Three-dimensional visual monitoring method for grouting amount of side slope anchor rod

Abstract

The invention discloses a side slope anchor rod grouting amount three-dimensional visual monitoring method, and relates to the technical field of anchor rod grouting quality monitoring, and the method comprises the steps: collecting real-time grouting parameter data based on a grouting parameter sensor arranged in a side slope anchor rod construction area; the method comprises the following steps: constructing a three-dimensional geometric model of a side slope anchor rod, generating three-dimensional grouting amount distribution data, and generating a real-time dynamic side slope anchor rod grouting amount three-dimensional visual monitoring image by adopting a three-dimensional visual rendering technology; based on a trained image anomaly judgment model, analyzing the three-dimensional visual monitoring image, judging whether an anomaly occurs or not, and outputting abnormal data; and outputting abnormal data based on the image anomaly judgment model, and outputting an anomaly type and anomaly prediction time by using the trained abnormal data analysis model. The slope anchor rod grouting amount full-process intelligent monitoring system is constructed, and the limitation of a traditional monitoring method on data precision, space presentation and abnormal response is broken through.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of anchor rod grouting quality monitoring, in particular to a three-dimensional visualization monitoring method for slope anchor rod grouting quantity. BACKGROUND

[0002] In slope engineering construction, the anchor rod grouting quality directly affects the stability and safety of the slope, and the grouting quantity is one of the core indicators for measuring the grouting quality. The traditional slope anchor rod grouting quantity monitoring method relies on manual recording or single-point sensor data collection, which has the following shortcomings: first, the monitoring data is scattered, and it is difficult to intuitively reflect the distribution of the grouting quantity in the full length range of the anchor rod, which is prone to local grouting deficiency or excess hidden dangers; second, the data processing is lagging, and it is difficult to dynamically track the grouting process in real time, making it difficult to give timely warnings when grouting anomalies occur; third, there is a lack of three-dimensional spatial perspective, and it is difficult for management personnel to quickly understand the correlation between the grouting quantity and the spatial position of the anchor rod, which brings inconvenience to the construction quality control and decision-making.

[0003] With the development of digital construction technology, although some monitoring methods have introduced visualization technology, they are mostly two-dimensional chart displays, which cannot realize three-dimensional spatial mapping of the grouting quantity, and the analysis of abnormal data relies on manual judgment, which is low in efficiency and poor in accuracy. In addition, the pre-processing of grouting parameters in the prior art is not fine enough, and is easily affected by environmental interference, leading to data deviation and affecting the accuracy of grouting quantity calculation. Therefore, there is an urgent need for a method that can realize three-dimensional visualization of grouting quantity, real-time anomaly monitoring and intelligent early warning, in order to improve the quality control level of slope anchor rod grouting construction. SUMMARY

[0004] The purpose of the present application is to solve the above-mentioned problems, and a three-dimensional visualization monitoring method for slope anchor rod grouting quantity is proposed.

[0005] The purpose of the present application can be achieved by the following technical solutions:

[0006] A three-dimensional visualization monitoring method for slope anchor rod grouting quantity, comprising:

[0007] Based on the grouting parameter sensors arranged in the slope anchor construction area, real-time grouting parameter data is collected, and the real-time grouting parameter data is pre-processed;

[0008] A three-dimensional geometric model of the slope anchor rod is constructed, the pre-processed grouting parameter data is mapped to the three-dimensional geometric model of the slope anchor rod, three-dimensional grouting quantity distribution data is generated, and a real-time dynamic three-dimensional visualization monitoring image of the slope anchor rod grouting quantity is generated by using three-dimensional visualization rendering technology;

[0009] Based on the trained image anomaly judgment model, the three-dimensional visualization monitoring image is analyzed to determine whether an anomaly occurs, and abnormal data is output;

[0010] Based on the image anomaly judgment model, the trained abnormal data analysis model is used to output the abnormal type and the abnormal prediction time, and the warning information is sent based on the abnormal type and the abnormal prediction time.

[0011] Preferably, the grouting parameter sensor based on the slope anchor construction area layout collects real-time grouting parameter data, and the pre-processing of the real-time grouting parameter data specifically includes:

[0012] The real-time collected grouting pressure and grout flow data are processed by using the moving average filtering method to remove high-frequency interference signals.

[0013] The grout density data is temperature compensated, and based on the pre-established grout density-temperature relationship curve, the density data at different temperatures is converted into density values at standard temperature.

[0014] The abnormal values in the grouting parameter data are removed, and when the parameter value exceeds the preset threshold range, the effective value at the adjacent time is used for interpolation replacement.

[0015] Preferably, the three-dimensional geometric model of the slope anchor is constructed, and the pre-processed grouting parameter data is mapped to the three-dimensional geometric model of the slope anchor to generate three-dimensional grouting amount distribution data, which specifically includes:

[0016] The geological survey report and anchor design drawings of the slope engineering are collected to obtain the topographic data of the slope, the length, diameter, spacing and inclination angle parameters of the anchor.

[0017] The three-dimensional modeling software is used to arrange the anchor model in the three-dimensional space based on the topographic data of the slope according to the design parameters of the anchor to form a preliminary three-dimensional geometric model of the slope anchor.

[0018] The actual shape data of the slope and the anchor are obtained by on-site three-dimensional laser scanning, and the preliminary constructed three-dimensional geometric model is corrected so that the error between the model and the actual situation is within the allowable range.

[0019] The cumulative grouting amount is calculated according to the grouting time and the grout flow.

[0020] The anchor is divided into a plurality of unit segments along the length direction, and the grouting amount of each unit segment is calculated according to the grouting time corresponding to each unit segment.

[0021] The grouting amount of each unit segment and the corresponding grouting pressure, grout density and other parameters are associated with the corresponding unit position in the three-dimensional geometric model to form three-dimensional grouting amount distribution data.

[0022] Preferably, the three-dimensional visualization rendering technology is used to generate real-time dynamic three-dimensional visualization monitoring images of the slope anchor grouting amount, which specifically includes:

[0023] The surface rendering algorithm is used to process the three-dimensional grouting amount distribution data to generate a surface three-dimensional image of the slope anchor grouting amount;

[0024] The color mapping is used to map the size of the grouting amount to different colors;

[0025] The transparency of the image is set so that the operator can clearly observe the internal anchor and the grouting amount distribution at different depths.

[0026] Preferably, the trained image anomaly judgment model analyzes the three-dimensional visual monitoring image to determine whether an anomaly occurs, and outputs the abnormal data, which specifically includes:

[0027] The three-dimensional visual monitoring image of the slope anchor grouting amount is collected, which includes normal grouting images and abnormal grouting images;

[0028] The collected images are labeled to mark the location, range and type of abnormal areas;

[0029] A convolutional neural network model based on deep learning is constructed, the labeled images are divided into a training set and a test set, the training set is used to train the model, and the performance of the model is evaluated through the test set.

[0030] Preferably, the trained image anomaly judgment model analyzes the three-dimensional visual monitoring image to determine whether an anomaly occurs, and outputs the abnormal data, which specifically includes:

[0031] The real-time generated three-dimensional visual monitoring image is input into the trained image anomaly judgment model;

[0032] The model extracts and analyzes the features of the image and compares them with the normal image features stored in the model;

[0033] When the deviation of the features in the image from the normal features exceeds the preset threshold, it is determined that an anomaly has occurred, and the three-dimensional coordinates of the abnormal area, the time of the anomaly and the corresponding grouting parameter data are output.

[0034] Preferably, the trained image anomaly judgment model specifically includes:

[0035] Historical abnormal data is collected, including grouting parameters, abnormal types, abnormal duration, treatment measures and treatment result data when the anomaly occurs;

[0036] The historical abnormal data is preprocessed to remove invalid data and duplicate data, and the data is standardized;

[0037] The support vector machine algorithm is used to construct an abnormal data analysis model, the preprocessed abnormal data is used as a training sample, the abnormal type and abnormal prediction time are used as the output of the model, and the parameters of the model are optimized through training, so that the model can accurately predict the abnormal type and abnormal prediction time.

[0038] Preferably, the support vector machine algorithm is used to construct an abnormal data analysis model, and the support vector machine algorithm is used to construct an abnormal data analysis model.

[0039] The normalized abnormal data is divided into an input feature vector and an output label, the input feature vector is the grouting pressure, grout flow and grout density parameters when the abnormality occurs, and the output label is the abnormal type and abnormal prediction time.

[0040] The radial basis kernel function is used as the kernel function of the support vector machine.

[0041] The penalty parameter and the kernel function parameter of the support vector machine are optimized by the grid search method, the average accuracy of five-fold cross-validation is used as an evaluation index, and the optimal parameter combination is selected.

[0042] The support vector machine model with the optimal parameter combination is used to train the training sample to generate an abnormal data analysis model.

[0043] Preferably, the warning information is sent based on the abnormal type and the abnormal prediction time.

[0044] The warning level is determined according to the abnormal type, wherein the slight abnormality corresponds to the third-level warning, the general abnormality corresponds to the second-level warning, and the serious abnormality corresponds to the first-level warning.

[0045] The warning response time is calculated in combination with the abnormal prediction time.

[0046] An alarm signal is sent through a sound-light alarm device on the construction site, and the warning information is sent to the mobile terminal of the relevant management personnel, the warning information including the abnormal type, the abnormal position, the abnormal prediction time and the recommended treatment measures.

[0047] Preferably, the warning information sending time, the receiving personnel and the response state are recorded to form a warning response log.

[0048] The grouting parameters of the abnormal area are re-collected and analyzed every 5-15 minutes, the abnormal type and the abnormal prediction time are updated, and the updated results are synchronized to the mobile terminal of the relevant management personnel.

[0049] When the abnormality is handled, the processed grouting parameter data and the three-dimensional visual monitoring image are collected, compared with the data when the abnormality occurs, and the abnormality handling process, the handling result and the comparison data are stored in the database.

[0050] 1. By constructing a three-dimensional geometric model of the slope anchor rod and mapping the grouting parameter data into the model, a real-time dynamic three-dimensional visual monitoring image is generated, which intuitively displays the distribution state of the grouting amount in the full length range of the anchor rod, enabling management personnel to accurately grasp the grouting process from a three-dimensional spatial perspective, and solving the limitations of traditional two-dimensional monitoring methods.

[0051] 2. The pre-processing methods such as moving average filtering, temperature compensation and abnormal value interpolation replacement effectively reduce the influence of environmental interference on grouting pressure, flow, density and other parameters, ensure the accuracy of the original data, provide reliable basis for accurate calculation of cumulative grouting amount and unit segment grouting amount, and with the help of trained image abnormality judgment model and abnormal data analysis model, abnormal conditions in the grouting process can be automatically identified, abnormal types can be quickly determined and abnormal influence time can be predicted, combined with the hierarchical early warning mechanism, early warning information can be sent in time, the abnormal response time is greatly shortened, and the engineering risk is reduced.

[0052] 3. The abnormality processing tracking mechanism records the warning response, processing process and effect evaluation, and stores the data into the database, providing data support for subsequent model optimization and construction scheme improvement, at the same time, the three-dimensional visual image and detailed abnormal information help management personnel to quickly develop processing strategies, improve the scientific nature and decision-making efficiency of construction management, the abnormal data analysis model constructed by using support vector machine algorithm has strong generalization ability through parameter optimization, which can adapt to grouting monitoring requirements under different geological conditions and construction scenes, at the same time, the establishment of the database lays a foundation for subsequent introduction of more advanced intelligent algorithms, and has good expansibility. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to facilitate those skilled in the art to understand, the present application will be further described below with reference to the accompanying drawings.

[0054] Figure 1 The overall method flowchart of the present application;

[0055] Figure 2 The step flowchart framework in the present application. DETAILED DESCRIPTION

[0056] In order to make those skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings, obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor belong to the scope of protection of the present application.

[0057] Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive of one another. As will be apparent to those of ordinary skill in the art, embodiments described herein can be combinable with other embodiments.

[0058] Referring to Figure 1 As shown in the drawings, a slope anchor grouting amount three-dimensional visualization monitoring method comprises:

[0059] Based on the grouting parameter sensors arranged in the slope anchor construction area, real-time grouting parameter data is collected, and the real-time grouting parameter data is preprocessed;

[0060] A three-dimensional geometric model of the slope anchor is constructed, the preprocessed grouting parameter data is mapped to the three-dimensional geometric model of the slope anchor, three-dimensional grouting amount distribution data is generated, a three-dimensional visualization rendering technology is used, and real-time dynamic slope anchor grouting amount three-dimensional visualization monitoring images are generated;

[0061] Based on the trained image anomaly judgment model, the three-dimensional visualization monitoring images are analyzed to determine whether an anomaly occurs, and anomaly data is output;

[0062] Based on the image anomaly judgment model anomaly data output, a trained anomaly data analysis model is used to output anomaly types and anomaly prediction times, and warning information is issued based on the anomaly types and anomaly prediction times.

[0063] Referring to Figure 2 As shown in the drawings, based on the grouting parameter sensors arranged in the slope anchor construction area, real-time grouting parameter data is collected, and the real-time grouting parameter data is preprocessed, which specifically comprises:

[0064] The real-time collected grouting pressure and grout flow data are denoised by using a moving average filtering method to remove high-frequency interference signals, wherein the moving average filtering method formula is:

[0065]

[0066] In the formula, is the filtered parameter value, x i is the i-th original parameter value in the sliding window, and n is the size of the sliding window, which is 5-15;

[0067] The grout density data is temperature compensated, the density data at different temperatures is converted to the density value at the standard temperature based on the pre-established grout density-temperature relationship curve, and the compensation formula is:

[0068] p 20 = p t + k(20-t)

[0069] wherein p 20 is the slurry density at 20℃, p t is the slurry density collected at temperature t, k is the temperature influence coefficient, and t is the real-time slurry temperature;

[0070] Abnormal values in the grouting parameter data are removed, and when the parameter value exceeds the preset threshold range, the effective value at the adjacent time is used for interpolation replacement, wherein the interpolation formula is:

[0071]

[0072] wherein x ab is the parameter value at the abnormal time after replacement, x a is the parameter value at the time before the abnormal time, x b is the parameter value at the time after the abnormal time, Δt a is the time interval between the abnormal time and the time before the abnormal time, and Δt b is the time interval between the abnormal time and the time after the abnormal time;

[0073] A multi-step cooperative data preprocessing mechanism is constructed, and differential processing methods (targeted noise reduction by sliding filtering, density deviation correction by temperature compensation, and abnormal value repair by interpolation replacement) are used for different parameter characteristics, which greatly improves the accuracy of the original data, provides a high-quality data basis for subsequent grouting quantity calculation, and solves the data deviation problem caused by the single traditional preprocessing method.

[0074] The construction of the slope anchor three-dimensional geometric model, mapping the preprocessed grouting parameter data to the slope anchor three-dimensional geometric model, generates three-dimensional grouting quantity distribution data specifically includes:

[0075] Collect the geological survey report and anchor design drawings of the slope project, obtain the topographic data of the slope, the length, diameter, spacing and inclination angle parameters of the anchor, use three-dimensional modeling software, and arrange the anchor model in three-dimensional space according to the design parameters of the anchor based on the topographic data of the slope, form a preliminary slope anchor three-dimensional geometric model, obtain the actual shape data of the slope and anchor through field three-dimensional laser scanning, correct the preliminarily constructed three-dimensional geometric model, so that the error between the model and the actual situation is within the allowable range, and the error calculation formula is:

[0076]

[0077] wherein e is the model error, (x mi , y mi , z mi ) is the coordinate value of a feature point in the model, (xai ,y ai ,z ai ) is the coordinate value of the actual corresponding feature point, and n is the total number of feature points;

[0078] According to the grouting time and the slurry flow, the cumulative grouting amount is calculated, and the formula is:

[0079]

[0080] In the formula, Q is the cumulative grouting amount, q(t) is the slurry flow at time t, t is the grouting time, and dt is the differential symbol in calculus;

[0081] The anchor rod is divided into a plurality of unit segments along the length direction, and the grouting amount of each unit segment is calculated according to the grouting time corresponding to each unit segment, and the unit segment grouting amount formula is:

[0082]

[0083] In the formula, Q i is the grouting amount of the i th unit segment, [t i-1 ,t i ] is the grouting time interval corresponding to the i th unit segment, and q(t) is the slurry flow in the time interval;

[0084] The grouting amount of each unit segment and the corresponding grouting pressure, slurry density and other parameters are associated with the corresponding unit position in the three-dimensional geometric model to form three-dimensional grouting amount distribution data;

[0085] The three-dimensional data generation logic of “design modeling-site correction-parameter mapping” is proposed, the model error is corrected by laser scanning, the spatial fine distribution of grouting amount is realized by combining unit segment division, the grouting parameters are accurately bound with the three-dimensional spatial position of the anchor rod for the first time, the problem that the grouting amount is disconnected with the physical position in the traditional method is solved, and the data foundation is laid for three-dimensional visualization.

[0086] The three-dimensional visualization rendering technology is adopted to generate real-time dynamic slope anchor rod grouting amount three-dimensional visualization monitoring image, which specifically includes:

[0087] The surface three-dimensional image of the slope anchor rod grouting amount is generated by processing the three-dimensional grouting amount distribution data by using the surface rendering algorithm;

[0088] The size of the grouting amount is mapped into different colors by color mapping, and the color mapping formula is:

[0089]

[0090] In the formula, C is the mapped color value, C min is the color corresponding to the minimum grouting amount, and C maxThe color corresponding to the maximum grouting amount, Q is the grouting amount of the current unit section, Q min The minimum grouting amount threshold, Q max The maximum grouting amount threshold;

[0091] The transparency of the image is set to enable the operator to clearly observe the internal structure of the anchor rod and the grouting amount distribution at different depths, and the transparency calculation formula is:

[0092]

[0093] In the formula, alpha is the transparency value, 0-1, 0 is completely transparent, and 1 is completely opaque, D is the current observation depth, D max The maximum observation depth, while ensuring real-time dynamic updating of the three-dimensional image, the update frequency is not less than 10Hz;

[0094] The fusion surface rendering algorithm and dynamic parameter mapping technology are combined to intuitively display the grouting amount difference through color gradient, realize internal structure observation through transparency adjustment, and realize three-dimensional dynamic visualization of the grouting process with a real-time update frequency of more than 10Hz, solving the defects of traditional two-dimensional display or static three-dimensional real-time tracking of grouting dynamics.

[0095] The trained image anomaly judgment model analyzes the three-dimensional visualization monitoring image, judges whether an anomaly occurs, and outputs abnormal data, which specifically includes:

[0096] The three-dimensional visualization monitoring image of the slope anchor rod grouting amount is collected, which contains normal grouting images and abnormal grouting images;

[0097] The collected images are labeled to mark the location, range and type of abnormal areas; a convolutional neural network model based on deep learning is constructed, the labeled images are divided into training set and test set, the training set is used to train the model, the performance of the model is evaluated through the test set, and the model accuracy calculation formula is:

[0098]

[0099] In the formula, A CC The model accuracy, TP is the number of correctly identified abnormal images, TN is the number of correctly identified normal images, FP is the number of normal images incorrectly identified as abnormal, and FN is the number of abnormal images incorrectly identified as normal. The model parameters are continuously adjusted until the recognition accuracy of the model reaches more than 90%;

[0100] Introduce deep learning convolutional neural network to recognize the abnormality of three-dimensional visualization image, realize the automatic positioning and judgment of abnormal area through large sample training and accuracy quantitative evaluation (≥90%), replace the traditional manual inspection, greatly improve the efficiency and accuracy of abnormal recognition, and solve the problem of strong subjectivity and high missing rate of artificial judgment.

[0101] The trained image abnormality judgment model analyzes the three-dimensional visualization monitoring image, judges whether an abnormality occurs, and outputs abnormal data, specifically including:

[0102] The real-time generated three-dimensional visualization monitoring image is input into the trained image abnormality judgment model;

[0103] The model extracts and analyzes the features of the image, and compares them with the normal image features stored in the model. The feature deviation calculation formula is:

[0104]

[0105] In the formula, ξ is the feature deviation value, f rj is the jth feature value of the real-time image, f nj is the jth feature value of the normal image, and m is the total number of features.

[0106] When the deviation of the features in the image from the normal features exceeds the preset threshold, it is determined that an abnormality occurs, and the three-dimensional coordinates of the abnormal area, the time when the abnormality occurs, and the corresponding grouting parameter data are output.

[0107] A feature deviation quantitative comparison mechanism is proposed, which calculates the multi-dimensional feature difference (such as grouting amount distribution gradient, color distribution entropy, etc.) between the real-time image and the normal image, realizes the objectivity and quantification of abnormality judgment, avoids the ambiguity of traditional qualitative judgment, and outputs the three-dimensional coordinates and parameter data of the abnormal area, providing accurate positioning information for subsequent analysis.

[0108] The trained image abnormality judgment model specifically includes:

[0109] Collect historical abnormal data, including grouting parameters, abnormal types, abnormal duration, treatment measures and treatment result data when the abnormality occurs; preprocess the historical abnormal data to remove invalid data and duplicate data, and standardize the data. The standardization formula is:

[0110]

[0111] In the formula, x' is the standardized parameter value, x is the original parameter value, μ is the average value of the parameter, and σ is the standard deviation of the parameter.

[0112] An anomaly data analysis model is constructed using the support vector machine algorithm. Preprocessed anomaly data is used as training samples, and anomaly type and anomaly prediction time are used as the model output. The model parameters are optimized through training so that the model can accurately predict anomaly type and anomaly prediction time.

[0113] By applying the support vector machine algorithm to anomaly data analysis and eliminating the influence of parameter dimensions through data standardization, we can achieve accurate classification of anomaly types and quantitative prediction of impact time, breaking through the limitations of traditional empirical anomaly analysis and providing data support for early warning level determination.

[0114] The specific steps of constructing the anomaly data analysis model using the support vector machine algorithm include:

[0115] The standardized abnormal data is divided into input feature vectors and output labels. The input feature vectors are the grouting pressure, grout flow rate and grout density parameters when the abnormality occurs, and the output labels are the abnormality type and the abnormality prediction time.

[0116] The radial basis function kernel is used as the kernel function for the support vector machine, and the kernel function formula is:

[0117] K(x i ,x j )=exp(-γ||x i -x j || 2 )

[0118] In the formula, K(x) i ,x j ) represents the kernel function value, x i and x j Given two distinct input feature vectors, γ is the kernel function parameter, |x i -x j | 2 Let be the Euclidean distance between the two vectors;

[0119] The penalty parameters and kernel function parameters of the support vector machine are optimized by grid search, and the average accuracy of five-fold cross-validation is used as the evaluation index to select the optimal parameter combination.

[0120] A support vector machine model with optimal parameter combinations is used to train the training samples to generate an anomaly data analysis model.

[0121] The support vector machine model construction process is optimized, and the radial basis kernel function is used to improve the fitting ability of nonlinear data. The global optimization of parameters is achieved through grid search and five-fold cross-validation, which solves the problem of insufficient generalization ability caused by the empirical setting of parameters in traditional models, so that the model can maintain high accuracy under different geological conditions.

[0122] The issuing of the early warning information based on the abnormal type and the abnormal prediction time specifically comprises:

[0123] The early warning level is determined according to the abnormal type, wherein a slight abnormality corresponds to a third-level early warning, a general abnormality corresponds to a second-level early warning, and a serious abnormality corresponds to a first-level early warning;

[0124] The early warning response time is calculated in combination with the abnormal prediction time, and the early warning response time formula is:

[0125] T = T p -T r

[0126] In the formula, T is the early warning response time, T p is the abnormal prediction time, and T r is the processing preparation time; when the abnormal prediction time is less than a preset emergency response time, an emergency early warning signal is issued;

[0127] An alarm signal is issued through a sound-light alarm device at the construction site, and early warning information is sent to a mobile terminal of a relevant management personnel, the early warning information including an abnormal type, an abnormal position, an abnormal prediction time and a suggested processing measure;

[0128] The "abnormal type + prediction time" two-dimensional early warning mechanism is designed, the emergency degree is quantified through the early warning response time, the hierarchical early warning is combined with the multi-channel information pushing (on-site sound-light + mobile terminal), the key information is ensured to be quickly acquired by the management personnel and the corresponding measures are taken, and the problems of single traditional early warning mode and unclear emergency degree are solved.

[0129] The early warning information issuing time, the receiving personnel and the response state are recorded to form an early warning response log;

[0130] The grouting parameters of the abnormal area are re-collected and analyzed every 5-15 minutes, the abnormal type and the abnormal prediction time are updated, and the updated results are synchronized to the mobile terminal of the relevant management personnel;

[0131] When the abnormality is processed, the processed grouting parameter data and the three-dimensional visual monitoring image are collected, compared with the data when the abnormality occurs, and the abnormal processing process, the processing result and the comparison data are stored in a database;

[0132] The abnormal full life cycle tracking mechanism is established, the dynamic monitoring, the processing result comparison and evaluation and the full-process data archiving are realized through the regular data updating, the closed-loop management of "early warning-processing-review" is formed, the effective processing of the abnormality is ensured, the empirical data for model iteration and construction process optimization are provided, and the continuous improvement capability of the monitoring system is improved.

[0133] In summary, the advantages of the present application are that:

[0134] The pre-processed grouting parameter data is accurately associated with the three-dimensional geometric model of the slope anchor rod, breaking through the limitations of traditional two-dimensional monitoring, and generating real-time dynamic monitoring images through three-dimensional visualization rendering technology. This fusion not only realizes the intuitive presentation of the distribution state of grouting volume in the full length range of the anchor rod, but also dynamically tracks the grouting process, allowing management personnel to real-time master the grouting situation from a three-dimensional spatial perspective, which is an important innovation in the dimension of grouting volume monitoring.

[0135] A multi-step data preprocessing system including moving average filtering, grout density temperature compensation, and abnormal value interpolation replacement is constructed. Different processing methods are used for different grouting parameters to effectively eliminate the interference of high-frequency interference, temperature influence, and abnormal values on data accuracy, providing a high-quality data basis for grouting volume calculation and solving the problem of calculation deviation caused by rough data preprocessing in traditional technology.

[0136] The dual-layer intelligent architecture of "image anomaly judgment model + abnormal data analysis model" is innovatively adopted. First, the three-dimensional visualization monitoring image is identified by the convolutional neural network model based on deep learning, and then the abnormal data analysis model constructed by the support vector machine algorithm is used to accurately output the abnormal type and prediction time. This architecture realizes the intelligent closed loop from anomaly detection to anomaly analysis, greatly improves the efficiency and accuracy of abnormal monitoring, and overcomes the lag and subjectivity of traditional manual judgment.

[0137] After the warning information is issued, an automatic tracking mechanism is introduced, including recording warning response logs, regularly updating abnormal data, evaluating processing effects, and storing relevant information in the database. This mechanism forms a full-process closed-loop management of abnormal processing, ensuring the timeliness and effectiveness of abnormal processing, providing valuable data support for subsequent model optimization and construction scheme improvement, and realizing the deep combination of monitoring and management.

[0138] In the construction of the abnormal data analysis model, the penalty parameter and kernel function parameter of the support vector machine are optimized by the grid search method, and the model performance is improved by five-fold cross-validation. This parameterized optimization design makes the model have strong generalization ability, which can adapt to different geological conditions and construction scenes of monitoring demand. At the same time, the establishment of the database reserves space for the introduction of more advanced intelligent algorithms, enhancing the adaptability and sustainability of the technology.

[0139] The preferred embodiments of the application disclosed above are only to facilitate the elucidation of the application. The preferred embodiments do not describe all the details of the application and limit the application to the specific embodiments. Obviously, many modifications and variations can be made in light of the teachings above. The description is chosen and described in order to provide the best illustration of the application and its practical application to those skilled in the art and to enable those skilled in the art to best utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. A three-dimensional visualization monitoring method for slope anchor grouting volume, characterized in that, include: Based on the grouting parameter sensors deployed in the slope anchor construction area, real-time grouting parameter data is collected, and the real-time grouting parameter data is preprocessed. A three-dimensional geometric model of the slope anchor is constructed, and the pre-processed grouting parameter data is mapped to the three-dimensional geometric model of the slope anchor to generate three-dimensional grouting volume distribution data. Three-dimensional visualization rendering technology is used to generate a real-time dynamic three-dimensional visualization monitoring image of the slope anchor grouting volume. Based on the trained image anomaly detection model, the three-dimensional visualization monitoring image is analyzed to determine whether an anomaly has occurred and anomaly data is output. Based on the image anomaly detection model, the model outputs abnormal data and uses a trained abnormal data analysis model to output the anomaly type and anomaly prediction time, and issues early warning information based on the anomaly type and anomaly prediction time.

2. The method for three-dimensional visualization monitoring of grouting volume of slope anchor bolts according to claim 1, characterized in that, The grouting parameter sensors deployed in the slope anchor construction area collect real-time grouting parameter data. The preprocessing of this real-time grouting parameter data specifically includes: The moving average filtering method is used to reduce noise in the real-time acquired grouting pressure and grout flow data, removing high-frequency interference signals. The formula for the moving average filtering method is as follows: In the formula, Here are the filtered parameter values, x i is the i-th original parameter value within the sliding window, and n is the size of the sliding window, which ranges from 5 to 15; Temperature compensation is applied to the slurry density data. Based on a pre-established slurry density-temperature relationship curve, the density data at different temperatures are converted to density values ​​at the standard temperature. The compensation formula is as follows: r 20 =ρ t +k(20-t) In the formula, ρ 20 ρ is the density of the slurry at 20℃. t denoted as slurry density collected at temperature t, where k is the temperature influence coefficient and t is the real-time slurry temperature. Outliers in the grouting parameter data are removed. When a parameter value exceeds a preset threshold range, the valid value from an adjacent time point is used for interpolation and replacement. The interpolation formula is as follows: In the formula, x ab The replaced parameter value at the time of the anomaly, x a x is the parameter value at the moment preceding the abnormal moment. b The parameter value Δt is the value at the time following the abnormal moment. a Δt represents the time interval between the abnormal moment and the previous moment. b This represents the time interval between the abnormal moment and the next moment.

3. The method for three-dimensional visualization monitoring of grouting volume of slope anchor bolts according to claim 1, characterized in that, The process of constructing a three-dimensional geometric model of the slope anchor, mapping the preprocessed grouting parameter data to the three-dimensional geometric model of the slope anchor, and generating three-dimensional grouting volume distribution data specifically includes: Collect geological survey reports and anchor bolt design drawings for slope engineering, and obtain topographic data of the slope, as well as the length, diameter, spacing, and inclination angle parameters of the anchor bolts; Using 3D modeling software, based on the topographic data of the slope, the anchor model is laid out in 3D space according to the design parameters of the anchor, forming a preliminary 3D geometric model of the slope anchor. The actual shape data of the slope and anchor bolts were obtained by on-site 3D laser scanning. The preliminary 3D geometric model was then corrected to ensure that the error between the model and the actual situation was within the allowable range. The error calculation formula is as follows: In the formula, e is the model error, (x mi ,y mi ,z mi (x) represents the coordinates of a feature point in the model. ai ,y ai ,z ai ) represents the coordinates of the actual corresponding feature point, and n represents the total number of feature points; The cumulative grouting volume is calculated based on the grouting time and grout flow rate, using the following formula: In the formula, Q is the cumulative grouting volume, q(t) is the grout flow rate at time t, t is the grouting time, and dt is the differential symbol in calculus; The anchor bolt is divided into several unit segments along its length. Based on the grouting time corresponding to each unit segment, the grouting volume of each unit segment is calculated. The formula for the grouting volume of each unit segment is as follows: In the formula, Q i Let [t] be the grouting volume of the i-th unit segment. i-1 ,t i ] represents the grouting time interval corresponding to the i-th unit segment, and q(t) represents the grout flow rate within this time interval; The grouting volume of each unit segment and its corresponding grouting pressure, grout density, and other parameters are associated with the corresponding unit positions in the three-dimensional geometric model to form three-dimensional grouting volume distribution data.

4. The method for three-dimensional visualization monitoring of grouting volume of slope anchor bolts according to claim 1, characterized in that, The specific steps of generating a real-time dynamic three-dimensional visualization monitoring image of the slope anchor grouting volume using three-dimensional visualization rendering technology include: A surface rendering algorithm is used to process the three-dimensional grouting volume distribution data to generate a surface three-dimensional image of the slope anchor grouting volume; The amount of grout injected is mapped to different colors using color mapping. The color mapping formula is as follows: In the formula, C is the mapped color value, C min C is the color corresponding to the minimum grouting volume. max The color corresponds to the maximum grouting volume, and Q represents the grouting volume of the current unit segment. min Q is the minimum grouting volume threshold. max The maximum grouting volume threshold; The image transparency is set to allow operators to clearly observe the interior of the anchor bolt and the distribution of grout volume at different depths. The transparency calculation formula is as follows: In the formula, α is the transparency value, ranging from 0 to 1, where 0 represents complete transparency and 1 represents complete opacity, and D is the current viewing depth. max To maximize the observation depth while ensuring real-time dynamic updates of the 3D image, the update frequency is no less than 10Hz.

5. The method for three-dimensional visualization monitoring of grouting volume of slope anchor bolts according to claim 1, characterized in that, The trained image anomaly detection model analyzes the 3D visualization monitoring image to determine whether an anomaly has occurred, and outputs anomaly data, specifically including: Collect three-dimensional visualization monitoring images of slope anchor grouting volume, including normal grouting images and abnormal grouting images; The acquired images are labeled to mark the location, extent, and type of abnormality. A deep learning-based convolutional neural network model is constructed. The labeled images are divided into a training set and a test set. The model is trained using the training set and its performance is evaluated using the test set. The model accuracy is calculated using the following formula: In the formula, A CC Here, TP represents the model accuracy, TN represents the number of correctly identified anomalous images, FP represents the number of correctly identified normal images, and FN represents the number of incorrectly identified normal images. The model parameters are continuously adjusted until the model's recognition accuracy reaches over 90%.

6. The method for three-dimensional visualization monitoring of grouting volume of slope anchor bolts according to claim 1, characterized in that, The trained image anomaly detection model analyzes the 3D visualization monitoring image to determine whether an anomaly has occurred, and outputs anomaly data, specifically including: The real-time generated 3D visualization monitoring images are input into the trained image anomaly detection model; The model extracts and analyzes features from the image, compares them with the features of normal images stored in the model, and calculates the feature deviation using the following formula: In the formula, ξ is the characteristic deviation value, f rj Let f be the j-th feature value of the real-time image. nj Let j be the j-th feature value of the normal image, and m be the total number of features; When the deviation between the features in the image and normal features exceeds a preset threshold, it is determined that an anomaly has occurred, and the three-dimensional coordinates of the abnormal area, the time of the anomaly, and the corresponding grouting parameter data are output.

7. The method for three-dimensional visualization monitoring of grouting volume of slope anchor bolts according to claim 6, characterized in that, The trained image anomaly detection model specifically includes: Collect historical anomaly data, including grouting parameters at the time of the anomaly, anomaly type, anomaly duration, handling measures, and handling results. Historical outlier data is preprocessed to remove invalid and duplicate data, and then standardized using the following formula: In the formula, x′ is the standardized parameter value, x is the original parameter value, μ is the mean of the parameter, and σ is the standard deviation of the parameter; An anomaly data analysis model is constructed using the support vector machine algorithm. Preprocessed anomaly data is used as training samples, and the anomaly type and anomaly prediction time are used as the model outputs. The model parameters are optimized through training, enabling the model to accurately predict anomaly types and anomaly prediction times.

8. The method for three-dimensional visualization monitoring of grouting volume of slope anchor bolts according to claim 7, characterized in that, The specific steps involved in constructing the anomaly data analysis model using the support vector machine algorithm are as follows: The standardized abnormal data is divided into input feature vectors and output labels. The input feature vectors are the grouting pressure, grout flow rate and grout density parameters when the abnormality occurs, and the output labels are the abnormality type and the abnormality prediction time. The radial basis function kernel is used as the kernel function for the support vector machine, and the kernel function formula is: K(x i ,x j )=exp(-γ||x i -x j || 2 ) In the formula, K(x) i ,x j ) represents the kernel function value, x i and x j Given two distinct input feature vectors, γ is the kernel function parameter, |x i -x j | 2 Let be the Euclidean distance between the two vectors; The penalty parameters and kernel function parameters of the support vector machine are optimized by grid search, and the average accuracy of five-fold cross-validation is used as the evaluation index to select the optimal parameter combination. A support vector machine model with optimal parameter combinations is used to train the training samples to generate an anomaly data analysis model.

9. A three-dimensional visualization monitoring method for slope anchor grouting volume according to claim 6, characterized in that, The issuance of early warning information based on anomaly type and anomaly prediction time specifically includes: The warning level is determined based on the type of anomaly, with minor anomalies corresponding to Level 3 warnings, general anomalies corresponding to Level 2 warnings, and severe anomalies corresponding to Level 1 warnings. The early warning response time is calculated by combining the anomaly prediction time with the early warning response time formula: T=T p -T r In the formula, T is the early warning response time, T p For the time of abnormal prediction, T r To allow for preparation time, an emergency warning signal will be issued when the anomaly prediction time is less than the preset emergency response time. An alarm signal is issued through the sound and light alarm device at the construction site, and the warning information is sent to the mobile terminal of relevant management personnel. The warning information includes the type of abnormality, the location of the abnormality, the predicted time of the abnormality, and the suggested handling measures.

10. A three-dimensional visualization monitoring method for slope anchor grouting volume according to claim 9, characterized in that: Record the time when the warning information is issued, the recipients, and the response status to create a warning response log; Every 5-15 minutes, the grouting parameters of the abnormal area are re-collected and analyzed, the abnormality type and abnormality prediction time are updated, and the updated results are synchronized to the mobile terminals of relevant management personnel. After the anomaly is handled, the grouting parameter data and three-dimensional visualization monitoring images after the handling are collected and compared with the data when the anomaly occurred. The anomaly handling process, handling results and comparison data are stored in the database.

Images