A method, device and system for intelligent construction control of splitting shotcrete mixing piles

CN117926802BActive Publication Date: 2024-11-19SHANDONG UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410033776.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-01-08
Publication Date
2024-11-19
Estimated Expiration
2044-01-08

Smart Images

  • Figure CN117926802B_ABST
    Figure CN117926802B_ABST
Patent Text Reader

Abstract

The present invention proposes a method, device and system for intelligent construction control of splitting shotcrete mixing piles, which relates to the field of geotechnical engineering technology. It includes receiving real-time perception information output from a construction information self-perception system, wherein the real-time perception information includes drilling depth, drilling rig output power, drill bit torque, drill rod axial force and drill bit pore water pressure data; inputting the real-time perception information into a trained formation information prediction machine learning model, establishing a nonlinear implicit correspondence between the real-time perception information and the construction formation conditions, and outputting real-time formation state information; inputting the real-time formation state information into a trained optimal construction parameter self-matching machine learning model, establishing a nonlinear implicit correspondence between the real-time formation state information and the optimal construction parameters, and outputting the current optimal construction parameters. The present invention can perform real-time intelligent inversion of formation information and further dynamically self-optimize the construction parameters.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of geotechnical engineering, and in particular relates to an intelligent construction control method, device and system for splitting shotcrete mixing piles. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Cement mixing piles usually have the advantages of simple construction technology, wide application of soil types, and large reinforcement depth. They can be used to treat saturated soft clay of various origins, including silt, silty soil, silt, sandy soil, peat soil, and clay soil with high water content and a standard value of foundation bearing capacity not exceeding 120kPa. However, a large number of practices have proved that when the clay content in clay soil is high (plasticity index is large), cement slurry is squeezed out from the slurry outlet of the drill bit at a low pressure and flow rate, and it is difficult to mix into the soft and sticky soil layer. It is easy to cause mud and cement slurry blocks in the pile body, serious cement slurry loss, and extremely uneven mixing, which seriously affects the quality of pile formation and reinforcement treatment effect.

[0004] In order to solve the above problems, splitting mixing piles are often used in actual projects. Splitting cement soil mixing piles have made two major improvements to traditional cement soil mixing piles: one is to change the large diameter slurry outlet to two symmetrical small diameter nozzles; the other is to change the low-pressure mud pump to a high-pressure mud pump. When the cement slurry is ejected from the nozzle at high pressure, it produces a cutting and splitting effect on the clay layer (similar to the high-pressure rotary spraying of the single-tube method), forcing the cement slurry to be mixed into the soft clay. In addition, the action of the stirring rod can make the cement slurry more evenly mixed into the soft clay, solving the common quality problems of traditional cement soil mixing piles. However, splitting cement soil mixing piles still have shortcomings and limitations in current engineering applications.

[0005] First, the strata where the splitting mixing piles are located are complex and changeable, and real-time information about the strata cannot be obtained during the construction process. Secondly, the splitting cement soil mixing piles are greatly affected by the construction environment and are highly dependent on the workers' experience. The parameters cannot be optimized and controlled according to the actual working conditions, which affects the construction quality and efficiency. In addition, the current level of intelligence of splitting mixing piles is insufficient, resulting in labor waste, difficulty in ensuring construction quality, and low efficiency.

[0006] Therefore, the current split shotcrete cement mixing piles cannot meet the actual engineering needs. There is an urgent need to solve the above problems by integrating a real-time intelligent inversion method of formation information and a dynamic self-optimization control technology of construction parameters, and developing intelligent construction equipment for split shotcrete pile foundations. Summary of the invention

[0007] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a splitting shotcrete mixing pile intelligent construction control method, device and system, which can realize real-time intelligent inversion of stratum information and further dynamic self-optimization control of construction parameters, and construct a set of splitting mixing pile foundation intelligent construction equipment to solve the problems in the prior art.

[0008] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0009] A first aspect of the present invention provides an intelligent construction control method for splitting shotcrete mixing piles.

[0010] A method for intelligent construction control of splitting shotcrete mixing piles, comprising the following steps:

[0011] Receiving real-time sensing information output from a construction information self-sensing system, the real-time sensing information including drilling depth, drilling rig output power, drill bit torque, drill rod axial force and drill bit pore water pressure data;

[0012] The real-time sensing information is input into a trained formation information prediction machine learning model, a nonlinear implicit correspondence between the real-time sensing information and the construction formation conditions is established, and real-time formation status information is output;

[0013] The real-time formation state information is input into the trained optimal construction parameter self-matching machine learning model, a nonlinear implicit correspondence between the real-time formation state information and the optimal construction parameters is established, and the current optimal construction parameters are output.

[0014] Optionally, the drilling depth and the drilling rig output power are obtained through the drilling rig, and the drill bit torque, drill rod axial force and drill bit pore water pressure data are obtained through a torque sensor, an axial force sensor and a pore water pressure sensor.

[0015] Optionally, train the formation information prediction machine learning model, specifically:

[0016] Build a random forest-DBSCAN integrated model;

[0017] Obtain existing geological status information and experimental data, and classify and annotate the stratum status information corresponding to the drilling depth;

[0018] The experimental data and the annotated formation state information are input into the random forest-DBSCAN integrated model, the DBSCAN algorithm is used to cluster the experimental data and train the random forest model;

[0019] Use the validation set or cross-validation method to evaluate and tune the model to obtain a trained stratigraphic information prediction machine learning model, that is, a trained random forest-DBSCAN integrated model;

[0020] Real-time perception information is input into the trained random forest-DBSCAN integrated model to obtain an estimate of the real-time formation status information.

[0021] Optionally, the optimal construction parameter self-matching machine learning model is represented as a mapping relationship between multiple inputs and multiple outputs:

[0022] f:G=f(s)

[0023] Among them, s is the input vector - formation conditions, s = (α1, α2, ... α n ),α1-α n represents the real-time formation status information component, G is the output vector - the optimal construction parameter combination, G = (β1, β2, ... β n ),β1-β n Represents the output performance component.

[0024] Optionally, the optimal construction parameter self-matching machine learning model is trained, specifically:

[0025] Obtain the optimal construction parameter database and pre-process the data in the database;

[0026] The annotated formation state information is used as the target variable, and the chi-square test method is used to evaluate the correlation between the feature and the target variable. The top M features with the highest correlation are selected and the features are standardized.

[0027] The optimal construction parameter database is divided into n datasets of similar size using the K-fold cross-validation method;

[0028] Select the custom polynomial kernel function of the SVM model and build a self-matching machine learning model for optimal construction parameters;

[0029] Perform K rounds of training set performance evaluation on the built construction parameter self-matching model. In each round, n-1 data sets are used for model training, and the remaining data set is used as a test set. Finally, the average value of the K rounds of evaluation results is calculated as the performance indicator of the model. If the requirements are not met, the next round of training is carried out until the training is completed.

[0030] Real-time formation status information is input into the trained optimal construction parameter self-matching machine learning model, and the predicted optimal construction parameters are obtained through forward propagation.

[0031] Optionally, the optimal construction parameter database is based on split pile construction parameters under typical formation conditions obtained by the cloud platform, including water-cement ratio, mud dosage, shotcrete pressure, grouting speed, drilling speed, drill rod lifting speed and fan blade rotation speed.

[0032] Optionally, also include:

[0033] The matched current optimal construction parameters are input into the construction parameter intelligent adjustment system to dynamically adjust the construction parameters to achieve intelligent control of the construction process.

[0034] A second aspect of the present invention provides an intelligent construction device for splitting shotcrete mixing piles.

[0035] An intelligent construction device for splitting shotcrete mixing piles, comprising a construction information self-sensing system, a control device and a construction parameter intelligent adjustment system, wherein:

[0036] The construction information self-sensing system includes a drilling rig and multi-source sensor components, wherein the drilling rig is used to detect the drilling depth and the drilling rig output power, and the multi-source sensor components are used to obtain the drill bit torque, the drill rod axial force and the drill bit pore water pressure data;

[0037] The control device is used to predict, output and display real-time stratum status information based on the detection value of the construction information self-sensing system, and output the current optimal construction parameters based on the real-time stratum status information;

[0038] The construction parameter intelligent adjustment system is used to perform real-time control of the variable frequency speed variable voltage adjustment equipment based on the current optimal construction parameters.

[0039] Optionally, the multi-source sensing components include a torque sensor installed at the drill bit, a pore water pressure sensor, and an axial force sensor installed at the bottom of the drill rod; the variable frequency speed and variable pressure regulating equipment includes a drilling drive motor and a high-pressure variable frequency pump, wherein the drilling drive motor is used to control the drilling speed, the lifting speed of the drill rod and the rotation speed of the fan blades, and the high-pressure variable frequency pump is used to control the spraying pressure and the grouting speed.

[0040] A third aspect of the present invention provides an intelligent construction control system for splitting shotcrete mixing piles.

[0041] An intelligent construction control system for splitting shotcrete mixing piles, comprising:

[0042] The real-time sensing information acquisition module is configured to: receive real-time sensing information output from the construction information self-sensing system, wherein the real-time sensing information includes drilling depth, drilling rig output power, drill bit torque, drill rod axial force and drill bit pore water pressure data;

[0043] The real-time formation state information prediction module is configured to: input the real-time perception information into a trained formation information prediction machine learning model, establish a nonlinear implicit correspondence between the real-time perception information and the construction formation conditions, and output the real-time formation state information;

[0044] The current optimal construction parameter acquisition module is configured to: input the real-time formation state information into the trained optimal construction parameter self-matching machine learning model, establish a nonlinear implicit correspondence between the real-time formation state information and the optimal construction parameters, and output the current optimal construction parameters.

[0045] One or more of the above technical solutions have the following beneficial effects:

[0046] The present invention provides a method, device and system for intelligent construction control of splitting shotcrete mixing piles, which can sense various parameters selected in the whole construction process in real time, synchronously invert accurate soil layer state parameters, and then control and optimize various construction parameters in the pile forming process in real time according to the inverted soil layer state parameters, thereby improving the effect of splitting cement soil mixing piles in treating and reinforcing soft foundations.

[0047] During the training process of the machine learning model for formation information prediction, the present invention integrates the DBSCAN algorithm on the basis of the general random forest model, uses the DBSCAN algorithm to cluster the real-time perception information data, and then uses the clustering results as data features, thereby reducing the amount of irrelevant features, thereby reducing the complexity of the model and improving the calculation accuracy, thereby overcoming the shortcomings of the traditional random forest model in processing high-dimensional sparse data poorly.

[0048] The optimal construction parameter self-matching machine learning model of the present invention is a SVM model. The kernel function of the model is a customized polynomial kernel function, which greatly improves the shortcomings of the general kernel function, such as complex operation and slow calculation speed, and can meet the requirements of real-time inversion of engineering.

[0049] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0051] Figure 1 This is a flow chart of the method of the first embodiment.

[0052] Figure 2 This is a schematic diagram of the machine learning structure of the first embodiment.

[0053] Figure 3 It is the overall structural diagram of the second embodiment.

[0054] Figure 4 It is a schematic diagram of the construction process of the present invention.

[0055] In the accompanying drawings, the components represented by the reference numerals are listed as follows:

[0056] Multi-source sensing probe 1; multi-source sensing components 2; beam 3; variable speed drive 4; wireless data transceiver 5; hub 6; single chip computer 7; display 8; workstation 9; pulping station 10; variable frequency spray pump 11. DETAILED DESCRIPTION

[0057] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0058] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0059] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0060] Embodiment 1

[0061] This embodiment discloses an intelligent construction control method for splitting shotcrete mixing piles.

[0062] like Figure 1 As shown, a method for intelligent construction control of splitting shotcrete mixing piles comprises the following steps:

[0063] Receiving real-time sensing information output from a construction information self-sensing system, the real-time sensing information including drilling depth, drilling rig output power, drill bit torque, drill rod axial force and drill bit pore water pressure data;

[0064] The real-time sensing information is input into a trained formation information prediction machine learning model, a nonlinear implicit correspondence between the real-time sensing information and the construction formation conditions is established, and real-time formation status information is output;

[0065] Input the real-time stratum state information into the trained optimal construction parameter self-matching machine learning model, establish a nonlinear implicit correspondence between the real-time stratum state information and the optimal construction parameters, and output the current optimal construction parameters;

[0066] The matched current optimal construction parameters are input into the construction parameter intelligent adjustment system to dynamically adjust the construction parameters to achieve intelligent control of the construction process.

[0067] In order to achieve real-time and accurate inversion of formation state information, in this embodiment, the DBSCAN algorithm is integrated on the basis of the general random forest model, thereby overcoming the shortcomings of the traditional random forest model in processing high-dimensional sparse data, specifically:

[0068] 1) The DBSCAN algorithm is used to cluster the real-time perception information data, and then the clustering results are used as data features, which reduces the amount of irrelevant features, thereby reducing the model complexity and improving the calculation accuracy;

[0069] 2) Classify and label the formation state (formation type, lithology or aquifer) corresponding to the drilling depth based on the existing geological information and experimental data;

[0070] 3) Using the annotated stratum state information as the target variable, the features and stratum state information are input into the random forest-DBSCAN integrated model for training;

[0071] 4) Use the validation set or cross-validation method to evaluate and tune the model. You can try different feature combinations, model parameters, and algorithms to improve the accuracy and generalization ability of the model, and obtain a trained stratigraphic information prediction machine learning model, that is, a trained random forest-DBSCAN integrated model;

[0072] 5) By inputting real-time perception information into the trained random forest-DBSCAN integrated model, an estimate of the real-time formation status information can be obtained.

[0073] The real-time stratum state information is input into the appropriate optimal construction parameter self-matching machine learning model, a nonlinear implicit correspondence between the real-time stratum state information and the optimal construction parameters is established, and the current optimal construction parameters are output. The specific steps are as follows:

[0074] 1) Data preprocessing: Remove data outliers based on statistical analysis methods, and use regression interpolation methods to fill in data for locally sparse data;

[0075] 2) Feature selection and standardization: The chi-square test method is used to evaluate the correlation between the features and the target variable. The five features with the highest correlation are selected, and the Z-score standardization method is used to convert the selected five features into a normal distribution with a mean of 0 and a variance of 1, so that they are comparable;

[0076] 3) Data partitioning: In order to better evaluate the generalization ability and performance of the model, the K-fold cross-validation method is used to divide the database into n datasets of similar size;

[0077] 4) Custom kernel function: Because this model involves a nonlinear parameter optimization problem based on high-dimensional data and requires extremely high efficiency of model operation, in order to ensure calculation accuracy and improve model calculation speed, a custom polynomial kernel function is used to map the data to a high-dimensional feature space, thereby converting the nonlinear problem into a linear problem;

[0078] 5) Model training and evaluation: Perform K rounds of training set performance evaluation on the built construction parameter self-matching model. In each round, n-1 data sets are used for model training, and the remaining data set is used as a test set. Finally, the average value of the K rounds of evaluation results is calculated as the performance indicator of the model. If it does not meet the requirements, the next round of training is carried out;

[0079] 6) Prediction of optimal construction parameters: Input real-time formation status information into the trained optimal construction parameter self-matching machine learning model, and obtain the predicted optimal construction parameters through forward propagation.

[0080] In this embodiment, the optimal construction parameter self-matching machine learning model is a specially designed SVM model. The kernel function of the model is a customized polynomial kernel function, which greatly improves the shortcomings of the general kernel function, such as complex operation and slow calculation speed, and can meet the requirements of real-time inversion of engineering.

[0081] The above control method is specifically as follows:

[0082] After receiving the real-time perception information output from the construction information self-perception system, various types of data transmitted by the construction information self-perception system are processed and analyzed in real time, and the soil conditions during the construction process are simultaneously inverted. This process is completed through the pre-trained formation information prediction machine learning model.

[0083] The machine learning model is trained by using the actual engineering geological database to improve the calculation accuracy. Various parameter data collected by the construction self-sensing parameter system are used as the input layer of the model to establish a nonlinear implicit correspondence between real-time perception information and construction formation conditions. The real-time formation conditions are used as the output layer of the model to invert the formation information in real time.

[0084] After obtaining the real-time inversion formation information, it is necessary to synchronously match the optimal construction parameters during the construction process and dynamically adjust the actual construction parameters of the pile driver through intelligent control equipment to achieve the optimal value. This process is completed through the pre-trained optimal construction parameter self-matching machine learning model.

[0085] The optimal construction parameter self-matching machine learning model is trained through the optimal construction parameter database to obtain the trained optimal construction parameter self-matching machine learning model. Among them, the optimal construction parameter database is a self-updating and continuously optimized optimal construction parameter database built using big data technology, and the machine deep learning algorithm is used to represent the construction parameter matching model as an implicit multi-point mapping relationship, so as to determine the optimal construction parameter combination under the current construction stratum conditions, and the intelligent control system is used to control and adjust the current construction parameters to the optimal value in real time.

[0086] More specifically, the optimal construction parameter database obtains the split pile construction parameters under typical geological conditions based on the cloud platform, including water-cement ratio, mud consumption, shotcrete pressure and grouting speed, as well as drilling speed, drill rod lifting speed and fan blade rotation speed; the database can be continuously updated and expanded based on the cloud platform.

[0087] The optimal construction parameter self-matching system adopts artificial machine learning to build an algorithm model, and represents the construction parameter matching model as a mapping relationship between multiple inputs and multiple outputs, which is specifically expressed as follows:

[0088] f:G=f(s)

[0089] Where s is the input vector - formation conditions, s = (α1, α2, ... α n ),α1-α n represents the formation information component, G is the output vector - the optimal construction parameter combination, G = (β1, β2, ... β n ),β1-β n Represents the output performance component.

[0090] After obtaining the output of the self-matching machine learning model of the optimal construction parameters, the optimal construction parameters are received by the construction parameter intelligent adjustment system, and the construction parameters are controlled in real time through the variable frequency speed and voltage regulating equipment to achieve intelligent control of the construction process and ensure that the construction is carried out according to the optimal construction parameters.

[0091] Embodiment 2

[0092] This embodiment discloses an intelligent construction device for splitting shotcrete mixing piles.

[0093] like Figure 3 As shown, an intelligent construction device for splitting shotcrete mixing piles includes a construction information self-sensing system, a control device and a construction parameter intelligent adjustment system, wherein:

[0094] The construction information self-sensing system includes a drilling rig and multi-source sensor components, wherein the drilling rig is used to detect the drilling depth and the drilling rig output power, and the multi-source sensor components are used to obtain the drill bit torque, the drill rod axial force and the drill bit pore water pressure data;

[0095] The control device is used to predict, output and display real-time stratum status information based on the detection value of the construction information self-sensing system, and output the current optimal construction parameters based on the real-time stratum status information;

[0096] The construction parameter intelligent adjustment system is used to perform real-time control of the variable frequency speed variable voltage adjustment equipment based on the current optimal construction parameters.

[0097] Specifically, the intelligent control system for splitting cement soil mixing pile construction provided in this embodiment includes: a construction information self-sensing system, a stratum information real-time inversion system, and a construction parameter dynamic self-optimization system:

[0098] The function of the construction information self-sensing system is to perceive in real time various soil state parameters and construction parameter data during the pile foundation construction process, including drilling rigs, multi-source sensor components, and wireless data transmission platforms; during construction, the construction information system synchronously collects various data and sends them to the formation information real-time inversion system through the wireless data transmission platform for the next step of data processing and analysis.

[0099] The drilling rig is a high-pressure mixing drilling rig, which can monitor the drilling depth and the output power of the drilling rig in real time. The multi-source sensing element is a vibrating string sensor, including a torque sensor and a pore water pressure sensor installed at the drill bit and an axial force sensor installed at the bottom of the drill rod.

[0100] The drilling rig and multi-source sensor components are used to sense the drilling depth, torque, axial force, pore water pressure during the drilling process, and the output power of the drilling rig in real time. The drilling depth, torque, axial force, pore water pressure and output power are used as input parameters for formation inversion.

[0101] The wireless data transmission platform includes a wireless transceiver chip, a low-power single-chip microcomputer, a hub and a receiving terminal module, which are used to receive sensor data and transmit it to the terminal system;

[0102] The single chip microcomputer adopts an embedded ultra-low power single chip microcomputer, which has the advantages of fast computing speed, strong stability and low power consumption.

[0103] The hub uses a stacked hub to regenerate and expand the received signal to extend the transmission interval of the network and quickly transmit the port data to the terminal system.

[0104] The terminal computer is used to receive sensor signals, display sensor data, and input the data into the real-time inversion system of formation information.

[0105] The intelligent control system includes variable frequency speed variable voltage regulating equipment and a construction parameter intelligent regulating system.

[0106] The variable frequency speed and pressure regulating equipment includes a drilling drive motor and a high-pressure variable frequency pump, wherein the drilling drive motor is used to control the drilling speed, the lifting speed of the drill rod and the rotation speed of the fan blades, and the high-pressure variable frequency pump is used to control the spraying pressure and the grouting speed.

[0107] like Figure 3As shown, the drill rod is connected to the frame beam, the drill rod is provided with multi-source sensor components, the drill bit is provided with a multi-source sensor probe, the variable frequency spray pump is connected to the pulping station through a pipeline, the drill rod is connected to the pulping station through a pipeline, the variable frequency driver drives the drilling rig and the fan blades, the multi-source sensor components and the multi-source sensor probes are connected to the hub through a wireless data transceiver, the hub is connected to the single-chip microcomputer, and the single-chip microcomputer is connected to the display.

[0108] The single chip microcomputer is used to store the trained formation information prediction machine learning model, establish a nonlinear implicit correspondence between real-time perception information and construction formation conditions, and output real-time formation status information.

[0109] It also includes a workstation computer, which is communicatively connected to the single-chip microcomputer. In this embodiment, the workstation computer is used to store the self-matching machine learning model of optimal construction parameters, receive predicted real-time formation status information, and obtain the current optimal construction parameters based on the self-matching machine learning model of optimal construction parameters.

[0110] like Figure 4 As shown, the construction process of the intelligent construction device for splitting shotcrete mixing piles according to this embodiment includes the following steps:

[0111] 1. Install multi-source sensor components on the drilling rig, complete sensor calibration, and build a mobile pulping station;

[0112] 2. Move the drilling rig to the designated location, align the drill bit with the center of the hole, and level the drilling rig to ensure it is stable and level;

[0113] 3. Drill holes at designated locations and use multi-source sensing components to monitor the drilling depth, drilling speed, soil pressure, pore water pressure and other sensing parameters in real time during the drilling process;

[0114] 4. The wireless sensor platform receives the information data collected by multi-source sensors in real time and uploads it to the terminal computer;

[0115] 5. Input the received data into the formation inversion model to invert the formation state information combination (soil type, strength, water content and permeability coefficient, etc.) in real time;

[0116] 6 Input the outputted formation state information combination into the optimal construction parameter self-matching system in the workstation computer, and match the best construction parameter combination based on the optimal construction parameter database;

[0117] 7. Input the best matching construction parameter combination into the construction parameter intelligent adjustment system in the workstation, dynamically adjust the construction parameters through variable frequency speed and pressure regulating equipment (drilling drive rig and high-pressure variable frequency pump), and realize intelligent control of the construction process;

[0118] 8. After drilling to the designed depth, turn off the mud pump, stir and lift the mud, and invert the formation state parameters synchronously, so as to intelligently adjust the speed of acceleration and stirring rod speed;

[0119] 9. When the slurry needs to be sprayed and mixed again, start the mud pump, lift the drill bit to more than 0.5m above the designed cement soil pile elevation, and stop slurrying;

[0120] 10. Move the drilling rig to the next construction location and repeat the above steps to complete the construction of the jet grouting piles.

[0121] Embodiment 3

[0122] This embodiment discloses an intelligent construction control system for splitting shotcrete mixing piles.

[0123] An intelligent construction control system for splitting shotcrete mixing piles, comprising:

[0124] The real-time sensing information acquisition module is configured to: receive real-time sensing information output from the construction information self-sensing system, wherein the real-time sensing information includes drilling depth, drilling rig output power, drill bit torque, drill rod axial force and drill bit pore water pressure data;

[0125] The real-time formation state information prediction module is configured to: input the real-time perception information into a trained formation information prediction machine learning model, establish a nonlinear implicit correspondence between the real-time perception information and the construction formation conditions, and output the real-time formation state information;

[0126] The current optimal construction parameter acquisition module is configured to: input the real-time formation state information into the trained optimal construction parameter self-matching machine learning model, establish a nonlinear implicit correspondence between the real-time formation state information and the optimal construction parameters, and output the current optimal construction parameters.

[0127] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0128] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A method for intelligent construction control of splitting shotcrete mixing piles, characterized in that: The following steps are involved: Receiving real-time sensing information output from a construction information self-sensing system, the real-time sensing information including drilling depth, drilling rig output power, drill bit torque, drill rod axial force and drill bit pore water pressure data; The real-time sensing information is input into a trained stratum information prediction machine learning model, a nonlinear implicit correspondence between the real-time sensing information and the construction stratum conditions is established, and real-time stratum status information is output; the machine learning model is a trained random forest-DBSCAN integrated model; The real-time stratum state information is input into the trained optimal construction parameter self-matching machine learning model, a nonlinear implicit correspondence between the real-time stratum state information and the optimal construction parameters is established, and the current optimal construction parameters are output. The specific steps are as follows: 1) Data preprocessing: Remove data outliers based on statistical analysis methods. For locally sparse data, use regression interpolation methods to fill in the data. 2) Feature selection and standardization: The chi-square test method is used to evaluate the correlation between the features and the target variable. The five features with the highest correlation are selected. The Z-score standardization method is used to convert the selected five features into a normal distribution with a mean of 0 and a variance of 1 to make them comparable. 3) Data partitioning: Use the K-fold cross-validation method to divide the database into n datasets of similar size; 4) Custom kernel function: Use custom polynomial kernel function to map data into high-dimensional feature space, thereby converting nonlinear problems into linear problems; 5) Model training and evaluation: Perform K rounds of training set performance evaluation on the built construction parameter self-matching model. In each round, n-1 data sets are used for model training, and the remaining data set is used as a test set. Finally, the average value of the K rounds of evaluation results is calculated as the performance indicator of the model. If it does not meet the requirements, the next round of training is carried out; 6) Prediction of optimal construction parameters: Input real-time formation state information into the trained optimal construction parameter self-matching machine learning model, and obtain the predicted optimal construction parameters through forward propagation; The optimal construction parameter self-matching machine learning model is a specially designed SVM model, and the kernel function of the model is a custom polynomial kernel function.

2. The intelligent construction control method for splitting shotcrete mixing piles according to claim 1, characterized in that: The drilling depth and the drilling rig output power are obtained through the drilling rig, and the drill bit torque, drill rod axial force and drill bit pore water pressure data are obtained through a torque sensor, an axial force sensor and a pore water pressure sensor.

3. The intelligent construction control method for splitting shotcrete mixing piles according to claim 1, characterized in that: The machine learning model for formation information prediction is trained as follows: Build a random forest-DBSCAN integrated model; Obtain existing geological status information and experimental data, and classify and label the stratum status information corresponding to the drilling depth; The experimental data and the annotated formation state information are input into the random forest-DBSCAN integrated model, the DBSCAN algorithm is used to cluster the experimental data and train the random forest model; Use the validation set or cross-validation method to evaluate and tune the model to obtain a trained stratigraphic information prediction machine learning model, that is, a trained random forest-DBSCAN integrated model; Real-time perception information is input into the trained random forest-DBSCAN integrated model to obtain an estimate of the real-time formation status information.

4. The intelligent construction control method for splitting shotcrete mixing piles according to claim 1, characterized in that: The optimal construction parameter self-matching machine learning model is expressed as a mapping relationship between multiple inputs and multiple outputs: in, s is the input vector - formation conditions, , Represents the real-time formation status information component, G is the output vector - the optimal construction parameter combination, , Represents the output performance component.

5. The intelligent construction control method for splitting shotcrete mixing piles according to claim 1, characterized in that: The optimal construction parameter database is based on the split pile construction parameters under typical formation conditions obtained on the cloud platform, including water-cement ratio, mud dosage, shotcrete pressure, grouting speed, drilling speed, drill rod lifting speed and fan blade rotation speed.

6. The intelligent construction control method for splitting shotcrete mixing piles according to claim 1, characterized in that: Also includes: The matched current optimal construction parameters are input into the construction parameter intelligent adjustment system to dynamically adjust the construction parameters to achieve intelligent control of the construction process.

7. An intelligent construction device for splitting shotcrete mixing piles, characterized in that: The construction device is used to implement the steps in the intelligent construction control method of the splitting shotcrete mixing pile according to any one of claims 1 to 6, including a construction information self-sensing system, a control device and a construction parameter intelligent adjustment system, wherein: The construction information self-sensing system includes a drilling rig and multi-source sensor components, wherein the drilling rig is used to detect the drilling depth and the drilling rig output power, and the multi-source sensor components are used to obtain the drill bit torque, the drill rod axial force and the drill bit pore water pressure data; The control device is used to predict, output and display real-time stratum status information based on the detection value of the construction information self-sensing system, and output the current optimal construction parameters based on the real-time stratum status information; The construction parameter intelligent adjustment system is used to perform real-time control of the variable frequency speed variable voltage adjustment equipment based on the current optimal construction parameters.

8. The intelligent construction device for splitting shotcrete mixing piles according to claim 7, characterized in that: The multi-source sensing components include a torque sensor installed at the drill bit, a pore water pressure sensor and an axial force sensor installed at the bottom of the drill rod; the variable frequency speed and variable pressure regulating equipment includes a drilling drive motor and a high-pressure variable frequency pump, wherein the drilling drive motor is used to control the drilling speed, the lifting speed of the drill rod and the rotation speed of the fan blades, and the high-pressure variable frequency pump is used to control the spraying pressure and the grouting speed.

9. An intelligent construction control system for splitting shotcrete mixing piles, characterized in that: include: The real-time sensing information acquisition module is configured to: receive real-time sensing information output from the construction information self-sensing system, wherein the real-time sensing information includes drilling depth, drilling rig output power, drill bit torque, drill rod axial force and drill bit pore water pressure data; The real-time formation state information prediction module is configured to: input the real-time perception information into a trained formation information prediction machine learning model, establish a nonlinear implicit correspondence between the real-time perception information and the construction formation conditions, and output the real-time formation state information; the machine learning model is a trained random forest-DBSCAN integrated model; The current optimal construction parameter acquisition module is configured to: input the real-time stratum state information into the trained optimal construction parameter self-matching machine learning model, establish a nonlinear implicit correspondence between the real-time stratum state information and the optimal construction parameters, and output the current optimal construction parameters; 1) Data preprocessing: Remove data outliers based on statistical analysis methods. For locally sparse data, use regression interpolation methods to fill in the data. 2) Feature selection and standardization: The chi-square test method is used to evaluate the correlation between the features and the target variable. The five features with the highest correlation are selected. The Z-score standardization method is used to convert the selected five features into a normal distribution with a mean of 0 and a variance of 1 to make them comparable. 3) Data partitioning: Use the K-fold cross-validation method to divide the database into n datasets of similar size; 4) Custom kernel function: Use custom polynomial kernel function to map data into high-dimensional feature space, thereby converting nonlinear problems into linear problems; 5) Model training and evaluation: Perform K rounds of training set performance evaluation on the built construction parameter self-matching model. In each round, n-1 data sets are used for model training, and the remaining data set is used as a test set. Finally, the average value of the K rounds of evaluation results is calculated as the performance indicator of the model. If it does not meet the requirements, the next round of training is carried out; 6) Prediction of optimal construction parameters: Input real-time formation state information into the trained optimal construction parameter self-matching machine learning model, and obtain the predicted optimal construction parameters through forward propagation; The optimal construction parameter self-matching machine learning model is a specially designed SVM model, and the kernel function of the model is a custom polynomial kernel function.