and ||E|| F represents the norm of the sparse matrix, m represents the number of acquisition points, M represents the total number of acquisition points, t represents the time step, and T is the total number of time points in the monitoring period;

[0048] Calculate the abnormal mean value at time step t, normalize it to the time abnormal distribution weight, and calculate the stress distribution abnormal coefficient. The calculation expression is:

[0049]

[0050] In the formula, γ represents the stress distribution abnormal coefficient, and ω t represents the abnormal weight at time step t, and ||E(:,t)|| 2 represents the norm of the column vector of the sparse matrix at time step t.

[0051] As a further solution of the present invention: The establishment of the influence relationship model using a machine learning algorithm specifically includes:

[0052] Obtain the regional settlement abnormal coefficient and regional stress stability coefficient of the settlement abnormal concentration area, construct a comprehensive feature vector from the regional settlement abnormal coefficient and regional stress stability coefficient, use it as the input of the machine learning model, input the constructed comprehensive feature vector into the feature set used as the model training data, and divide the data set into a training set and a test set;

[0053] Construct a gradient boosting decision tree model, initialize the model parameters, including the maximum depth of the decision tree, the learning rate, and the number of decision trees; use the training data set for model iterative training, optimize the model parameters according to the target loss function, in each iteration, gradually construct the decision tree, predict the influence degree score through the trained model, and use the influence degree score as the output of the influence relationship model.

[0054] As a further solution of the present invention: The analysis of the influence degree of stress and strain on the settlement rate specifically includes:

[0055] Obtain the influence degree score of the settlement anomaly concentration area, and determine whether the influence degree score is greater than or equal to the preset threshold. If so, the anomaly of the settlement rate is affected by the anomaly of stress. If not, the anomaly of the settlement rate is not affected by the anomaly of stress.

[0056] As a further solution of the present invention: the regulation of the stress distribution and strain value specifically includes:

[0057] If the anomaly of the settlement rate is affected by the anomaly of stress, then regulate the stress distribution and strain value. Specifically: adjust the external load of the soil structure, and increase the support strength in the settlement anomaly concentration area, apply prestress to the soil to relieve stress concentration; reinforce the soil, and when the settlement rate is abnormal, real-time feedback stress and strain data, and adjust the construction in time.

[0058] The beneficial effects of the present invention:

[0059] (1) By constructing an intelligent monitoring and early warning system with multi-modularization as the core, the present invention integrates the comprehensive calculation methods of the settlement rate fluctuation anomaly coefficient, settlement gradient anomaly coefficient, strain value anomaly coefficient and stress distribution anomaly coefficient, and realizes the high-precision quantitative analysis of the anomaly degree of the cofferdam monitoring area. In data processing, the present invention introduces the sparse matrix decomposition technology to extract the potential correlation features in the monitoring data, and enhances the suppression effect on multi-dimensional data noise through the marginal likelihood maximization optimization algorithm. At the same time, combined with the local outlier factor calculation method, it accurately identifies the potential local instability or concentrated settlement trend in the cofferdam area. In addition, during the monitoring data fusion analysis process, the present invention dynamically constructs a causal association network between the settlement behavior and the stress characteristics, and effectively identifies the inducing factors of the cofferdam structure anomaly through multi-level model training, significantly improving the system's ability to identify key risk points. Compared with the traditional monitoring method based on a single index or empirical judgment, the systematic quantitative analysis and dynamic association evaluation strategy of the present invention significantly reduces the false alarms and missed alarms caused by insufficient monitoring accuracy, ensuring the scientificity, reliability and forward-looking of the monitoring results, and thus providing a more accurate risk assessment and regulation basis for the cofferdam project.

[0060] (2) The present invention constructs a multivariate influence relationship model based on machine learning algorithms, deeply explores the non-linear correlations among key parameters such as stress, strain, and settlement rate, dynamically evaluates the comprehensive influence degree of each parameter on settlement anomalies, and generates accurate hierarchical warning information for the monitored area in combination with the risk weight results output by the model. During the warning generation process, the present invention further combines specific engineering scenarios and proposes targeted control measures, including optimizing soil reinforcement strategies, adjusting construction load distribution, and enhancing the strength of support structures, etc., to achieve a closed-loop management from risk identification to engineering disposal. At the same time, the warning information is presented graphically through a visualization platform. Using color gradient maps, risk level identifiers, and dynamic area distribution maps, it real-time displays the risk distribution and change trends of each monitored area of the cofferdam, providing a clear and intuitive risk dynamic perception ability. This visualization platform supports interactive operations. Managers can quickly locate high-risk areas according to specific warning levels and anomaly distributions, formulate targeted disposal plans, and implement efficient response measures. Through this data-driven intelligent management mode, the present invention not only significantly improves the comprehensive control efficiency of the safe operation of the cofferdam, but also strengthens the scientific nature, real-time nature, and decision-making reference value of the warning information, providing a technologically leading intelligent monitoring and control solution for the cofferdam project. Brief Description of the Drawings

[0061] The present invention will be further described below with reference to the accompanying drawings.

[0062] Figure 1 is a flowchart of an intelligent monitoring and warning system for cofferdam settlement deformation according to the present invention;

[0063] Figure 2 is a flowchart of the acquisition process of the regional settlement anomaly coefficient in the present invention. Detailed Embodiments

[0064] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0065] Example 1, please refer to Figure 1 As shown, the present invention is an intelligent monitoring and warning system for cofferdam settlement deformation, including:

[0066] A data acquisition module, which divides the cofferdam structure into several identical monitored areas, and real-time monitors the settlement gradient changes, stress distributions, and strain values of each monitored area of the cofferdam, and generates original monitored data;

[0067] An area recognition module, which analyzes the settlement rate and gradient change in each monitoring area, and identifies the areas with concentrated settlement anomalies according to the fluctuation degree of the settlement rate and the change situation of the settlement gradient;

[0068] An area stress analysis module, which dynamically analyzes the stress distribution and strain values in the areas with concentrated settlement anomalies, and evaluates the stress stability in the corresponding areas according to the stress state and deformation characteristics of the materials inside the cofferdam in each monitoring area;

[0069] An influence analysis module, which, based on the abnormal settlement rate situation, corresponding stress distribution and strain values in the areas with concentrated settlement anomalies, uses machine learning algorithms to establish an influence relationship model and analyzes the influence degree of stress and strain on the settlement rate;

[0070] An early warning and regulation module, which, based on the judgment result, regulates the stress distribution and strain values, generates corresponding graded early warning information for the areas with concentrated settlement anomalies, and displays the risk distribution of the cofferdam through a visualization platform.

[0071] Example 2: Divide the cofferdam structure into several identical monitoring areas, and monitor the settlement gradient change, stress distribution and strain values of each monitoring area of the cofferdam in real time to generate original monitoring data, specifically including:

[0072] Divide the cofferdam structure into several identical monitoring areas, and determine the layout density and position of the monitoring points according to the geological conditions, structural form and key node distribution of the cofferdam. Install sensing devices in each monitoring area, including settlement sensors, stress sensors and strain sensors. These sensors are respectively used to collect key parameters such as settlement gradient change, stress distribution and strain values in the area in real time. The sensing devices are connected to the data acquisition unit (DAU) through wired or wireless communication methods to ensure the integrity and timeliness of the monitoring data.

[0073] The sensors collect real-time data at a preset sampling frequency (once a minute) to generate original monitoring data including settlement gradient, stress distribution and strain values. The data acquisition unit synchronizes the output signals of each sensor and eliminates environmental noise or abnormal interference through filtering technology. After the preprocessing of the data is completed, the processed monitoring data is transmitted to the central control system for storage and analysis by using a communication module (such as LoRa, 5G or fiber optic network). At the same time, the central control system performs validity verification on the transmitted data to prevent data loss or errors and ensure the accuracy of subsequent analysis.

[0074] Example 3, please refer to Figure 2As shown, the settlement rate and gradient changes in each monitoring area are analyzed. According to the fluctuation degree of the settlement rate and the change of the settlement gradient, the areas with concentrated settlement anomalies are identified, specifically including:

[0075] The settlement rate in each monitoring area is calculated. According to the fluctuation degree of the settlement rate within the monitoring period, the settlement rate fluctuation anomaly coefficient is calculated. According to the change of the settlement gradient in each monitoring area within the monitoring period, the settlement gradient anomaly coefficient is calculated. The settlement rate fluctuation anomaly coefficient and the settlement gradient anomaly coefficient are subjected to normalization calculation and processing to obtain the regional settlement anomaly coefficient. It is judged whether the regional settlement anomaly coefficient is greater than or equal to the preset threshold. If so, it is recorded as the area with concentrated settlement anomalies. If not, it is recorded as the area without concentrated settlement anomalies;

[0076] The acquisition logic of the settlement rate fluctuation anomaly coefficient is as follows:

[0077] Within the monitoring period, the settlement rate in each monitoring area is obtained according to the time series, and the settlement rate time series data in each monitoring area is obtained, expressed as:

[0078]

[0079] where i represents the i-th monitoring area, is the total number of time points in the monitoring period, and y i represents the settlement rate time series of the i-th monitoring area;

[0080] Construct the feature matrix of the time series

[0081] where x d represents the basic settlement rate feature of the time series, d represents the total number of settlement rate features, and X represents the feature matrix, initially set as the identity matrix;

[0082] The settlement rate fluctuation is represented by the feature matrix and the sparse weight vector, and the calculation expression is:

[0083]

[0084] In the formula, w represents the sparse weight vector, w d represents the weight of the d-th settlement rate feature, ∈ represents the Gaussian noise term, and σ 2 represents the noise variance, represents the normal distribution,

[0085] The sparse prior distribution is:

[0086]

[0087] Among them, α represents the sparse hyperparameter, and w j represents the weight of the j-th settlement rate feature, and α j represents the sparse hyperparameter of the j-th settlement rate feature, and j represents the j-th settlement rate feature;

[0088] Optimize α and the noise variance σ based on the maximization of the marginal likelihood 2 , where the marginal likelihood formula is:

[0089]

[0090] Among them, A = diag(α), and I represents the identity matrix;

[0091] After iterative update, obtain the mean value of the posterior distribution of the sparse weight, and the calculation expression is:

[0092]

[0093] Among them, μ w represents the mean value of the posterior distribution, A represents the diagonal matrix, and C represents the covariance matrix;

[0094] According to the reconstruction error of the sparse weight vector and the time series, calculate the settlement rate fluctuation anomaly coefficient, and the calculation expression is:

[0095]

[0096] Among them, A i represents the settlement rate fluctuation anomaly coefficient in the i-th monitoring area;

[0097] The acquisition logic of the settlement gradient anomaly coefficient is as follows:

[0098] Obtain the settlement data d of each monitoring area within the monitoring period according to the time series a ={d 1 , d 2 ,…, d n}, where d a represents the settlement value at the a-th time acquisition point, and n represents the total number of acquisition time points;

[0099] Based on the change in the settlement value of adjacent monitoring points, calculate the settlement gradient of each monitoring point, and the calculation expression is:

[0100]

[0101] Among them, represents the settlement gradient at the a-th time acquisition point, Δt represents the time interval between adjacent two points, and d a+1 represents the settlement value at the (a + 1)-th time acquisition point;

[0102] For each monitoring point d a , calculate the reciprocal of the difference measure of the settlement gradients of adjacent points to obtain the local density of each monitoring point. The calculation expression is:

[0103]

[0104] In the formula, ρ(d a ) represents the local density at the a-th time acquisition point, represents the settlement gradient at the d b -th time acquisition point, and N k (a) represents the k-neighborhood of the a-th time acquisition point;

[0105] Among them, the Euclidean distance is used as the difference measure between settlement gradients;

[0106] For each monitoring point, calculate the local outlier factor value, which represents the degree of abnormality of the corresponding monitoring point relative to its neighborhood. The calculation expression is:

[0107]

[0108] Among them, |N k (a)| is the number of points in the neighborhood of point d a , ρ(d b ) represents the local density at the b-th time acquisition point, and LOF(d a ) represents the local outlier factor value at the a-th time acquisition point in the monitoring area;

[0109] Calculate the mean value of the local outlier factor values of all time acquisition points in the monitoring area to obtain the average local outlier factor value of the current area. Use the standard deviation calculation formula to calculate the standard deviation value of the local outlier factor values in the current area. Calculate the ratio of the average local outlier factor value to the standard deviation value of the local outlier factor values to obtain the settlement gradient anomaly coefficient of the current area;

[0110] The calculation process of the regional settlement anomaly coefficient is as follows:

[0111]

[0112] In the formula, AD i represents the regional settlement anomaly coefficient in the i-th monitoring area, A i represents the settlement rate fluctuation anomaly coefficient in the i-th area, B i represents the settlement gradient anomaly coefficient in the i-th area, c 1 and c 2 are preset proportionality coefficients, and c 1 and c2 All are greater than 0.

[0113] It should be noted that: In this embodiment, through the dynamic analysis of the settlement rate and settlement gradient in the monitoring area, the settlement rate fluctuation anomaly coefficient and settlement gradient anomaly coefficient are calculated respectively. Combining advanced algorithms such as sparse matrix decomposition, marginal likelihood maximization optimization, and local outlier factor, the anomaly degree of the regional settlement behavior is accurately quantified, and finally the regional settlement anomaly coefficient is generated by fusion. Using the sparse weight vector to perform dimensionality reduction modeling on the time series features improves the expression ability for complex settlement features; at the same time, based on the calculation of the local outlier factor and neighborhood density, the changes in the local anomaly gradient are accurately captured, significantly enhancing the sensitivity and recognition ability for the potential settlement concentration trend within the region. The overall solution can effectively identify the abnormal settlement concentration area of the cofferdam, improve the early warning ability of the monitoring system for the local instability or concentrated settlement of the cofferdam, greatly reduce the risk of potential structural instability, and provide important technical support for the operation safety of the cofferdam.

[0114] Embodiment 4: Dynamically analyze the stress distribution and strain values in the abnormal settlement concentration area, and evaluate the stress stability in the corresponding area according to the stress state and deformation characteristics of the materials inside the cofferdam in each monitoring area, specifically including:

[0115] The evaluation of the stress stability in the corresponding area specifically includes:

[0116] Based on the abnormal settlement concentration area, the strain values in the abnormal settlement concentration area are monitored in real time. According to the degree of fluctuation of the strain values during the monitoring period, the strain value anomaly coefficient is calculated. According to the degree of anomaly of the stress distribution during the monitoring period, the stress distribution anomaly coefficient is calculated. The strain value anomaly coefficient and the stress distribution anomaly coefficient are fused and calculated to obtain the regional stress stability coefficient;

[0117] The process of obtaining the strain value anomaly coefficient is as follows:

[0118] In the abnormal settlement concentration area, the stress values during the monitoring period are collected according to the time series to form the time series data s m =[s 1 , s 2 , …, s M , where s m represents the strain value at the m-th collection point, m represents the number of collection points, and M represents the total number of collection points;

[0119] The time series data is denoised, and its mean value is calculated. After removing the mean value, the normalized sequence s mz is obtained;

[0120] For the normalized sequence s mzPerform a discrete Fourier transform to convert the time-domain signal into a frequency-domain signal, calculate the Fourier coefficients, and the calculation expression is:

[0121]

[0122] In the formula, F(f) represents the Fourier coefficient of frequency f, j represents the imaginary unit, and e represents the natural logarithm base;

[0123] Calculate the amplitude at frequency f, and the calculation expression is:

[0124]

[0125] where Re represents the real part of the Fourier coefficient, Im represents the imaginary part of the Fourier coefficient, and A f represents the amplitude at frequency f;

[0126] Calculate the energy of each frequency component, and the calculation expression is:

[0127]

[0128] In the formula, E f represents the energy of the frequency component;

[0129] Sum up the energies of all frequency components to obtain the total energy of the entire time series;

[0130] According to the preset high-frequency threshold, calculate the energy proportion of the high-frequency components, and the calculation expression is:

[0131]

[0132] In the formula, R high represents the energy proportion of the high-frequency components, f thresh represents the preset high-frequency threshold, and E total represents the total energy of the entire time series;

[0133] Calculate the average amplitude of the high-frequency components, and calculate the ratio of the high-frequency energy proportion to the average high-frequency amplitude to obtain the strain value anomaly coefficient;

[0134] The process of obtaining the stress distribution anomaly coefficient is as follows:

[0135] Obtain the time series data of the stress values and construct a stress distribution matrix;

[0136] Perform normalization processing on the stress distribution matrix to obtain a normalized matrix, and decompose the normalized stress distribution matrix into a low-rank matrix and a sparse matrix, and satisfy that the sum of the low-rank matrix and the sparse matrix is equal to the stress distribution matrix;

[0137] Calculate the norm of the sparse matrix, and the calculation expression is:

[0138]

[0139] Wherein, E represents a sparse matrix, and ||E|| F represents the norm of the sparse matrix, m represents the number of acquisition points, M represents the total number of acquisition points, t represents the time step, and T is the total number of time points in the monitoring period;

[0140] Calculate the abnormal mean value at the time step t, and normalize it to the time abnormal distribution weight, and calculate the stress distribution abnormal coefficient. The calculation expression is:

[0141]

[0142] Wherein, γ represents the stress distribution abnormal coefficient, and ω t represents the abnormal weight at the time step t, and ||E(:,t)|| 2 represents the norm of the column vector of the sparse matrix at the time step t;

[0143] The calculation process of the regional stress stability coefficient is as follows:

[0144]

[0145] Wherein, Az g represents the regional stress stability coefficient of the g-th settlement abnormal concentration area, g represents the number of settlement abnormal concentration areas, and δ g represents the strain value abnormal coefficient of the g-th settlement abnormal concentration area, and γ g represents the stress distribution abnormal coefficient of the g-th settlement abnormal concentration area, and c 3 and c 4 are preset proportionality coefficients, and c 3 and c 4 are both greater than 0.

[0146] It should be noted that: in this embodiment, by real-time monitoring the strain value and stress distribution in the settlement abnormal concentration area, using advanced algorithms such as Fourier transform and sparse matrix decomposition, dynamic analysis of the frequency and spatial components of time series data is carried out, the strain value abnormal coefficient and stress distribution abnormal coefficient are accurately calculated, and then the regional stress stability coefficient is obtained by fusion. This embodiment can effectively capture the stress state and deformation characteristics of the materials inside the cofferdam in the settlement concentration area, and comprehensively evaluate the stress stability in the area. Through the ratio analysis of the high-frequency energy ratio and the amplitude mean value and the normalization processing of the sparse matrix time abnormal weight, the sensitivity to high-frequency abnormal fluctuations and local stress imbalance is improved, thereby improving the ability to identify the potential instability risk of the cofferdam structure, and providing a more accurate and comprehensive scientific basis for the safety warning of the cofferdam.

[0147] Example 5. Based on the settlement rate anomalies, corresponding stress distributions, and strain values in the concentrated area of settlement anomalies, a relationship model is established using a machine learning algorithm to analyze the influence of stress and strain on the settlement rate. Specifically, it includes:

[0148] Obtain the regional settlement anomaly coefficient and regional stress stability coefficient of the concentrated area of settlement anomalies, construct a comprehensive feature vector from the regional settlement anomaly coefficient and regional stress stability coefficient, and use it as the input of the machine learning model. Input the constructed comprehensive feature vector into the feature set used as the model training data, and divide the data set into a training set and a test set with a ratio of 8:2.

[0149] Construct a gradient boosting decision tree model, and initialize the model parameters, including the maximum depth of the decision tree, learning rate, and number of decision trees. Use the training data set for iterative training of the model, optimize the model parameters according to the target loss function. In each iteration, gradually construct decision trees, and improve the fitting ability of the model to the relationship between input features and target labels through residual calculation and weighted update. Use the test data set to evaluate the model performance, calculate the accuracy, recall rate, and F1 score of the model prediction, verify the generalization ability of the model, predict the influence degree score through the trained model, and use the influence degree score as the output of the relationship model.

[0150] Obtain the influence degree score of the concentrated area of settlement anomalies, and determine whether the influence degree score is greater than or equal to the preset threshold. If so, the anomaly of the settlement rate is affected by the anomaly of stress; if not, the anomaly of the settlement rate is not affected by the anomaly of stress.

[0151] Example 6. Based on the judgment result, adjust the stress distribution and strain values, generate corresponding hierarchical warning information for the concentrated area of settlement anomalies, and display the cofferdam risk distribution through a visualization platform. Specifically, it includes:

[0152] If the anomaly of the settlement rate is affected by the anomaly of stress, then adjust the stress distribution and strain values. Specifically: adjust the external load of the soil structure, increase the support strength in the concentrated area of settlement anomalies, apply prestress to the soil to relieve stress concentration; reinforce the soil, such as grouting, foundation treatment, increasing the foundation depth, and when the settlement rate shows anomalies, real-time feedback of stress and strain data, and timely adjustment of construction.

[0153] For the concentrated area of settlement anomalies before regulation, the warning information is presented in the form of a color gradient map or risk label through the visualization platform, intuitively showing the risk distribution of each area of the cofferdam, facilitating managers to quickly locate the abnormal area and take corresponding intervention measures to ensure the safe operation of the cofferdam and the efficiency of risk management decisions.

[0154] Working principle of the present invention: The system of the present invention includes a data acquisition module, a region identification module, a regional stress analysis module, an impact analysis module, and an early warning and control module. The data acquisition module monitors the settlement gradient, stress distribution, and strain value of each monitoring region of the cofferdam in real time through sensors, and generates original monitoring data; the region identification module analyzes the settlement rate fluctuation and gradient change of each region to identify regions with abnormal settlement concentration; the regional stress analysis module evaluates the stress stability within the region based on the stress state and deformation characteristics of the material; the impact analysis module uses machine learning algorithms to construct an impact relationship model between stress, strain, and settlement rate, and quantifies the impact degree of each factor on abnormal settlement; the early warning and control module generates hierarchical early warning information according to the analysis results, regulates the regional stress and strain when necessary, and simultaneously displays the risk distribution through a visualization platform. The present invention integrates advanced algorithms such as sparse matrix decomposition, local outlier factor, Fourier transform, and gradient boosting decision tree to accurately capture the abnormal settlement trend and stress change, improve the accuracy and response efficiency of early warning, and provide technical guarantee for the safe operation of the cofferdam.

[0155] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0156] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired or wireless (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0157] It should be understood that the term "and / or" in this text is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. Additionally, the character " / " in this text generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood by referring to the context before and after.

[0158] It should be understood that in various embodiments of this application, the magnitudes of the serial numbers of the above processes do not imply the sequence of execution. The execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of this application.

[0159] The above has described in detail one embodiment of the present invention, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made in accordance with the scope of the application of the present invention should still fall within the scope covered by the patent of the present invention.

Claims

1. An intelligent monitoring and early warning system for cofferdam settlement and deformation, characterized in that: include: A data acquisition module, wherein the data acquisition module divides the cofferdam structure into a number of identical monitoring areas, monitors the settlement gradient change, stress distribution and strain value of each monitoring area of ​​the cofferdam in real time, and generates original monitoring data; A region identification module, which analyzes the sedimentation rate and gradient changes in each monitoring area, and identifies the area where sedimentation anomalies are concentrated according to the degree of sedimentation rate fluctuation and sedimentation gradient changes; A regional stress analysis module, which dynamically analyzes the stress distribution and strain value of the settlement abnormality concentrated area, and evaluates the stress stability in the corresponding area according to the stress state and deformation characteristics of the internal material of the cofferdam in each monitoring area; An impact analysis module, which uses a machine learning algorithm to establish an impact relationship model based on the abnormal settlement rate in the settlement abnormality concentrated area and the corresponding stress distribution and strain value, and analyzes the influence of stress and strain on the settlement rate; An early warning and control module regulates stress distribution and strain values ​​based on the judgment results, generates corresponding graded early warning information for areas where settlement anomalies are concentrated, and displays the risk distribution of cofferdams through a visualization platform.

2. The intelligent monitoring and early warning system for cofferdam settlement deformation according to claim 1 is characterized in that: The identification of the concentrated area of ​​abnormal settlement specifically includes: The sedimentation rate in each monitoring area is calculated, and the sedimentation rate fluctuation anomaly coefficient is calculated according to the degree of sedimentation rate fluctuation during the monitoring period. The sedimentation gradient anomaly coefficient is calculated according to the change of sedimentation gradient in each monitoring area during the monitoring period. The sedimentation rate fluctuation anomaly coefficient and the sedimentation gradient anomaly coefficient are normalized and calculated to obtain the regional sedimentation anomaly coefficient. It is determined whether the regional sedimentation anomaly coefficient is greater than or equal to the preset threshold. If so, it is recorded as a sedimentation anomaly concentrated area. If not, it is recorded as a non-sedimentation anomaly concentrated area.

3. The intelligent monitoring and early warning system for cofferdam settlement deformation according to claim 2 is characterized in that: The logic for obtaining the sedimentation rate fluctuation anomaly coefficient is: During the monitoring period, the sedimentation rate in each monitoring area is obtained according to the time series to obtain the sedimentation rate time series data in each monitoring area; Construct the feature matrix X of the time series. The sedimentation rate fluctuation is represented by the feature matrix and the sparse weight vector, and the sparse prior distribution is calculated. Based on the marginal likelihood maximization, we optimize α and noise square σ 2 ; After iterative updating, the mean of the posterior distribution of the sparse weights is obtained, and the calculation expression is: According to the sparse weight vector and the reconstruction error of the time series, the sedimentation rate fluctuation anomaly coefficient is calculated, and the calculation expression is: Among them, A i represents the sedimentation rate fluctuation anomaly coefficient in the ith monitoring area, y i represents the sedimentation rate time series of the ith monitoring area, represents a sparse weight vector, and i represents the i-th monitoring area.

4. The intelligent monitoring and early warning system for cofferdam settlement deformation according to claim 2 is characterized in that: The logic for obtaining the sedimentation gradient anomaly coefficient is: Obtain the settlement data d of each monitoring area during the monitoring period according to the time series a ={d1,d2,…,d n }, where d a represents the sedimentation value at the ath time collection point, and n represents the total number of collection time points; Calculate the difference in settlement values ​​between adjacent monitoring points and perform ratio calculation with the time interval to obtain the settlement gradient of each monitoring point; For each monitoring point d a , calculate the inverse of the difference measure of the sedimentation gradient of adjacent points, and obtain the local density of each monitoring point. The calculation expression is: In the formula, ρ(d a ) represents the local density of the a-th time acquisition point, Indicates the d b The sedimentation gradient at each time point is represents the sedimentation gradient at the ath time point, N k (a) represents the k-neighborhood of the a-th time acquisition point; For each monitoring point, calculate the local anomaly factor value; The local anomaly factor values ​​of all time collection points in the monitoring area are averaged to obtain the average local anomaly factor value of the current area. The standard deviation calculation formula is used to calculate the standard deviation of the local anomaly factor values ​​in the current area. The average local anomaly factor value is calculated and the standard deviation of the local anomaly factor value is calculated to obtain the sedimentation gradient anomaly coefficient of the current area.

5. The intelligent monitoring and early warning system for cofferdam settlement deformation according to claim 1 is characterized in that: The stress stability in the evaluation corresponding to the area specifically includes: Based on the settlement anomaly concentrated area, the strain value in the settlement anomaly concentrated area is monitored in real time. The strain value anomaly coefficient is calculated according to the strain value fluctuation degree within the monitoring period. The stress distribution anomaly coefficient is calculated according to the abnormal degree of stress distribution within the monitoring period. The strain value anomaly coefficient and the stress distribution anomaly coefficient are fused and calculated to obtain the regional stress stability coefficient.

6. The intelligent monitoring and early warning system for cofferdam settlement and deformation according to claim 5 is characterized in that: The process of obtaining the strain value anomaly coefficient is as follows: In the area where settlement anomalies are concentrated, the stress values ​​within the monitoring period are collected according to the time series to form the time series data of the stress values ​​s m =[s1,s2,…,s M ], where s m represents the strain value of the mth collection point, m represents the number of collection points, and M represents the total number of collection points; For the normalized sequence s mz Perform discrete Fourier transform to convert the time domain signal into frequency domain signal and calculate Fourier coefficients; Calculate the amplitude at frequency f and calculate the energy of each frequency component; The energy of all frequency components is summed up to obtain the total energy of the entire time series; According to the preset high-frequency threshold, the energy proportion of the high-frequency component is calculated, and the calculation expression is: In the formula, R high Indicates the energy proportion of high-frequency components, f thresh Indicates the preset high frequency threshold, E total Represents the total energy of the entire time series, E f Represents the energy of a frequency component; The mean amplitude of the high-frequency component is calculated, and the ratio of the high-frequency energy proportion to the mean high-frequency amplitude is calculated to obtain the strain value anomaly coefficient.

7. The intelligent monitoring and early warning system for cofferdam settlement and deformation according to claim 5 is characterized in that: The process of obtaining the stress distribution anomaly coefficient is as follows: Obtain time series data of stress values ​​and construct a stress distribution matrix; Normalizing the stress distribution matrix to obtain a normalized matrix, decomposing the normalized stress distribution matrix into a low-rank matrix and a sparse matrix, and satisfying that the sum of the low-rank matrix and the sparse matrix is ​​equal to the stress distribution matrix; Calculate the norm of the sparse matrix. The calculation expression is: In the formula, E represents a sparse matrix, ||E|| F represents the norm of the sparse matrix, m represents the number of acquisition points, M represents the total number of acquisition points, t represents the time step, and T represents the total number of time points in the monitoring period; Calculate the abnormal mean at time step t and normalize it to the time abnormal distribution weight, calculate the stress distribution abnormal coefficient, and the calculation expression is: In the formula, γ represents the stress distribution abnormal coefficient, ω t represents the abnormal weight at time step t, and ||E(:,t)||2 represents the l2 norm of the column vector of the sparse matrix at time step t.

8. The intelligent monitoring and early warning system for cofferdam settlement and deformation according to claim 1 is characterized in that: The use of a machine learning algorithm to establish an influence relationship model specifically includes: Obtain the regional settlement anomaly coefficient and regional stress stability coefficient of the settlement anomaly concentrated area, construct a comprehensive feature vector from the regional settlement anomaly coefficient and the regional stress stability coefficient, use them as the input of the machine learning model, input the constructed comprehensive feature vector as the feature set of the model training data, and divide the data set into a training set and a test set; Build a gradient boosting decision tree model and initialize the model parameters, including the maximum depth of the decision tree, learning rate, and number of decision trees. Use the training data set to iteratively train the model and optimize the model parameters according to the target loss function. In each iteration, gradually build a decision tree, predict the impact score through the trained model, and use the impact score as the output of the impact relationship model.

9. The intelligent monitoring and early warning system for cofferdam settlement deformation according to claim 1 is characterized in that: The analysis of the influence of stress and strain on the sedimentation rate specifically includes: Obtain the impact score of the settlement anomaly concentrated area, and determine whether the impact score is greater than or equal to a preset threshold. If so, the settlement rate anomaly is affected by the stress anomaly; if not, the settlement rate anomaly is not affected by the stress anomaly.

10. The intelligent monitoring and early warning system for cofferdam settlement and deformation according to claim 1, characterized in that: The regulating and controlling of stress distribution and strain value specifically includes: If the abnormal settlement rate is affected by the abnormal stress, the stress distribution and strain value will be regulated. Specifically: adjust the external load of the soil structure, increase the support strength in the area where the abnormal settlement is concentrated, apply prestress to the soil, and relieve stress concentration; reinforce the soil, and when the settlement rate is abnormal, provide real-time feedback on the stress and strain data, and adjust the construction in time.

Images