A method, system, electronic device and storage medium for predicting the diameter of a rotary jetting pile

Abstract

The application discloses a rotary jetting pile diameter prediction method and system, electronic equipment and a storage medium, and relates to the fields of geotechnical engineering and artificial intelligence. The method first acquires multi-dimensional characteristics of a site and a measured pile diameter, and introduces a grouting specific energy index to quantize energy input. A historical data set is constructed and subjected to numerical coding and standardization processing. Then, a deep neural network model based on a Transformer encoder is built, and a multi-head self-attention mechanism is used to capture global coupling rules between strata and construction parameters in parallel. Meanwhile, an Optuna automatic optimization framework is established, and a TPE algorithm and a pruning strategy are used to iteratively search for a globally optimal super parameter combination. Finally, preset parameters are input to complete pile diameter prediction. The application effectively solves the problems of insufficient feature extraction capability, dependence on artificial experience for super parameter adjustment and poor generalization of traditional methods, and realizes deep fusion of heterogeneous data.

Description

Technical Field

[0001] This invention relates to the field of geotechnical engineering and artificial intelligence, and in particular to a method, system, electronic device and storage medium for predicting the diameter of jet grouting piles. Background Technology

[0002] High-pressure jet grouting is a key technology for treating weak foundations, reinforcing existing building foundations, and constructing cutoff walls for deep foundation pits, playing a crucial role in modern underground space development. Its basic principle involves using a drilling rig to lower a grouting pipe with a nozzle to a predetermined depth in the stratum. A high-pressure generator then propels cement grout or water into a high-speed jet horizontally from the nozzle. This powerful fluid kinetic energy impacts and cuts the soil, breaking it up and thoroughly mixing it with the grout. After solidification, it forms a consolidated pile with a specific strength and geometric shape.

[0003] In the design and construction quality control of jet grouting piles, the pile diameter is a core physical indicator for measuring the reinforcement effect, seepage prevention performance, and engineering bearing capacity. Since jet grouting pile construction is an underground concealed project, the pile diameter is affected by multiple coupling factors, including the construction environment, the physical and mechanical properties of the strata (such as soil quality, density, and moisture content), and construction process parameters (such as grouting pressure, lifting speed, and grout flow rate), exhibiting extremely strong nonlinear characteristics.

[0004] Currently, commonly used prediction methods in engineering mainly include empirical formula methods and traditional machine learning methods. Empirical formula methods are usually based on fitting limited data from specific experimental sites. Although they are simple to calculate, the applicable range of parameters is narrow. When faced with complex and variable geological conditions, they often have large prediction errors, which leads to the need for a large number of test piles on site, increasing construction costs and time.

[0005] Although algorithms such as Support Vector Regression (SVR), Artificial Neural Networks (ANN), and Long Short-Term Memory (LSTM) have been applied to pile diameter prediction in recent years, these technologies have revealed significant shortcomings in practical applications. First, traditional deep learning models have limitations when processing the feature data of jet grouting piles. Influencing factors include discrete geological parameters and continuous construction parameters, which are typical multi-source heterogeneous data. There are global physical coupling relationships between the parameters. Existing shallow models or recurrent neural networks cannot capture the complex global dependencies between these heterogeneous features in parallel, resulting in insufficient feature extraction. Second, existing models mostly rely on manual experience or grid search for hyperparameter setting, which is not only inefficient but also prone to getting trapped in local optima when facing small sample high-dimensional data, resulting in unstable model generalization ability.

[0006] Therefore, there is an urgent need for a high-precision method for predicting the diameter of jet grouting piles that can deeply capture the complex physical dependencies between heterogeneous features and has fully automated optimization capabilities. This is to enable timely optimization of construction plans and thus improve project quality. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a method, system, electronic device, and storage medium for predicting the diameter of jet grouting piles. Addressing the problems of insufficient feature extraction capabilities, reliance on manual experience for hyperparameter adjustment, and poor generalization ability in existing jet grouting pile diameter prediction methods, this invention provides a jet grouting pile diameter prediction method based on a Transformer encoder. This method utilizes a self-attention mechanism to uncover the global coupling patterns between the formation and construction parameters, and combines this with the Optuna automated search framework to achieve optimal model configuration.

[0008] On the one hand, a method for predicting the diameter of jet grouting piles is provided, including: Acquire multidimensional feature data of the jet grouting pile construction site and the corresponding on-site measured pile diameter; A historical dataset of jet grouting piles is constructed and preprocessed. The preprocessed dataset is divided into a training set and a test set. The historical dataset of jet grouting piles includes: stratum parameters, construction parameters, and jet grouting pile diameter. The preprocessing includes: numerical coding and standardization preprocessing. A jet grouting pile diameter prediction model is constructed, which is used to capture the nonlinear mapping relationship between input features and pile diameter. The training set is input into the jet grouting pile diameter prediction model to train the model. During the training process, the Optuna hyperparameter automatic optimization framework is used to search for parameters, aiming to minimize the prediction error and find the globally optimal hyperparameter combination to obtain the trained jet grouting pile diameter prediction model. The jet grouting pile diameter prediction model can be used to predict the diameter of jet grouting piles. Input the preset jet grouting pile construction parameters and stratum parameters into the trained jet grouting pile diameter prediction model, and output the corresponding predicted value of the jet grouting pile diameter.

[0009] On the other hand, a system for predicting the diameter of jet grouting piles is provided, including: The acquisition module is configured to acquire multi-dimensional feature data of the jet grouting pile construction site and the corresponding on-site measured pile diameter. The preprocessing module is configured to: construct a historical dataset of jet grouting piles and preprocess it, dividing the preprocessed dataset into a training set and a test set; the historical dataset of jet grouting piles includes: stratum parameters, construction parameters, and jet grouting pile diameter; the preprocessing includes: numerical coding and standardization preprocessing; The model building module is configured to: build a jet grouting pile diameter prediction model, which is used to capture the nonlinear mapping relationship between input features and pile diameter; The model training module is configured to: input the training set into the jet grouting pile diameter prediction model, train the jet grouting pile diameter prediction model, and use the Optuna hyperparameter automatic optimization framework to search for parameters during the training process, with the goal of minimizing the prediction error and finding the globally optimal hyperparameter combination to obtain the trained jet grouting pile diameter prediction model; the jet grouting pile diameter prediction model can be used to predict the diameter of jet grouting piles. The output module is configured to input the preset jet grouting pile construction parameters and stratum parameters into the trained jet grouting pile diameter prediction model and output the corresponding predicted value of the jet grouting pile diameter.

[0010] Furthermore, an electronic device is also provided, including: Memory, used for non-transitory storage of computer-readable instructions; and Processor, for executing the computer-readable instructions, When the computer-readable instructions are executed by the processor, they perform the method described in the first aspect above.

[0011] In another aspect, a storage medium is also provided for non-transitory storage of computer-readable instructions, wherein when the non-transitory computer-readable instructions are executed by a computer, the method described in the first aspect is performed.

[0012] In another aspect, a computer program product is also provided, including a computer program that, when run on one or more processors, is used to implement the method described in the first aspect above.

[0013] The above technical solution has the following advantages or beneficial effects: First, this invention employs a Transformer encoder model, which captures the coupling relationship between heterogeneous geological data and construction specific energy through input embedding and self-attention mechanisms. This effectively addresses the high complexity and uncertainty of the jet grouting pile formation process. Simultaneously, it adaptively allocates the weights of influencing factors, significantly improving the feature extraction depth compared to traditional models, thus ensuring prediction accuracy meets the construction requirements under complex conditions.

[0014] Secondly, this invention employs the Optuna automatic optimization framework, which uses the TPE algorithm for intelligent sampling and combines it with a pruning mechanism to dynamically adjust the model structure. This overcomes the blindness of manual parameter tuning, improves optimization efficiency, and enhances model robustness. Because it possesses global search capabilities under small sample conditions, it effectively prevents model overfitting, ensures generalization stability under different construction site environments, and provides more scientific algorithmic support for pile diameter prediction.

[0015] Third, by establishing an end-to-end prediction process, this invention enables real-time dynamic prediction of pile diameter based on geological conditions, enhancing construction predictability. Construction personnel can pre-adjust operating parameters to obtain the target pile diameter, reduce the number of test piles, save costs, and better guide refined construction, which is the current trend in jet grouting pile diameter prediction. Attached Figure Description

[0016] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0017] Figure 1 This is a flowchart of a jet grouting pile diameter prediction method based on a Transformer encoder, according to an embodiment of the present invention. Figure 2 This is a diagram illustrating the overall structure of a deep neural network model based on a Transformer encoder, according to an embodiment of the present invention. Figure 3 This is an internal structure diagram of the Transformer encoder according to an embodiment of the present invention; Figure 4 This is a scatter plot comparing the predicted and measured values ​​of the jet grouting pile diameter in the preset test set of this invention. Detailed Implementation

[0018] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0020] All data acquisition in this embodiment is carried out in accordance with laws and regulations and with user consent, and the data is used legally.

[0021] Example 1 This embodiment provides a method for predicting the diameter of jet grouting piles; like Figure 1 As shown, a method for predicting the diameter of a jet grouting pile includes: S101: Obtain multi-dimensional feature data of the jet grouting pile construction site and the corresponding on-site measured pile diameter; S102: Construct a historical dataset of jet grouting piles and preprocess it, dividing the preprocessed dataset into a training set and a test set; the historical dataset of jet grouting piles includes: stratum parameters, construction parameters, and jet grouting pile diameter; the preprocessing includes: numerical coding and standardization preprocessing; S103: Construct a jet grouting pile diameter prediction model, which is used to capture the nonlinear mapping relationship between input features and pile diameter; S104: Input the training set into the jet grouting pile diameter prediction model to train the jet grouting pile diameter prediction model. During the training process, the Optuna hyperparameter automatic optimization framework is used to search for parameters, with the goal of minimizing the prediction error, to find the globally optimal hyperparameter combination, and obtain the trained jet grouting pile diameter prediction model; the jet grouting pile diameter prediction model can be used to predict the diameter of jet grouting piles. S105: Input the preset jet grouting pile construction parameters and stratum parameters into the trained jet grouting pile diameter prediction model, and output the corresponding predicted value of the jet grouting pile diameter.

[0022] In this embodiment, step S101 involves acquiring multidimensional feature data of the jet grouting pile construction site and the corresponding measured pile diameter on site. The multidimensional feature data includes soil type, standard penetration test N value (SPT-NValue), and grouting specific energy at the nozzle.

[0023] In this embodiment, on-site construction data of high-pressure jet grouting piles are collected based on a soft soil foundation reinforcement project. The multi-dimensional feature data acquired in this embodiment covers the two core factors determining the quality of jet grouting pile formation: geological conditions and construction technology.

[0024] The N-value in the standard penetration test is the number of blows made by the penetration hammer in the standard penetration test.

[0025] Among them, the grouting specific energy at the nozzle is a comprehensive physical index used to quantify the effective energy input received per unit pile length. Its calculation formula is as follows:

[0026] In the formula, The grouting specific energy at the nozzle is expressed in MJ / m. The grouting pressure is expressed in MPa. This refers to the slurry flow rate, expressed in L / min. The nozzle's average lifting speed is expressed in cm / min; the coefficient 0.9 represents the energy utilization efficiency coefficient.

[0027] In this embodiment, the preprocessing includes: numerical encoding and standardization preprocessing; Numerical encoding refers to using a label encoder to map discrete "soil type" identifiers into a sequence of integer indices, with the mapping being: no fine-grained coarse-grained soil = 0, containing fine-grained coarse-grained soil = 1, and fine-grained soil = 2.

[0028] The standardization process refers to the use of a standard scaler to normalize the mean and variance of the N-value and grouting specific energy of continuous standard penetration tests. The calculation formula is as follows:

[0029] In the formula, This is the original data; The data is standardized. The mean of the sample; The standard deviation is denoted as .

[0030] In this embodiment, the preprocessed dataset is divided into a training set and a test set, specifically including: A stratified sampling method was used, dividing the preprocessed dataset into a training set (80%) and a test set (20%), containing input and output data. The input data consisted of a soil type index, a standardized N-value, and grouting specific energy; the output data was the pile diameter.

[0031] In this embodiment, as Figure 2 and Figure 3 As shown, the jet grouting pile diameter prediction model includes: The system comprises an input embedding module, a position encoding module, a feature extraction module, and a fully connected regression module, which are connected in sequence. The input embedding module is used to achieve semantic fusion of heterogeneous data, the position encoding module is used to introduce sequence topological information, the feature extraction module is used to extract high-order features through a multi-head self-attention mechanism, and the fully connected regression module is used to perform deep nonlinear mapping and output the predicted pile diameter.

[0032] The input embedding module sets the embedding dimension. The soil type index is mapped to a 64-dimensional vector through an embedding layer, and the standard penetration test N value and the grouting ratio at the nozzle are projected to a 64-dimensional vector through a linear layer. These vectors are then stitched together to form... The feature sequence. The input embedding module is used for heterogeneous data semantic fusion. It maps the soil index into a dense vector through the embedding layer and maps the N value and specific energy into a feature vector through the linear projection layer.

[0033] The position encoding module is used to introduce sequence topological information, generate a position encoding matrix using sine and cosine functions, and add the position encoding matrix to the feature sequence element by element to introduce the relative position information of the features in the sequence.

[0034] The feature extraction module includes: a multi-layer stacked encoder layer. Each encoder layer includes: a multi-head attention mechanism layer, a first-layer normalization module, a feedforward neural network (FFN), and a second-layer normalization module, connected in sequence. The input of the multi-head self-attention mechanism layer is also connected to the input of the first-layer normalization module; The output of the first-level normalization module is connected to the input of the second-level normalization module; The input of the feedforward neural network is connected to the input of the second-layer normalization module.

[0035] The Transformer feature extraction module comprises multiple stacked encoder layers, extracting high-order features through a multi-head self-attention mechanism. Residual connections and layer normalization are configured to stabilize gradient propagation and accelerate model convergence.

[0036] In this embodiment, the multi-head self-attention mechanism is used to calculate the association weights between features:

[0037] In the formula, These are respectively query, key, and value vectors; The scaling factor is used. A multi-head mechanism is employed to perform calculations in parallel, capturing the complex global dependencies between soil properties, formation hardness, and grouting energy.

[0038] The calculation formula for a feedforward neural network is:

[0039] In the formula, , This is the weight matrix. , For bias terms, This represents the ReLU activation function. A feedforward neural network is used to perform nonlinear spatial transformations on the features extracted by the attention mechanism, enhancing the model's expressive power.

[0040] The fully connected regression module includes a flattening layer, a multilayer perceptron (MLP), and a linear output layer. This module maps the high-dimensional abstract features output by the feature extraction module to the final predicted pile diameter.

[0041] The multilayer perceptron is equipped with a Dropout layer, and the random inactivation rate is set to 0.1 to 0.3 to improve the model's generalization performance and prevent overfitting. The fully connected regression module maps the extracted high-order features to the final continuous pile diameter values.

[0042] In this embodiment, the Optuna hyperparameter automatic optimization framework is adopted. The TPE algorithm is used to perform iterative trials on the training set, repeatedly adjusting the parameters until the optimal trial count of 50 is found. The specific operation is as follows: (a) Define the search space: feature dimension Attention count number of encoder layers Learning rate .

[0043] (b) Sampling and iteration: Using the TPE algorithm, based on Bayesian optimization, hyperparameter combinations are sampled and experimental trials are constructed. The model is dynamically adjusted based on historical trial feedback. (c) Pruning strategy: Enable the median pruner. If the intermediate validation set error of the current trial is higher than the historical average during training, the training will be terminated early.

[0044] After 50 iterations of optimization, the globally optimal hyperparameter combination is obtained as follows: Attention count number of encoder layers Learning rate A training model for predicting the diameter of jet grouting piles is generated using the optimal combination of hyperparameters.

[0045] Furthermore, the method also includes: The test set is input into the trained jet grouting pile diameter prediction model, and the accuracy of the trained jet grouting pile diameter prediction model is evaluated using a cost function. The cost function uses the coefficient of determination. The mean absolute error (MAE) is used for evaluation, and the calculation formula is as follows:

[0046]

[0047] In the formula, For the actual measured pile diameter, To predict the pile diameter, For the sample size, This represents the average measured pile diameter. (Coefficient of determination) The mean absolute error (MAE) is used to characterize the extent to which the model explains the variability of pile diameter. It is used to characterize the average absolute deviation between the predicted pile diameter and the measured pile diameter.

[0048] Coefficient of determination The hyperparameters are used to characterize the model's ability to explain pile diameter variability; a value closer to 1 indicates a better fit. The mean absolute error (MAE) characterizes the average absolute deviation between the predicted and measured pile diameters. By calculating these metrics on the test set, the search direction for hyperparameters is guided, and the optimal model instance is determined. By calculating and evaluating these metrics on the test set, the optimal model instance is identified, validating the model's generalization ability under unknown working conditions.

[0049] like Figure 4 As shown, the retained test set data is input into the model trained by this invention for verification. Figure 4 A scatter plot showing the comparison between the predicted and measured values ​​of pile diameter in the preset test set is presented.

[0050] Calculations show that the prediction model based on the Transformer encoder in this embodiment exhibits excellent evaluation metrics on the test set, with the optimal coefficient of determination being [missing information]. The mean absolute error (MAE) reached 0.8864, with a mean absolute error (MAE) of only 0.1116m. Furthermore, the mean square error (MSE) was as low as 0.0207m. 2 .

[0051] The results show that the predicted values ​​are closely distributed in Near the diagonal line, it is evident that the method of this invention can effectively capture the nonlinear relationship between the formation and construction parameters. Compared to existing technologies... Compared to traditional empirical formulas and shallow neural network methods, which generally have low prediction accuracy, this invention significantly improves prediction accuracy and generalization stability through the self-attention mechanism of the Transformer, resulting in minimal prediction bias. It can meet the dynamic prediction accuracy requirements for jet grouting pile construction under complex geological conditions. The prediction accuracy is significantly superior to traditional empirical formulas and shallow neural network methods.

[0052] Example 2 This embodiment provides a system for predicting the diameter of jet grouting piles, including: The acquisition module is configured to acquire multi-dimensional feature data of the jet grouting pile construction site and the corresponding on-site measured pile diameter. The preprocessing module is configured to: construct a historical dataset of jet grouting piles and preprocess it, dividing the preprocessed dataset into a training set and a test set; the historical dataset of jet grouting piles includes: stratum parameters, construction parameters, and jet grouting pile diameter; the preprocessing includes: numerical coding and standardization preprocessing; The model building module is configured to: build a jet grouting pile diameter prediction model, which is used to capture the nonlinear mapping relationship between input features and pile diameter; The model training module is configured to: input the training set into the jet grouting pile diameter prediction model, train the jet grouting pile diameter prediction model, and use the Optuna hyperparameter automatic optimization framework to search for parameters during the training process, with the goal of minimizing the prediction error and finding the globally optimal hyperparameter combination to obtain the trained jet grouting pile diameter prediction model; the jet grouting pile diameter prediction model can be used to predict the diameter of jet grouting piles. The output module is configured to input the preset jet grouting pile construction parameters and stratum parameters into the trained jet grouting pile diameter prediction model and output the corresponding predicted value of the jet grouting pile diameter.

[0053] It should be noted that the acquisition module, preprocessing module, model building module, model training module, and output module mentioned above correspond to steps S101 to S105 in Embodiment 1. The examples and application scenarios implemented by these modules and their corresponding steps are the same, but they are not limited to the content disclosed in Embodiment 1. It should be noted that these modules, as part of the system, can be executed in a computer system such as a set of computer-executable instructions.

[0054] The descriptions of each embodiment in the above embodiments have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0055] The proposed system can be implemented in other ways. For example, the system embodiments described above are merely illustrative, and the division of modules described above is only a logical functional division. In actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed.

[0056] Example 3 This embodiment also provides an electronic device, including: one or more processors, one or more memories, and one or more computer programs; wherein, the processor is connected to the memory, and the one or more computer programs are stored in the memory. When the electronic device is running, the processor executes the one or more computer programs stored in the memory to cause the electronic device to perform the method described in Embodiment 1.

[0057] It should be understood that in this embodiment, the processor can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0058] Memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of memory may also include non-volatile random access memory. For example, memory may also store information about the device type.

[0059] In the implementation process, each step of the above method can be completed by the integrated logic circuits in the processor hardware or by software instructions.

[0060] The method in Embodiment 1 can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor. The software modules can reside in readily available storage media in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, a detailed description is not provided here.

[0061] Those skilled in the art will recognize that the units and algorithm steps described in connection with the various examples of this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.

[0062] Example 4 This embodiment also provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the method described in Embodiment 1.

[0063] Example 5 This embodiment also provides a computer program product, including a computer program that, when executed by a processor, implements the method in Embodiment 1.

[0064] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which execute in a device on a target real or virtual processor to perform the processes / methods described above. Typically, program modules include routines, programs, libraries, objects, classes, components, data structures, etc., that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of program modules can be combined or divided among program modules as needed. The machine-executable instructions for the program modules can execute within a local or distributed device. In a distributed device, the program modules can reside in both local and remote storage media.

[0065] The computer program code used to implement the methods of the present invention may be written in one or more programming languages. This computer program code may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the computer or other programmable data processing device, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a computer, partially on a computer, as a stand-alone software package, partially on a computer and partially on a remote computer, or entirely on a remote computer or server.

[0066] In the context of this invention, computer program code or related data may be carried by any suitable carrier to enable a device, apparatus, or processor to perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals may include electrical, optical, radio, sound, or other forms of propagation signals, such as carrier waves, infrared signals, etc.

[0067] Those skilled in the art will recognize that the units and algorithm steps described in conjunction with the embodiments herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0068] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting the diameter of jet grouting piles, characterized in that, include: Acquire multidimensional feature data of the jet grouting pile construction site and the corresponding on-site measured pile diameter; Construct a historical dataset of jet grouting piles and preprocess it, then divide the preprocessed dataset into a training set and a test set; The historical dataset of jet grouting piles includes: geological parameters, construction parameters, and jet grouting pile diameter; the preprocessing includes: numerical coding and standardized preprocessing. A jet grouting pile diameter prediction model is constructed, which is used to capture the nonlinear mapping relationship between input features and pile diameter. The training set is input into the jet grouting pile diameter prediction model to train the model. During the training process, the Optuna hyperparameter automatic optimization framework is used to search for parameters, aiming to minimize the prediction error and find the globally optimal hyperparameter combination to obtain the trained jet grouting pile diameter prediction model. The jet grouting pile diameter prediction model can be used to predict the diameter of jet grouting piles. Input the preset jet grouting pile construction parameters and stratum parameters into the trained jet grouting pile diameter prediction model, and output the corresponding predicted value of the jet grouting pile diameter.

2. The method for predicting the diameter of a jet grouting pile as described in claim 1, characterized in that, The preprocessing includes: numerical encoding and standardization preprocessing; Numerical encoding refers to using a tag encoder to map discrete soil type identifiers into a sequence of integer indices. Standardization processing refers to the process of removing the mean and normalizing the variance of the N-value and grouting ratio energy of continuous standard penetration tests using a standard scaler. The calculation formula is as follows: ； In the formula, This is the original data; The data is standardized. The mean of the sample; The standard deviation is denoted as .

3. The method for predicting the diameter of a jet grouting pile as described in claim 1, characterized in that, The jet grouting pile diameter prediction model includes: The system comprises an input embedding module, a position encoding module, a feature extraction module, and a fully connected regression module, which are connected in sequence. The input embedding module is used to achieve semantic fusion of heterogeneous data, the position encoding module is used to introduce sequence topological information, the feature extraction module is used to extract high-order features through a multi-head self-attention mechanism, and the fully connected regression module is used to perform deep nonlinear mapping and output the predicted pile diameter.

4. The method for predicting the diameter of a jet grouting pile as described in claim 3, characterized in that, The feature extraction module includes: a multi-layered stacked encoder layer; Each encoder layer includes: a multi-head self-attention mechanism layer, a first-layer normalization module, a feedforward neural network (FFN), and a second-layer normalization module, connected in sequence. The input of the multi-head self-attention mechanism layer is also connected to the input of the first-layer normalization module; The output of the first-level normalization module is connected to the input of the second-level normalization module; The input of the feedforward neural network is connected to the input of the second-layer normalization module; Multi-head self-attention mechanisms are used to calculate the association weights between features: ； In the formula, These are respectively query, key, and value vectors; The scaling factor is used; the calculation is performed in parallel through a multi-head mechanism to capture the complex global dependencies between soil properties, formation hardness and grouting energy. The calculation formula for a feedforward neural network is: ； In the formula, , This is the weight matrix. , For bias terms, Represents the ReLU activation function; a feedforward neural network used to perform nonlinear spatial transformations on the features extracted by the attention mechanism, enhancing the model's expressive power.

5. The method for predicting the diameter of a jet grouting pile as described in claim 1, characterized in that, The Optuna hyperparameter automatic optimization framework was adopted, and the TPE algorithm was used to conduct iterative trials on the training set. After repeated adjustments, the optimal trial was found to be 50 times. The specific operation is as follows: (a) Define the search space: feature dimension Attention count number of encoder layers Learning rate ; (b) Sampling and iteration: Using the TPE algorithm, based on Bayesian optimization, hyperparameter combinations are sampled and experimental trials are constructed. The model is dynamically adjusted based on historical trial feedback. (c) Pruning strategy: Enable median pruning. If the intermediate validation set error of the current Trial during training is higher than the historical average, the training will be terminated early. After 50 iterations of optimization, the globally optimal hyperparameter combination is obtained as follows: Attention count number of encoder layers Learning rate A training model for predicting the diameter of jet grouting piles is generated using the optimal combination of hyperparameters.

6. The method for predicting the diameter of a jet grouting pile as described in claim 1, characterized in that, The method further includes: The test set is input into the trained jet grouting pile diameter prediction model, and the accuracy of the trained jet grouting pile diameter prediction model is evaluated using a cost function. The cost function uses the coefficient of determination. The mean absolute error (MAE) is used for evaluation, and the calculation formula is as follows: ； ； In the formula, For the actual measured pile diameter, To predict the pile diameter, For the sample size, The mean of the measured pile diameter; coefficient of determination The mean absolute error (MAE) is used to characterize the extent to which the model explains the variability of pile diameter. It is used to characterize the average absolute deviation between the predicted pile diameter and the measured pile diameter.

7. The method for predicting the diameter of a jet grouting pile as described in claim 3, characterized in that, The input embedding module sets the embedding dimension. The soil type index is mapped to a 64-dimensional vector through an embedding layer, and the standard penetration test N value and the grouting ratio at the nozzle are projected to a 64-dimensional vector through a linear layer. These vectors are then stitched together to form... The feature sequence; the input embedding module is used for heterogeneous data semantic fusion, which maps the soil index into a dense vector through the embedding layer, and maps the N value and specific energy into a feature vector through the linear projection layer; The position encoding module is used to introduce sequence topological information, generate a position encoding matrix using sine and cosine functions, and add the position encoding matrix to the feature sequence element by element to introduce the relative position information of the feature in the sequence.

8. A prediction system for the diameter of jet grouting piles, characterized in that, include: The acquisition module is configured to acquire multi-dimensional feature data of the jet grouting pile construction site and the corresponding on-site measured pile diameter. The preprocessing module is configured to: construct a historical dataset of jet grouting piles and preprocess it, dividing the preprocessed dataset into a training set and a test set; The historical dataset of jet grouting piles includes: geological parameters, construction parameters, and jet grouting pile diameter; the preprocessing includes: numerical coding and standardized preprocessing. The model building module is configured to: build a jet grouting pile diameter prediction model, which is used to capture the nonlinear mapping relationship between input features and pile diameter; The model training module is configured to: input the training set into the jet grouting pile diameter prediction model, train the jet grouting pile diameter prediction model, and use the Optuna hyperparameter automatic optimization framework to search for parameters during the training process, with the goal of minimizing the prediction error and finding the globally optimal hyperparameter combination to obtain the trained jet grouting pile diameter prediction model; the jet grouting pile diameter prediction model can be used to predict the diameter of jet grouting piles. The output module is configured to input the preset jet grouting pile construction parameters and stratum parameters into the trained jet grouting pile diameter prediction model and output the corresponding predicted value of the jet grouting pile diameter.

9. An electronic device, characterized in that it comprises: Memory is used to store computer-readable instructions in a non-transitory manner. as well as Processor, for executing the computer-readable instructions, When the computer-readable instructions are executed by the processor, they perform the method described in any one of claims 1-7.

10. A storage medium, characterized in that, Non-transitory storage of computer-readable instructions, wherein when the non-transitory computer-readable instructions are executed by a computer, the method of any one of claims 1-7 is performed.

Images