In-situ injection method and device of barrier material

Abstract

The invention relates to the technical field of barrier construction, in particular to an in-situ injection method and device for a barrier material. An injection scheme is optimized through a computer simulation system, standard data are obtained, actual data are obtained through the Internet of Things in the actual injection process, the actual data are compared with the standard data to analyze the working state, and an unexpected state is evaluated, adjusted and controlled through a gradient lifting tree model constructed based on historical feature data. The injection accuracy and effectiveness of the barrier material can be improved, the injection scheme is simulated and optimized in advance, the unexpected state is monitored in real time and adjusted in time, errors and faults in the injection process are reduced, and the injection effect of the barrier material in the target area is ensured.

Description

Technical Field

[0001] The invention relates to the technical field of barrier construction, in particular to an in-situ injection method and device of a barrier material. Background Art

[0002] With the development of industries, agriculture, and environmental protection, many engineering projects involve tasks such as groundwater pollution prevention and control and soil remediation. Especially in terms of the barrier and isolation of harmful substances, the demand is gradually increasing. Although traditional barrier material injection technology has been applied in some projects, there are still a series of problems such as difficulty in uniformly controlling the injection effect, material waste, and long construction period. Therefore, the development of a more accurate, stable, and efficient in-situ injection technology has become a research hotspot in the industry.

[0003] At present, common barrier materials include alkali-activated gel and modified geomembrane barrier materials. These materials usually have good physical and chemical properties and can effectively form barriers to prevent the spread of pollutants. However, the traditional barrier material injection method mainly relies on manual operation, which makes it difficult to achieve high-precision and high-efficiency injection control. Therefore, how to achieve precise in-situ injection and ensure that these materials can form a uniform and lasting barrier effect in the target area is still a technical difficulty. In addition, the traditional injection device has a complex structure and cumbersome operation, which makes it difficult to meet the needs of modern engineering for efficiency and intelligence. Summary of the invention

[0004] The present invention overcomes the shortcomings of the prior art and provides an in-situ injection method and device for a barrier material.

[0005] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is:

[0006] The present invention discloses an in-situ injection method of a barrier material, comprising the following steps:

[0007] Obtaining a preset injection plan for the barrier material, importing the preset injection plan into a computer simulation system for simulated injection evaluation, and outputting a final injection plan for the in-situ injection device and a standard dynamic injection data pool during the working process after the simulation result satisfies the preset requirements;

[0008] The final injection plan is transmitted to the controller of the in-situ injection device, and the in-situ injection device is controlled to inject the barrier material into the target area according to the final injection plan;

[0009] During the injection process, actual injection data of the in-situ injection device at each preset injection time point is obtained based on the Internet of Things; and standard injection data of the in-situ injection device at each preset injection time point is obtained in the standard dynamic injection data pool;

[0010] Analyze the working state of the in-situ injection device according to the actual injection data and the standard injection data at each preset injection time point; if the working state of the in-situ injection device is the expected state, there is no need to correct the injection data of the in-situ injection device;

[0011] Acquire historical feature data corresponding to various abnormal operation events of the in-situ injection device, and construct a gradient boosting tree model according to the historical feature data;

[0012] If the working state of the in-situ injection device is unexpected, the actual injection data of the in-situ injection device at each preset injection time point is imported into the gradient boosting tree model for evaluation, and the in-situ injection device is adjusted and controlled according to the evaluation result; wherein, the barrier material includes alkali-activated gel and modified geomembrane barrier material.

[0013] Furthermore, a preset injection scheme of the barrier material is obtained, and the preset injection scheme is imported into a computer simulation system for simulated injection evaluation, until the simulation result satisfies the preset requirements, and the final injection scheme of the in-situ injection device and its standard dynamic injection data pool during the working process are output, specifically:

[0014] Obtain a three-dimensional structural model diagram of the in-situ injection device and a regional characteristic model diagram of the target area, and construct a computer simulation system for injecting barrier materials into the target area based on the three-dimensional structural model diagram of the in-situ injection device and the regional characteristic model diagram of the target area in combination with computer simulation software; wherein the target area is the area to be blocked;

[0015] Obtaining a preset injection scheme for the barrier material, importing the preset injection scheme into the computer simulation system for injection simulation analysis, and outputting the simulation results after the simulation is completed;

[0016] Obtaining in the simulation result a predicted permeability coefficient of the barrier material after the barrier material is simulated and injected into the target area through the preset injection scheme; comparing the predicted permeability coefficient with a preset threshold value;

[0017] If the predicted permeability coefficient is greater than a preset threshold, there is no need to adjust and optimize the preset injection plan, and the preset injection plan is output as the final injection plan;

[0018] If the predicted permeability coefficient is not greater than the preset threshold, a genetic algorithm is introduced to iteratively adjust and optimize the preset injection scheme. After each iterative optimization cycle, an optimized injection scheme is obtained, and the predicted permeability coefficient of the optimized injection scheme is evaluated by a simulation system until the predicted permeability coefficient of the optimized injection scheme is greater than the preset threshold. Then, the iteration is stopped and the optimized injection scheme is output as the final injection scheme.

[0019] After the final injection plan is output, the standard injection data of the in-situ injection device based on the time series when the in-situ injection device is controlled by the final injection plan to perform simulation work is obtained in the computer simulation system, and the various standard injection data are aggregated to obtain a standard dynamic injection data pool of the in-situ injection device during the injection work.

[0020] Furthermore, the working state of the in-situ injection device is analyzed according to the actual injection data and the standard injection data at each preset injection time point, specifically:

[0021] Constructing a dynamic transformation curve graph of actual injection data according to actual injection data of the in-situ injection device at each preset injection time point; and constructing a dynamic transformation curve graph of standard injection data according to standard injection data of the in-situ injection device at each preset injection time point;

[0022] Performing coincidence analysis on the actual injection data dynamic transformation curve graph and the standard injection data dynamic transformation curve graph to obtain the curve coincidence degree between the actual injection data dynamic transformation curve graph and the standard injection data dynamic transformation curve graph;

[0023] Comparing the curve overlap between the actual injection data dynamic transformation curve diagram and the standard injection data dynamic transformation curve diagram with a preset overlap;

[0024] If the curve overlap between the actual injection data dynamic transformation curve diagram and the standard injection data dynamic transformation curve diagram is greater than the preset overlap, it means that the actual working state of the in-situ injection device is consistent with the expected working state, and the injection effect is in line with expectations, then the working state of the in-situ injection device is calibrated as the expected state;

[0025] If the curve overlap between the actual injection data dynamic transformation curve and the standard injection data dynamic transformation curve is not greater than the preset overlap, it means that the actual working state of the in-situ injection device deviates from the expected working state, and the injection effect does not meet expectations, then the working state of the in-situ injection device is calibrated as an unexpected state.

[0026] Furthermore, historical feature data corresponding to various abnormal operation events of the in-situ injection device are obtained, and a gradient boosting tree model is constructed according to the historical feature data, specifically:

[0027] Obtain historical characteristic data corresponding to various abnormal operation events of the in-situ injection device through the big data network;

[0028] The collected historical characteristic data of abnormal events in the operation of the in-situ injection device are sorted into a structured table form, where each record corresponds to an abnormal event name and a characteristic value of the corresponding historical characteristic data;

[0029] The eigenvalues ​​in the table are extracted to form a feature matrix X, and the abnormal results are formed into a label vector y; where X contains the eigenvalues ​​used to train the model, and y contains the corresponding abnormal labels;

[0030] Introduce a gradient boosting tree algorithm, initialize a gradient boosting tree model, and set the number of iterations; use the gradient boosting framework and combine the feature matrix X to iteratively construct a decision tree, and correct the residual of the previous decision tree according to the corresponding label vector y to form a strong learner;

[0031] At the same time, in each iteration, the gradient and second-order derivative of the loss function are calculated and used as the input of the new weak learner to update the model to minimize the loss function;

[0032] After reaching the number of iterations, a gradient boosting tree model consisting of multiple decision trees is constructed.

[0033] Furthermore, if the working state of the in-situ injection device is unexpected, the actual injection data of the in-situ injection device at each preset injection time point is imported into the gradient boosting tree model for evaluation, and the in-situ injection device is adjusted and controlled according to the evaluation result, specifically:

[0034] If the working state of the in-situ injection device is an unexpected state, obtaining actual injection data of the in-situ injection device at each preset injection time point;

[0035] Importing the actual injection data of the in-situ injection device at each preset injection time point into the gradient boosting tree model to diagnose the in-situ injection device in an unexpected state, and obtaining the name of the abnormal event currently occurring in the in-situ injection device in the unexpected state;

[0036] Determine the characteristic type of the name of the abnormal event currently occurring in the in-situ injection device in an unexpected state; wherein the characteristic type of the name of the abnormal event includes an uncontrollable abnormal operation event and a controllable abnormal operation event;

[0037] If the name of the abnormal event currently occurring in the in-situ injection device in an unexpected state is an uncontrollable abnormal operation event, the in-situ injection device is controlled to suspend the injection work, and the name of the abnormal event currently occurring in the in-situ injection device is sent to the preset terminal for display.

[0038] Furthermore, if the working state of the in-situ injection device is an unexpected state, the actual injection data of the in-situ injection device at each preset injection time point is imported into the gradient boosting tree model for evaluation, and the in-situ injection device is adjusted and controlled according to the evaluation result, which also includes the following steps:

[0039] Obtaining control log information of the in-situ injection device, obtaining adjustment measures corresponding to various deviations of various injection data in the in-situ injection device according to the control log information, and obtaining the adjustment accuracy after adjusting the injection data with deviations through various adjustment measures;

[0040] Obtaining the adjustment measure corresponding to the maximum adjustment accuracy, taking the adjustment measure corresponding to the maximum adjustment accuracy as the best adjustment measure for the corresponding deviation of the corresponding injection data, and so on, obtaining the best adjustment measures corresponding to various deviations of the injection data in the in-situ injection device;

[0041] Constructing a knowledge matching library, and importing the best adjustment measures corresponding to various deviations of various injection data in the in-situ injection device into the knowledge matching library;

[0042] If the abnormal event currently occurring in the in-situ injection device in an unexpected state is named as an adjustable abnormal operation event, then obtaining various real-time injection data of the in-situ injection device at the current injection time point; and obtaining various standard injection data of the in-situ injection device at the current injection time point in the standard dynamic injection data pool;

[0043] Calculate the data deviation value between various real-time injection data of the in-situ injection device at the current injection time point and the corresponding standard injection data;

[0044] Marking the real-time injection data whose data deviation value is greater than the preset difference value as operation deviation injection data, and obtaining the data deviation value between the operation deviation injection data and the corresponding standard injection data;

[0045] The data deviation value between the operation deviation injection data and the corresponding standard injection data is imported into the knowledge matching library for matching, and the best adjustment measure is obtained by matching. The best adjustment measure obtained by matching is sent to the controller to correct the operation deviation injection data.

[0046] Another aspect of the present invention discloses an in-situ injection device for barrier materials, which is applied to any one of the in-situ injection methods for barrier materials, wherein the in-situ injection device comprises a material tank and a syringe, wherein the material tank and the syringe are connected via a feeding pipe;

[0047] The injector comprises a base, and hydraulic cylinders are arranged on both sides of the base, the output end of the hydraulic cylinder is connected with a pull rod, the top end of the pull rod is connected with a mounting platform, a material pump is arranged on the mounting platform, the input end of the material pump is connected with the feeding pipe, the output end of the material pump is connected with a steel pipe, and the bottom end of the steel pipe is connected with an injection head;

[0048] The syringe is also provided with a pressure gauge and a controller.

[0049] The present invention solves the technical defects existing in the background technology, and has the following beneficial effects: the in-situ injection method of the barrier material optimizes the injection scheme and obtains standard data through a computer simulation system, obtains actual data with the help of the Internet of Things during the actual injection process, and compares and analyzes the working status with the standard data. For unexpected conditions, a gradient boosting tree model constructed based on historical feature data is used to evaluate and adjust the control, which can improve the accuracy and effectiveness of the barrier material injection, optimize the injection scheme in advance through simulation, monitor and adjust unexpected conditions in real time, reduce errors and failures in the injection process, and ensure the injection effect of the barrier material in the target area. It is particularly suitable for the in-situ injection of alkali-activated gel and modified geomembrane barrier materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, drawings of other embodiments can be obtained based on these drawings without paying creative work.

[0051] Figure 1 A first method flow chart of a method for in-situ injection of a barrier material;

[0052] Figure 2 The present invention is a method flow chart of an in-situ injection method of a barrier material;

[0053] Figure 3 This is a schematic diagram of the structure of an in-situ injection device for barrier materials. DETAILED DESCRIPTION

[0054] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0055] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0056] like Figure 1 As shown, the present invention discloses an in-situ injection method of a barrier material, comprising the following steps:

[0057] S102, obtaining a preset injection plan for the barrier material, importing the preset injection plan into a computer simulation system for simulated injection evaluation, and outputting a final injection plan of the in-situ injection device and a standard dynamic injection data pool during the working process after the simulation result satisfies the preset requirements;

[0058] S104, transmitting the final injection plan to the controller of the in-situ injection device, and controlling the in-situ injection device to inject the barrier material into the target area according to the final injection plan;

[0059] S106. During the injection process, actual injection data of the in-situ injection device at each preset injection time point is obtained based on the Internet of Things; and standard injection data of the in-situ injection device at each preset injection time point is obtained from the standard dynamic injection data pool;

[0060] S108, analyzing the working state of the in-situ injection device according to the actual injection data and the standard injection data at each preset injection time point; if the working state of the in-situ injection device is the expected state, there is no need to correct the injection data of the in-situ injection device;

[0061] S110, acquiring historical feature data corresponding to various abnormal operation events occurring in the in-situ injection device, and constructing a gradient boosting tree model according to the historical feature data;

[0062] S112. If the working state of the in-situ injection device is an unexpected state, the actual injection data of the in-situ injection device at each preset injection time point is imported into the gradient boosting tree model for evaluation, and the in-situ injection device is adjusted and controlled according to the evaluation result; wherein the barrier material includes alkali-activated gel and modified geomembrane barrier material.

[0063] Among them, the final injection plan is a set of static parameter configurations to guide actual operations, aiming to ensure that the barrier effect meets the standards; and the standard dynamic working data pool is a set of dynamic data generated when simulating the operation of the device in a simulation environment based on the plan, reflecting the expected performance of the device in the process of executing the plan, providing reference and monitoring basis for actual operations.

[0064] It should be noted that the preset injection scheme of the barrier material is first obtained and imported into the computer simulation system for simulated injection evaluation. In this process, the computer simulation system will simulate the virtual injection operation of the preset injection scheme according to the characteristics of the in-situ injection device and the relevant parameters of the target area. The preset injection scheme includes parameters such as injection speed, injection pressure, and injection volume changes over time. Only when the simulation results meet the preset requirements will the final injection scheme of the in-situ injection device and the corresponding standard dynamic injection data pool be output. The standard dynamic injection data pool is a collection of standard injection data recorded in time series when simulating the final injection scheme, and these data will be used as a reference standard in the subsequent actual injection process. The final injection scheme obtained is transmitted to the controller of the in-situ injection device. The controller controls the in-situ injection device to inject the barrier material into the target area according to this scheme. In this process, the in-situ injection device operates according to the parameters set in the final injection scheme to inject the barrier material (such as alkali-activated gel and modified geomembrane barrier material) into the target area. During the injection work, the actual injection data of the in-situ injection device at each preset injection time point is obtained by using the Internet of Things. At the same time, the standard injection data of the same preset injection time point is obtained from the standard dynamic injection data pool obtained previously. Then, the working state of the in-situ injection device is analyzed based on these two sets of data. If the actual injection data is consistent with the standard injection data, that is, the working state of the in-situ injection device is the expected state, it means that the injection work is proceeding normally and there is no need to correct the injection data. When the working state of the in-situ injection device is an unexpected state, the actual injection data of each preset injection time point is imported into this gradient boosting tree model for evaluation. The model will determine the possible abnormal conditions of the device based on the input data, and then adjust and control the in-situ injection device based on the evaluation results, so that the working state of the device can be restored to the expected state as much as possible.

[0065] The in-situ injection method of the barrier material optimizes the injection plan and obtains standard data through a computer simulation system. In the actual injection process, the Internet of Things is used to obtain actual data and compare and analyze the working status with the standard data. For unexpected conditions, a gradient boosting tree model constructed based on historical feature data is used to evaluate and adjust the control. This can improve the accuracy and effectiveness of the barrier material injection, optimize the injection plan in advance through simulation, monitor and adjust unexpected conditions in real time, reduce errors and failures in the injection process, and ensure the injection effect of the barrier material in the target area. It is particularly suitable for the in-situ injection of alkali-activated gel and modified geomembrane barrier materials.

[0066] Further, a preset injection scheme of the barrier material is obtained, and the preset injection scheme is imported into a computer simulation system for simulated injection evaluation, until the simulation result satisfies the preset requirements, and the final injection scheme of the in-situ injection device and its standard dynamic injection data pool during the working process are output, such as Figure 2 As shown, specifically:

[0067] S202, obtaining a three-dimensional structural model diagram of the in-situ injection device and a regional characteristic model diagram of the target area, and constructing a computer simulation system for injecting barrier materials into the target area based on the three-dimensional structural model diagram of the in-situ injection device and the regional characteristic model diagram of the target area in combination with computer simulation software; wherein the target area is the area to be blocked;

[0068] S204, obtaining a preset injection scheme for the barrier material, importing the preset injection scheme into the computer simulation system for injection simulation analysis, and outputting the simulation results after the simulation is completed;

[0069] S206, obtaining in the simulation result a predicted permeability coefficient of the barrier material after the barrier material is simulated and injected into the target area through the preset injection scheme; and comparing the predicted permeability coefficient with a preset threshold value;

[0070] S208: If the predicted permeability coefficient is greater than a preset threshold, there is no need to adjust and optimize the preset injection plan, and the preset injection plan is output as a final injection plan;

[0071] S210, if the predicted permeability coefficient is not greater than the preset threshold, a genetic algorithm is introduced to iteratively adjust and optimize the preset injection scheme, and after each iterative optimization cycle is completed, an optimized injection scheme is obtained, and the predicted permeability coefficient of the optimized injection scheme is evaluated by a simulation system until the predicted permeability coefficient of the optimized injection scheme is greater than the preset threshold, then the iteration is stopped, and the optimized injection scheme is output as the final injection scheme;

[0072] S212. After the final injection plan is output, the standard injection data of the in-situ injection device based on the time series when the in-situ injection device is controlled by the final injection plan to perform simulation work is obtained in the computer simulation system, and each standard injection data is aggregated to obtain a standard dynamic injection data pool of the in-situ injection device during the injection work.

[0073] Among them, the preset injection plan is designed by relevant technical personnel based on their experience; the regional characteristic model diagram is a graphical model representation that can reflect the target area (area to be blocked) in terms of geological structure (such as soil type, stratum distribution, etc.), spatial form (shape, size, slope, etc.), physical properties (such as porosity, permeability, etc.) and other relevant characteristics that may affect the injection effect of the barrier material.

[0074] Specifically, the three-dimensional structural model diagram of the in-situ injection device and the regional characteristic model diagram of the target area (area to be blocked) are obtained, and then a computer simulation system specifically used to simulate the injection of barrier materials is constructed with the help of computer simulation software (such as ANSYS software, COMSOL Multiphysics software, etc.). This is the basis of the entire simulation evaluation work. By integrating the relevant models of the device and the target area, the injection process can be accurately simulated in a virtual environment. The preset injection scheme of the barrier material is input into the constructed computer simulation system for injection simulation analysis, and the simulation results are obtained after the simulation is completed. In this simulation result, the focus is on the predicted permeability coefficient after the barrier material simulation injection is performed on the target area through the preset injection scheme. This predicted permeability coefficient is a key indicator for measuring the effectiveness of the injection scheme. It is compared with the preset threshold value. If the predicted permeability coefficient is greater than the preset threshold value, it means that the preset injection scheme has met the requirements and does not need to be adjusted. It is directly output as the final injection scheme. If the predicted permeability coefficient is not greater than the preset threshold value, it is necessary to introduce a genetic algorithm to iteratively adjust and optimize the preset injection scheme. Genetic algorithm is an optimization algorithm that simulates the biological evolution process to find the optimal solution. After each iteration optimization cycle is completed, the optimized injection scheme is obtained, and the predicted permeability coefficient of the optimized injection scheme is evaluated by the simulation system. This process is repeated until the predicted permeability coefficient of the optimized injection scheme is greater than the preset threshold. At this time, the iteration is stopped and the optimized injection scheme is output as the final injection scheme. When the final injection scheme is determined, the standard injection data of the in-situ injection device based on the time series is obtained in the computer simulation system when the in-situ injection device is controlled by the final injection scheme to perform simulation work. By gathering these standard injection data, a standard dynamic injection data pool of the in-situ injection device during the injection process is obtained. This data pool provides a standard reference for data comparison and analysis in the subsequent actual injection process.

[0075] Through the above series of steps, the preset injection scheme can be comprehensively and carefully evaluated and optimized using the computer simulation system before the actual barrier material injection work is carried out. The computer simulation system is constructed to combine the model of the target area and the injection device, so that the simulation is more in line with the actual situation. Through the analysis of the predicted permeability coefficient and the optimization of the genetic algorithm, it is ensured that the final injection scheme can meet the requirements of the barrier effect, that is, to ensure that the permeability coefficient of the barrier material in the target area reaches the expected standard. The establishment of the standard dynamic injection data pool provides an accurate reference standard for the subsequent monitoring and adjustment of the actual injection process, which helps to improve the accuracy, effectiveness and reliability of the in-situ injection work, thereby realizing efficient and accurate in-situ injection of barrier materials.

[0076] Furthermore, the working state of the in-situ injection device is analyzed according to the actual injection data and the standard injection data at each preset injection time point, specifically:

[0077] Constructing a dynamic transformation curve graph of actual injection data according to actual injection data of the in-situ injection device at each preset injection time point; and constructing a dynamic transformation curve graph of standard injection data according to standard injection data of the in-situ injection device at each preset injection time point;

[0078] Performing coincidence analysis on the actual injection data dynamic transformation curve graph and the standard injection data dynamic transformation curve graph to obtain the curve coincidence degree between the actual injection data dynamic transformation curve graph and the standard injection data dynamic transformation curve graph;

[0079] Comparing the curve overlap between the actual injection data dynamic transformation curve diagram and the standard injection data dynamic transformation curve diagram with a preset overlap;

[0080] If the curve overlap between the actual injection data dynamic transformation curve diagram and the standard injection data dynamic transformation curve diagram is greater than the preset overlap, it means that the actual working state of the in-situ injection device is consistent with the expected working state, and the injection effect is in line with expectations, then the working state of the in-situ injection device is calibrated as the expected state;

[0081] If the curve overlap between the actual injection data dynamic transformation curve and the standard injection data dynamic transformation curve is not greater than the preset overlap, it means that the actual working state of the in-situ injection device deviates from the expected working state, and the injection effect does not meet expectations, then the working state of the in-situ injection device is calibrated as an unexpected state.

[0082] Specifically, first, a dynamic transformation curve of actual injection data is constructed for the actual injection data of the in-situ injection device at each preset injection time point. This curve can intuitively show the dynamic change trend of the actual injection data (such as the change of injection volume, injection pressure, etc. over time) during the entire injection process. At the same time, a dynamic transformation curve of standard injection data is constructed according to the standard injection data of the same preset injection time point. The curve represents the change of injection data over time that the injection device should achieve under ideal conditions. The constructed dynamic transformation curve of actual injection data and the dynamic transformation curve of standard injection data are analyzed for coincidence, with the purpose of calculating the curve coincidence between the two. This curve coincidence is a quantitative indicator used to measure the similarity between the actual injection process and the ideal injection process. If the curve coincidence is greater than the preset coincidence, it means that the actual injection process is closer to the ideal situation, and the actual working state of the in-situ injection device is consistent with the expected working state, indicating that the injection effect is in line with expectations. At this time, the working state of the in-situ injection device is calibrated as the expected state. On the contrary, if the curve overlap is not greater than the preset overlap, it means that there is a large deviation between the actual injection process and the ideal situation, the actual working state of the in-situ injection device deviates from the expected working state, and the injection effect does not meet expectations, thereby calibrating the working state of the in-situ injection device as an unexpected state.

[0083] By constructing a dynamic transformation curve diagram of actual and standard injection data, calculating the overlap between the two curves, and comparing them with the preset overlap, the working state of the in-situ injection device can be determined. The working state of the in-situ injection device can be evaluated in an intuitive and quantitative way, and the difference between the actual injection process and the ideal state can be discovered in time, thereby providing a basis for subsequent adjustment or optimization of the injection device, ensuring that the injection work is carried out as expected, and improving the accuracy and effectiveness of the in-situ injection of barrier materials.

[0084] Furthermore, historical feature data corresponding to various abnormal operation events of the in-situ injection device are obtained, and a gradient boosting tree model is constructed according to the historical feature data, specifically:

[0085] Obtain historical characteristic data corresponding to various abnormal operation events of the in-situ injection device through the big data network;

[0086] The collected historical characteristic data of abnormal events in the operation of the in-situ injection device are sorted into a structured table form, where each record corresponds to an abnormal event name and a characteristic value of the corresponding historical characteristic data;

[0087] The eigenvalues ​​in the table are extracted to form a feature matrix X, and the abnormal results are formed into a label vector y; where X contains the eigenvalues ​​used to train the model, and y contains the corresponding abnormal labels;

[0088] Introduce a gradient boosting tree algorithm, initialize a gradient boosting tree model, and set the number of iterations; use the gradient boosting framework and combine the feature matrix X to iteratively construct a decision tree, and correct the residual of the previous decision tree according to the corresponding label vector y to form a strong learner;

[0089] At the same time, in each iteration, the gradient and second-order derivative of the loss function are calculated and used as the input of the new weak learner to update the model to minimize the loss function;

[0090] After reaching the number of iterations, a gradient boosting tree model consisting of multiple decision trees is constructed.

[0091] Specifically, the historical feature data of various abnormal operation events of the in-situ injection device are first obtained through the big data network. These data come from a wide range of sources and contain various relevant information when the device has abnormal conditions in the past operation process. Then these data are organized into a structured table form, in which each record clearly corresponds to an abnormal event name and the characteristic value of the historical feature data related to it. This arrangement method makes the data have a clear structure, which is convenient for subsequent model construction. The characteristic values ​​are extracted from the organized table to form a characteristic matrix X, and the abnormal results are formed into a label vector y. Here, the characteristic matrix X contains various characteristic values ​​used to train the model, such as the values ​​of relevant parameters such as injection pressure, injection speed, and equipment temperature when abnormal events occur, and the label vector y contains the corresponding abnormal labels, such as abnormal types such as blockage and leakage. Then the gradient boosting tree algorithm is introduced, the gradient boosting tree model is initialized and the number of iterations is set. The gradient boosting tree algorithm is an ensemble learning algorithm based on decision trees, which improves the accuracy of the model by iteratively constructing decision trees. When building the model, the gradient boosting framework is used to iteratively construct the decision tree in combination with the characteristic matrix X. After each decision tree is constructed, the residual of the previous decision tree is corrected according to the corresponding label vector y, so that a strong learner is gradually constructed. At the same time, in each iteration, the gradient and second-order derivative of the loss function are calculated and used as the input of the new weak learner (i.e., the new decision tree), and the loss function is minimized by continuously updating the model. This process is repeated until the preset number of iterations is reached, and finally a gradient boosting tree model composed of multiple decision trees is constructed. This model can comprehensively consider the relationship between multiple eigenvalues ​​and accurately predict and classify the abnormal operation of the in-situ injection device.

[0092] By acquiring the historical characteristic data of abnormal operation events of the in-situ injection device, a gradient boosting tree model is constructed. This method can make full use of the information in the historical data and integrate and analyze various characteristic values ​​that may affect the operation of the device. The constructed gradient boosting tree model can accurately classify and predict the abnormal operation of the in-situ injection device, providing an effective tool for timely discovery and resolution of problems in the operation of the device, helping to improve the stability and reliability of the operation of the in-situ injection device and ensure the smooth operation of the barrier material injection work.

[0093] Furthermore, if the working state of the in-situ injection device is unexpected, the actual injection data of the in-situ injection device at each preset injection time point is imported into the gradient boosting tree model for evaluation, and the in-situ injection device is adjusted and controlled according to the evaluation result, specifically:

[0094] If the working state of the in-situ injection device is an unexpected state, obtaining actual injection data of the in-situ injection device at each preset injection time point;

[0095] Importing the actual injection data of the in-situ injection device at each preset injection time point into the gradient boosting tree model to diagnose the in-situ injection device in an unexpected state, and obtaining the name of the abnormal event currently occurring in the in-situ injection device in the unexpected state;

[0096] Determine the characteristic type of the name of the abnormal event currently occurring in the in-situ injection device in an unexpected state; wherein the characteristic type of the name of the abnormal event includes an uncontrollable abnormal operation event and a controllable abnormal operation event;

[0097] If the name of the abnormal event currently occurring in the in-situ injection device in an unexpected state is an uncontrollable abnormal operation event, the in-situ injection device is controlled to suspend the injection work, and the name of the abnormal event currently occurring in the in-situ injection device is sent to the preset terminal for display.

[0098] Among them, uncontrollable abnormal operation events refer to those abnormal conditions that occur during the operation of the in-situ injection device and cannot be corrected by operating the device itself or adjusting its parameters. For example, the key components of the device suddenly suffer irreparable mechanical damage, such as the steel pipe of the syringe ruptures and cannot be repaired on site, or due to external force majeure factors such as earthquakes and floods, the device is severely damaged, resulting in the device being unable to operate normally. These situations are beyond the scope of the device's own adjustment capabilities.

[0099] Adjustable abnormal operation events are abnormal conditions that occur when the in-situ injection device is running and can be restored to normal operation by adjusting the internal parameters, operation mode or operation process of the device. For example, if the injection pressure deviates from the standard value, it may be due to the improper opening of a valve. By adjusting the valve opening, the injection pressure can be restored to normal; or the injection speed is abnormal, which may be due to the unreasonable frequency setting of the pump. The injection speed problem can be solved by resetting the frequency of the pump.

[0100] It should be noted that when it is determined that the working state of the in-situ injection device is an unexpected state, the actual injection data of the device at each preset injection time point is first obtained. These actual injection data contain various information of the device during the actual operation, such as injection volume, injection pressure, etc. Then these actual injection data are imported into the previously constructed gradient boosting tree model. The gradient boosting tree model is constructed based on a large number of historical feature data of abnormal events in the operation of the in-situ injection device, and has the ability to diagnose abnormal situations. By importing the actual injection data into the model, the in-situ injection device in an unexpected state can be diagnosed, so as to obtain the name of the abnormal event currently occurring in the device. After obtaining the name of the abnormal event, it is necessary to determine its feature type. Here, the feature types of the abnormal event names are divided into uncontrollable abnormal operation events and adjustable abnormal operation events. This classification is to take different countermeasures according to different abnormal types. If it is determined that the abnormal event currently occurring in the in-situ injection device in an unexpected state is an uncontrollable abnormal operation event, since this type of abnormality cannot be resolved by the regulating device itself, the measure taken is to control the in-situ injection device to suspend the injection work to avoid possible more serious consequences, such as equipment damage or serious deterioration of the injection effect. At the same time, the name of the abnormal event currently occurring in the in-situ injection device is sent to the preset terminal for display, so that relevant personnel can be informed of the abnormal situation in time and carry out subsequent processing, such as repairing or replacing the equipment.

[0101] By importing the actual injection data into the gradient boosting tree model for diagnosis when the working state of the in-situ injection device is unexpected, the name of the current abnormal event of the device and its characteristic type can be accurately determined. For uncontrollable abnormal operation events, measures such as suspending the injection work and notifying relevant personnel are taken to help avoid further damage to the equipment and ensure the safety of the injection work. This method realizes intelligent diagnosis and targeted processing of abnormal conditions of the in-situ injection device, improving the reliability of the device operation and the controllability of the injection work.

[0102] Furthermore, if the working state of the in-situ injection device is an unexpected state, the actual injection data of the in-situ injection device at each preset injection time point is imported into the gradient boosting tree model for evaluation, and the in-situ injection device is adjusted and controlled according to the evaluation result, which also includes the following steps:

[0103] Obtaining control log information of the in-situ injection device, obtaining adjustment measures corresponding to various deviations of various injection data in the in-situ injection device according to the control log information, and obtaining the adjustment accuracy after adjusting the injection data with deviations through various adjustment measures;

[0104] Obtaining the adjustment measure corresponding to the maximum adjustment accuracy, taking the adjustment measure corresponding to the maximum adjustment accuracy as the best adjustment measure for the corresponding deviation of the corresponding injection data, and so on, obtaining the best adjustment measures corresponding to various deviations of the injection data in the in-situ injection device;

[0105] Constructing a knowledge matching library, and importing the best adjustment measures corresponding to various deviations of various injection data in the in-situ injection device into the knowledge matching library;

[0106] If the abnormal event currently occurring in the in-situ injection device in an unexpected state is named as an adjustable abnormal operation event, then obtaining various real-time injection data of the in-situ injection device at the current injection time point; and obtaining various standard injection data of the in-situ injection device at the current injection time point in the standard dynamic injection data pool;

[0107] Calculate the data deviation value between various real-time injection data of the in-situ injection device at the current injection time point and the corresponding standard injection data;

[0108] Marking the real-time injection data whose data deviation value is greater than the preset difference value as operation deviation injection data, and obtaining the data deviation value between the operation deviation injection data and the corresponding standard injection data;

[0109] The data deviation value between the operation deviation injection data and the corresponding standard injection data is imported into the knowledge matching library for matching, and the best adjustment measure is obtained by matching. The best adjustment measure obtained by matching is sent to the controller to correct the operation deviation injection data.

[0110] Specifically, first obtain the control log information of the in-situ injection device, and find out the corresponding adjustment measures when the injection data deviates from this information, as well as the adjustment accuracy after the implementation of these adjustment measures. By comparing the adjustment accuracy of different adjustment measures, find out the adjustment measure with the highest adjustment accuracy, and determine it as the best adjustment measure for the corresponding deviation of each injection data. For example, for injection pressure deviation, there may be multiple adjustment measures such as adjusting the power of the pump, changing the valve opening, etc. After analyzing the control log, determine which measure has the highest adjustment accuracy for the injection pressure deviation in the past, and use it as the best adjustment measure for the injection pressure deviation. Then build the best adjustment measures for each injection data into a knowledge matching library.

[0111] When the in-situ injection device is in an unexpected state and the current abnormal event is an adjustable abnormal operation event, it is necessary to obtain various real-time injection data and corresponding standard injection data at the current injection time point. Calculate the data deviation value between them, calibrate the real-time injection data with a deviation value greater than the preset difference as the operation deviation injection data, and obtain a more accurate data deviation value between it and the standard injection data again. Finally, import this data deviation value into the knowledge matching library for matching, find the corresponding optimal adjustment measure, and send the measure to the controller to correct the operation deviation injection data. For example, if it is found that the real-time data of the injection amount deviates too much from the standard data, find the best adjustment measure for the injection amount deviation (such as adjusting the pumping speed of the pumping pump, etc.) through the knowledge matching library, and then the controller corrects the injection amount according to this measure.

[0112] By analyzing the control log, the optimal adjustment measures for each injection data deviation are determined and a knowledge matching library is constructed. When an abnormal operation event of a controllable type occurs, the deviation between the real-time injection data and the standard injection data can be quickly calculated, and the optimal adjustment measures are matched in the knowledge matching library and sent to the controller for correction. This method realizes the rapid and accurate processing of controllable anomalies, improves the ability of the in-situ injection device to cope with abnormal situations, ensures the accuracy and stability of the injection work, and reduces the risk of poor injection results due to abnormal situations.

[0113] In practical application, the method may further include the following steps:

[0114] Obtaining operation log information of the in-situ injection device, obtaining historical blockage time of the in-situ injection device according to the operation log information, and obtaining characteristic working conditions of the in-situ injection device when various historical blockage times occur;

[0115] Conduct statistical analysis on the working characteristic conditions of the in-situ injection device when various historical silting times occurred, and obtain the silting risk coefficient of the in-situ injection device under various working characteristic conditions;

[0116] Constructing a knowledge base, dividing the knowledge base into a plurality of sub-base spaces, presetting encoding rules, and encoding various working characteristic conditions based on the encoding rules to encode the various working characteristic conditions into binary digits;

[0117] Compress and bundle the corresponding binary digits and the corresponding siltation risk coefficients to obtain a number of knowledge data packets, and map each knowledge data packet into the corresponding sub-library space;

[0118] Acquire the real-time working characteristic condition of the in-situ injection device, encode the real-time working characteristic condition according to the encoding rule, and obtain a string of binary digits corresponding to the real-time working characteristic condition;

[0119] Importing a string of binary digits corresponding to the real-time working characteristic conditions into the knowledge base to match the binary digits of each knowledge data packet to determine whether there are identical binary digits;

[0120] If there are identical binary digits, the silting risk coefficient of the knowledge data packet to which the identical binary digits belong is obtained, and the silting risk coefficient of the in-situ injection device operating under the real-time working characteristic conditions is obtained;

[0121] If the clogging risk coefficient of the in-situ injection device operating under the real-time working characteristic conditions is greater than the preset coefficient threshold, an early warning message is generated.

[0122] Specifically, firstly, the operation log information of the in-situ injection device is obtained to find out the historical silting time and the working characteristic conditions of the device when silting occurs each time. These working characteristic conditions include but are not limited to injection pressure, injection speed, viscosity of the material, ambient temperature, total operation time of the device and other factors. These working characteristic conditions are statistically analyzed to determine the silting risk coefficient of the in-situ injection device under different working characteristic conditions. This step is based on a large amount of historical data to mine the relationship between device silting and working conditions, providing a data basis for subsequent early warning. A knowledge base is constructed and divided into several sub-library spaces. In order to facilitate data storage and retrieval, encoding rules are preset to encode various working characteristic conditions into binary digits. Then the corresponding binary digits are compressed and bundled with the corresponding silting risk coefficients to form knowledge data packets, and these knowledge data packets are mapped to the corresponding sub-library spaces respectively. This processing method stores complex working characteristic conditions and silting risk coefficients in the knowledge base in a compact and orderly manner, which improves the efficiency of data management. The real-time working characteristic conditions of the in-situ injection device are obtained and encoded into binary digits according to the previous encoding rules. This binary digit is imported into the knowledge base and matched with the binary digits of each knowledge data packet. If the same binary digits exist, the siltation risk coefficient corresponding to the knowledge data packet is obtained. This coefficient is the siltation risk coefficient of the device under the current real-time working characteristic conditions. If this siltation risk coefficient is greater than the preset coefficient threshold, an early warning message is generated. This process realizes the rapid assessment and early warning of the siltation risk of the in-situ injection device under the real-time working state. By analyzing the historical siltation data of the in-situ injection device, a knowledge base containing the working characteristic condition coding and the siltation risk coefficient is constructed, and then the real-time working characteristic condition coding is matched with the knowledge base, the siltation risk coefficient of the device under the current working state can be quickly obtained and early warning information can be generated when the risk is high. The siltation risk of the in-situ injection device can be predicted in advance, which helps to take timely measures to prevent the occurrence of siltation, ensure the normal operation of the device, and improve the continuity and reliability of the in-situ injection of barrier materials.

[0123] In practical application, the method may further include the following steps:

[0124] Obtaining an operation log of the in-situ injection device, and obtaining electrical characteristic data corresponding to various operating states of the in-situ injection device according to the operation log; wherein the operating state includes a normal state, a state requiring regulation, and an unregulatable state (i.e., a fault state);

[0125] Introducing a factor analysis algorithm, based on which the correlation factor feature text between various electrical characteristic data under various operating conditions is analyzed;

[0126] Generate a feature text topology structure diagram according to the correlation factor feature text, and analyze the state transition probability between various operating states under various electrical feature data conditions according to the feature text topology structure diagram in combination with Markov chain analysis;

[0127] Constructing a state transition probability matrix according to the state transition probability between various operating states under various electrical characteristic data conditions;

[0128] Building a Markov model based on a deep learning network, embedding the state transition probability matrix into the Markov model for coding learning, and outputting the learned Markov model after the model parameters meet the preset requirements;

[0129] Acquire real-time electrical characteristic data of the in-situ injection device at a preset time node, import the real-time electrical characteristic data into the learned Markov model for deduction, and obtain the state transition probability value of the in-situ injection device in the future time;

[0130] If the state transition probability value of the in-situ injection device in the future time is greater than the preset probability value, an early warning message is generated.

[0131] It should be noted that the corresponding electrical characteristic data under different operating states (normal state, state requiring regulation, and unregulatory state, i.e., fault state) are first obtained from the operation log of the in-situ injection device. These electrical characteristic data include but are not limited to parameters such as voltage, current, and power factor, which can reflect the electrical characteristics of the device under different operating states. The factor analysis algorithm can find a few common factors hidden behind many variables (electrical characteristic data). These common factors can explain the correlation between variables, thereby generating a correlation factor characteristic text. This text describes the intrinsic connection between the electrical characteristic data.

[0132] A feature text topology diagram is constructed based on the correlation factor feature text. This diagram graphically displays the relationship between the electrical feature data. Then, the state transition probability between the operating states under different electrical feature data conditions is analyzed by combining the Markov chain. A state transition probability matrix is ​​constructed based on these state transition probabilities. This matrix is ​​a two-dimensional matrix, where rows and columns represent different operating states, and the elements in the matrix represent the probability of transitioning from one state to another.

[0133] A Markov model is constructed based on a deep learning network, and the state transition probability matrix is ​​embedded in it for encoding learning. The deep learning network can learn and optimize complex probability relationships by continuously adjusting the model parameters until the preset requirements are met. This process enables the Markov model to more accurately describe the operating state transition law of the in-situ injection device.

[0134] The real-time electrical characteristic data of the in-situ injection device is obtained at the preset time node and imported into the learned Markov model for deduction. The model calculates the state transition probability value of the in-situ injection device in the future time based on the real-time data and the previously learned state transition law. If this probability value is greater than the preset probability value, it means that the device is more likely to undergo state transition in the future (for example, from normal state to fault state), and an early warning message is generated at this time.

[0135] By acquiring the electrical characteristic data of the in-situ injection device under different operating conditions, the Markov model is constructed and trained using factor analysis algorithm, Markov chain analysis and deep learning network, and then the real-time electrical characteristic data is imported into the model for deduction. This method can predict the state transition probability of the in-situ injection device in the future in advance, and generate early warning information in time when the probability value exceeds the preset value, which helps to take measures in advance to prevent the device from malfunctioning or other adverse conditions, improve the reliability and stability of the device operation, and ensure the smooth progress of the in-situ injection of barrier materials.

[0136] Another aspect of the present invention discloses an in-situ injection device for a barrier material, which is applied to any of the in-situ injection methods for a barrier material, such as Figure 3 As shown, the in-situ injection device includes a material tank 2011 and a syringe, and the material tank and the syringe are connected through a feeding pipe 2012; wherein the material tank may be provided with a stirring and mixing mechanism, such as a spiral blade stirring mechanism, etc.;

[0137] The injector comprises a base 2013, and hydraulic cylinders 2014 are arranged on both sides of the base, and the output end of the hydraulic cylinder is connected with a pull rod 2015, and the top of the pull rod is connected with a mounting platform 2016, and a material pump 2017 is arranged on the mounting platform, and the input end of the material pump is connected with the feeding pipe, and the output end of the material pump is connected with a steel pipe 2018, and the bottom end of the steel pipe is connected with an injection head 2019;

[0138] The syringe is also provided with a pressure gauge 2020 and a controller 2021.

[0139] It should be noted that before the barrier material injection work begins, the material tank 2011 is used to store the barrier material. The material tank may be provided with a stirring and mixing mechanism (such as a spiral blade stirring mechanism) to stir and mix the barrier material in the material tank. This helps to ensure that the composition of the barrier material is uniform, especially for some barrier materials that may have precipitation or need to mix multiple components (such as alkali-activated gel, etc.), the stirring operation can ensure the consistency of the performance of the material during injection. The hydraulic cylinders 2014 on both sides of the base 2013 of the syringe are in the initial state, and the pull rod 2015 is in a suitable starting position under the action of the hydraulic cylinder. The material pump 2017 on the mounting platform 2016 is also in standby mode, the pressure gauge 2020 displays the initial pressure value (usually zero or a value balanced with the ambient pressure), and the controller 2021 is in a state ready to receive instructions. When the controller 2021 receives the instruction to start injection, the material pump 2017 starts. Since the input end of the pump is connected to the material tank 2011 through the feed pipe 2012, under the action of the pump, the barrier material is pumped from the material tank into the pump through the feed pipe. The barrier material pumped into the pump 2017 flows to the injection head 2019 through the steel pipe 2018 at its output end. In this process, the hydraulic cylinders 2014 on both sides of the base 2013 can act according to the instructions of the controller 2021. For example, if it is necessary to adjust the depth of the injection head inside the soil, the hydraulic cylinder can push the pull rod 2015, and the pull rod drives the mounting platform 216 and the pump 2017 thereon to move up and down as a whole, thereby adjusting the position of the injection head. When the injection head reaches the appropriate position, the barrier material is injected into the target area through the injection head. During the injection process, the pressure gauge 2020 continuously monitors the pressure situation, and the controller 2021 controls the entire injection process according to the pressure value and other preset injection parameters (such as injection speed, injection volume, etc.). If the injection speed needs to be changed, the controller can adjust the rotation speed of the pump; if the injection pressure needs to be changed, in addition to adjusting the power of the pump, it can also be achieved by adjusting the insertion depth of the injection head through the hydraulic cylinder.

[0140] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for in-situ injection of a barrier material, characterized in that: The following steps are involved: Obtaining a preset injection plan for the barrier material, importing the preset injection plan into a computer simulation system for simulated injection evaluation, and outputting a final injection plan for the in-situ injection device and a standard dynamic injection data pool during the working process after the simulation result satisfies the preset requirements; The final injection plan is transmitted to the controller of the in-situ injection device, and the in-situ injection device is controlled to inject the barrier material into the target area according to the final injection plan; During the injection process, the actual injection data of the in-situ injection device at each preset injection time point is obtained based on the Internet of Things; and obtaining standard injection data of the in-situ injection device at each preset injection time point in the standard dynamic injection data pool; Analyze the working state of the in-situ injection device according to the actual injection data and the standard injection data at each preset injection time point; if the working state of the in-situ injection device is the expected state, there is no need to correct the injection data of the in-situ injection device; Acquire historical feature data corresponding to various abnormal operation events of the in-situ injection device, and construct a gradient boosting tree model according to the historical feature data; If the working state of the in-situ injection device is unexpected, the actual injection data of the in-situ injection device at each preset injection time point is imported into the gradient boosting tree model for evaluation, and the in-situ injection device is adjusted and controlled according to the evaluation result; wherein, the barrier material includes alkali-activated gel and modified geomembrane barrier material.

2. The in-situ injection method of a barrier material according to claim 1, characterized in that: Obtain a preset injection plan for the barrier material, import the preset injection plan into a computer simulation system for simulated injection evaluation, and output the final injection plan of the in-situ injection device and its standard dynamic injection data pool during the working process until the simulation results meet the preset requirements, specifically: Obtain a three-dimensional structural model diagram of the in-situ injection device and a regional characteristic model diagram of the target area, and construct a computer simulation system for injecting barrier materials into the target area based on the three-dimensional structural model diagram of the in-situ injection device and the regional characteristic model diagram of the target area in combination with computer simulation software; wherein the target area is the area to be blocked; Obtaining a preset injection scheme for the barrier material, importing the preset injection scheme into the computer simulation system for injection simulation analysis, and outputting the simulation results after the simulation is completed; Obtaining in the simulation result a predicted permeability coefficient of the barrier material after the barrier material is simulated and injected into the target area through the preset injection scheme; comparing the predicted permeability coefficient with a preset threshold value; If the predicted permeability coefficient is greater than a preset threshold, there is no need to adjust and optimize the preset injection plan, and the preset injection plan is output as the final injection plan; If the predicted permeability coefficient is not greater than the preset threshold, a genetic algorithm is introduced to iteratively adjust and optimize the preset injection scheme. After each iterative optimization cycle, an optimized injection scheme is obtained, and the predicted permeability coefficient of the optimized injection scheme is evaluated by a simulation system until the predicted permeability coefficient of the optimized injection scheme is greater than the preset threshold. Then, the iteration is stopped and the optimized injection scheme is output as the final injection scheme. After the final injection plan is output, the standard injection data of the in-situ injection device based on the time series when the in-situ injection device is controlled by the final injection plan to perform simulation work is obtained in the computer simulation system, and the various standard injection data are aggregated to obtain a standard dynamic injection data pool of the in-situ injection device during the injection work.

3. The in-situ injection method of a barrier material according to claim 1, characterized in that: The working status of the in-situ injection device is analyzed based on the actual injection data and the standard injection data at each preset injection time point, specifically: Constructing a dynamic transformation curve diagram of actual injection data according to actual injection data of the in-situ injection device at each preset injection time point; and constructing a standard injection data dynamic transformation curve diagram according to the standard injection data of the in-situ injection device at each preset injection time point; Performing coincidence analysis on the actual injection data dynamic transformation curve graph and the standard injection data dynamic transformation curve graph to obtain the curve coincidence degree between the actual injection data dynamic transformation curve graph and the standard injection data dynamic transformation curve graph; Comparing the curve overlap between the actual injection data dynamic transformation curve diagram and the standard injection data dynamic transformation curve diagram with a preset overlap; If the curve overlap between the actual injection data dynamic transformation curve diagram and the standard injection data dynamic transformation curve diagram is greater than the preset overlap, it means that the actual working state of the in-situ injection device is consistent with the expected working state, and the injection effect is in line with expectations, then the working state of the in-situ injection device is calibrated as the expected state; If the curve overlap between the actual injection data dynamic transformation curve and the standard injection data dynamic transformation curve is not greater than the preset overlap, it means that the actual working state of the in-situ injection device deviates from the expected working state, and the injection effect does not meet expectations, then the working state of the in-situ injection device is calibrated as an unexpected state.

4. The in-situ injection method of a barrier material according to claim 1, characterized in that: Obtain the historical feature data corresponding to various abnormal operation events of the in-situ injection device, and build a gradient boosting tree model based on the historical feature data, specifically: Obtain historical characteristic data corresponding to various abnormal operation events of the in-situ injection device through the big data network; The collected historical characteristic data of abnormal events in the operation of the in-situ injection device are sorted into a structured table form, where each record corresponds to an abnormal event name and a characteristic value of the corresponding historical characteristic data; The eigenvalues ​​in the table are extracted to form a feature matrix X, and the abnormal results are formed into a label vector y; where X contains the eigenvalues ​​used to train the model, and y contains the corresponding abnormal labels; Introduce a gradient boosting tree algorithm, initialize a gradient boosting tree model, and set the number of iterations; use the gradient boosting framework and combine the feature matrix X to iteratively construct a decision tree, and correct the residual of the previous decision tree according to the corresponding label vector y to form a strong learner; At the same time, in each iteration, the gradient and second-order derivative of the loss function are calculated and used as the input of the new weak learner to update the model to minimize the loss function; After reaching the number of iterations, a gradient boosting tree model consisting of multiple decision trees is constructed.

5. The in-situ injection method of a barrier material according to claim 1, characterized in that: If the working state of the in-situ injection device is unexpected, the actual injection data of the in-situ injection device at each preset injection time point is imported into the gradient boosting tree model for evaluation, and the in-situ injection device is adjusted and controlled according to the evaluation results, specifically: If the working state of the in-situ injection device is an unexpected state, obtaining actual injection data of the in-situ injection device at each preset injection time point; Importing the actual injection data of the in-situ injection device at each preset injection time point into the gradient boosting tree model to diagnose the in-situ injection device in an unexpected state, and obtaining the name of the abnormal event currently occurring in the in-situ injection device in the unexpected state; Determine the characteristic type of the name of the abnormal event currently occurring in the in-situ injection device in an unexpected state; wherein the characteristic type of the name of the abnormal event includes an uncontrollable abnormal operation event and a controllable abnormal operation event; If the name of the abnormal event currently occurring in the in-situ injection device in an unexpected state is an uncontrollable abnormal operation event, the in-situ injection device is controlled to suspend the injection work, and the name of the abnormal event currently occurring in the in-situ injection device is sent to the preset terminal for display.

6. The in-situ injection method of a barrier material according to claim 5, characterized in that: If the working state of the in-situ injection device is an unexpected state, the actual injection data of the in-situ injection device at each preset injection time point is imported into the gradient boosting tree model for evaluation, and the in-situ injection device is adjusted and controlled according to the evaluation result, which also includes the following steps: Obtaining control log information of the in-situ injection device, obtaining adjustment measures corresponding to various deviations of various injection data in the in-situ injection device according to the control log information, and obtaining the adjustment accuracy after adjusting the injection data with deviations through various adjustment measures; Obtaining the adjustment measure corresponding to the maximum adjustment accuracy, taking the adjustment measure corresponding to the maximum adjustment accuracy as the best adjustment measure for the corresponding deviation of the corresponding injection data, and so on, obtaining the best adjustment measures corresponding to various deviations of the injection data in the in-situ injection device; Constructing a knowledge matching library, and importing the best adjustment measures corresponding to various deviations of various injection data in the in-situ injection device into the knowledge matching library; If the abnormal event currently occurring in the in-situ injection device in an unexpected state is named as an adjustable abnormal operation event, then obtaining various real-time injection data of the in-situ injection device at the current injection time point; and obtaining various standard injection data of the in-situ injection device at the current injection time point in the standard dynamic injection data pool; Calculate the data deviation value between various real-time injection data of the in-situ injection device at the current injection time point and the corresponding standard injection data; Marking the real-time injection data whose data deviation value is greater than the preset difference value as operation deviation injection data, and obtaining the data deviation value between the operation deviation injection data and the corresponding standard injection data; The data deviation value between the operation deviation injection data and the corresponding standard injection data is imported into the knowledge matching library for matching, and the best adjustment measure is obtained by matching. The best adjustment measure obtained by matching is sent to the controller to correct the operation deviation injection data.

7. An in-situ injection device for barrier materials, applied to an in-situ injection method for barrier materials according to any one of claims 1 to 6, characterized in that: The in-situ injection device comprises a material tank and a syringe, and the material tank and the syringe are connected through a feeding pipe; The injector comprises a base, and hydraulic cylinders are arranged on both sides of the base, the output end of the hydraulic cylinder is connected with a pull rod, the top end of the pull rod is connected with a mounting platform, a material pump is arranged on the mounting platform, the input end of the material pump is connected with the feeding pipe, the output end of the material pump is connected with a steel pipe, and the bottom end of the steel pipe is connected with an injection head; The syringe is also provided with a pressure gauge and a controller.

Images