Intelligent shield muck improvement method and system based on machine learning

Abstract

The application discloses an intelligent shield muck improvement method and system based on machine learning, which realizes the intelligentization and precision of muck improvement of a shield machine in the tunneling process by integrating muck image classification, improvement parameter prediction and optimization model. First, the YOLOV8 network is used to classify the muck image in real time, the XGBOOST model is combined to correct the randomness of classification, and the classification accuracy and stability are improved. Second, the LSTM network is used to predict the time sequence of the muck improvement parameters, and the linear regression and genetic algorithm are introduced to improve the prediction accuracy. Finally, the particle swarm optimization (PSO) algorithm is used to solve the optimal solution of the muck improvement parameters, which significantly improves the stability and efficiency of shield tunneling. The application solves the randomness and hysteresis problems caused by manual adjustment in the prior art, has the advantages of high efficiency and precision, and is suitable for automatic application scenarios of shield engineering.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of shield tunneling construction, and particularly relates to an intelligent shield muck improvement method and system based on machine learning. BACKGROUND

[0002] Shield construction is a widely used tunneling technology in underground engineering. In the process of shield tunneling, muck improvement is an important link to ensure the stability of the pressure in the tunneling cabin, the smoothness of the muck discharge, and the efficient operation of the shield machine. The traditional muck improvement method relies on the experience of the operator to manually adjust the improvement parameters, such as the foam, water, and bentonite flow, by observing the tunneling state and the image of the muck discharge port. This method has obvious randomness and hysteresis, which not only affects the tunneling efficiency, but also may cause equipment damage or engineering quality problems.

[0003] With the development of artificial intelligence technology, machine learning algorithms are increasingly applied in the field of engineering, but existing research on the intelligentization of shield muck improvement is still limited to a single task, such as muck classification or parameter prediction, and lacks a systematic solution from data processing, prediction to optimization. Therefore, there is an urgent need for a comprehensive method to realize the intelligentization and efficiency of shield muck improvement based on machine learning technology. SUMMARY

[0004] To solve the above technical problems, the application proposes an intelligent shield muck improvement method and system based on machine learning, which integrates image classification, parameter prediction, and optimization algorithms to establish a full-process intelligent muck improvement system.

[0005] The first aspect of the application is an intelligent shield muck improvement method based on machine learning, comprising the following steps:

[0006] (1) Collecting muck images and related sensor parameters during shield tunneling;

[0007] (2) Using an improved YOLOV8 network to classify the muck images, and using an XGBOOST model to analyze the feature importance in the muck image classification, adjust the model output probability to correct the classification randomness, and obtain the accurate muck type probability;

[0008] (3) Using an LSTM network to input the time series of sensor parameters during tunneling, combining the muck type probability, and combining linear regression and genetic algorithm to predict the muck improvement parameters needed in the future tunneling process;

[0009] (4) Using a particle swarm optimization PSO algorithm to take the muck type probability and tunneling stability as the objective function, and solving the optimal solution of the bentonite flow, foam liquid flow, and foam air flow;

[0010] (5) The optimized parameters in step (4) are fed back to the shield machine control system to realize real-time monitoring and dynamic adjustment during tunneling.

[0011] Further, the YOLOV8 network is based on an improved VanillaNet structure, and through feature extraction, enhancement and classification prediction, the accuracy of the muck classification is improved to more than 98%.

[0012] Specifically, the output calculation formula of YOLOV8 in step (2) is:

[0013] P cls (i)=σ(W cls ·f i )+b cls

[0014] Where P cls (i) is the classification probability of the i-th class, W cls is the classification weight, f i is the image feature vector, σ is the sigmoid activation function, and b cls is the bias term.

[0015] Specifically, the correction process of the XGBOOST model in step (2) uses the following expression:

[0016]

[0017] Where P XGBoost (i) is the corrected classification probability of the i-th class, α k is the weight of each feature, f k (x) is the contribution of the k-th feature to sample x, and N is the total number of features.

[0018] Specifically, the formula of the LSTM network in step (3) is:

[0019] f t =σ(W f ·[h t-1 ,x t ]+b f );

[0020] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0021]

[0022]

[0023] ot = σ(W o · [h t-1 , x t ] + b o ) ;

[0024] h t = o t · tanh(C t ) ;

[0025] where f t , i t , are forget gate, input gate and candidate memory state respectively, C t is memory cell state, h t is the output of the current time, h t-1 is the hidden state of the previous time step, x t is the input of the current time step, σ is the Sigmoid activation function, W f , b f are the weights and biases of the forget gate respectively, W i , b i are the weights and biases of the input gate respectively, W C , b C are the weights and biases of the candidate memory state respectively, W o , b o are the weights and biases of the output gate respectively;

[0026] The calculation formula of the linear regression is:

[0027]

[0028] where w is the weight of the linear regression, b is the bias term, X scaled is the normalized feature column, is the target column;

[0029] The fitness function of the genetic algorithm is:

[0030]

[0031] where individual is an individual in the genetic algorithm, representing a feature value vector; w and b are the weights and bias terms of the linear regression model; is the target value (processed probability of slag) predicted by the current individual; penalty is a penalty term for processing the penalty when the feature value exceeds the predetermined range.

[0032] Specifically, the fitness calculation formula of the PSO algorithm in step (4) is:

[0033]

[0034] wherein Z is the value of the fitness function, i.e., the value of the objective function; c0is a bias term; c i is the coefficient of the linear term, representing the coefficients of the bentonite flow rate, the foam liquid flow rate, and the foam air flow rate; c ij is the coefficient of the quadratic term, representing the cross terms of x i and x j ; x i is the i-th component of the particle position vector, corresponding to the bentonite flow rate, the foam liquid flow rate, or the foam air flow rate.

[0035] The second aspect of the present application: an intelligent shield muck improvement system based on machine learning, comprising the following modules:

[0036] Parameter acquisition module: acquiring muck images and related sensor parameters during shield tunneling; Muck image real-time classification module: classifying muck images using an improved YOLOV8 network, correcting classification randomness through an XGBOOST model, and obtaining accurate muck type probability;

[0037] Muck improvement parameter prediction module: classifying muck images using an improved YOLOV8 network, adjusting model output probability to correct classification randomness through XGBOOST model analysis of feature importance in muck image classification, and obtaining accurate muck type probability;

[0038] Prediction parameter module: inputting sensor parameter time series during tunneling into an LSTM network, combining muck type probability, and combining linear regression and genetic algorithm to predict muck improvement parameters needed in future tunneling;

[0039] Parameter optimization module: using particle swarm optimization PSO algorithm to solve the optimal solution of bentonite flow rate, foam liquid flow rate, and foam air flow rate with muck type probability and tunneling stability as objective functions;

[0040] Real-time adjustment module: feeding the parameters optimized by the parameter optimization module to the shield machine control system to realize real-time monitoring and dynamic adjustment during tunneling.

[0041] The third aspect of the present application: an electronic device, characterized by comprising:

[0042] One or more processors;

[0043] Memory for storing one or more programs;

[0044] When the one or more programs are executed by the one or more processors, the one or more processors implement the one kind of intelligent shield muck improvement method based on machine learning.

[0045] The fourth aspect of the present application is a computer readable storage medium having stored thereon computer instructions, wherein the instructions, when executed by a processor, implement the steps of the intelligent shield muck improvement method based on machine learning.

[0046] The beneficial effects of the present application are as follows:

[0047] The present application can significantly improve the efficiency of shield tunneling, reduce the randomness and hysteresis caused by human intervention, and has high popularization value in actual engineering. BRIEF DESCRIPTION OF DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0049] Figure 1 The intelligent shield segment muck improvement method based on machine learning of the present application is a technical roadmap.

[0050] Figure 2 The muck improvement process diagram of the present application is a technical roadmap.

[0051] Figure 3 The muck improvement parameter prediction route diagram of the present application is a technical roadmap.

[0052] Figure 4 The muck improvement parameter optimization route diagram of the present application is a technical roadmap. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the drawings of the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0054] The present application discloses an intelligent shield muck improvement method and system based on machine learning, and the embodiments of the present application are further described in detail below. Figure 1 、 Figure 2 and Figure 3 The embodiments of the present application are further described in detail.

[0055] Step 1: Real-time classification of muck images

[0056] (1.1) Data collection and preprocessing

[0057] During shield tunneling, the muck sent out by the screw conveyor is real-time collected by the high-resolution camera at the muck outlet, ensuring clear images without obvious noise. The image data is first extracted by the camera, and then manually classified by the shield driver according to the actual situation, labeling the muck type (such as dry, normal, wet, etc.). Subsequently, the YOLOv8 network is used to process these image data, and the input of the model includes images that have been standardized and enhanced in resolution. In order to improve the inference speed, the network backbone structure uses a lightweight optimized version of VanillaNet.

[0058] (1.2) Classification network training

[0059] First, the YOLOV8 network is trained using a dataset containing different types of muck (dry, normal, wet). The proportion of each type of sample is approximately 1:1:1 to ensure balanced data distribution. During training, considering the randomness of classification results, the XGBOOST model is used to analyze the correlation between historical data (such as thrust pressure, torque, etc.) and YOLOV8 classification results, thereby correcting possible biases in the classification network and improving the accuracy of the model.

[0060] (1.3) Classification result output

[0061] After training, the YOLOV8 network classifies each muck image as dry, normal, or wet muck, outputting the corresponding classification result ("0, 1, 2" representing dry, normal, and wet muck, respectively). This classification result will serve as a reference label for the XGBOOST model, which further calculates the classification probability of each category based on historical data (such as thrust pressure, torque, etc.) and the classification results of YOLOV8. The output calculation formula of YOLOV8 is:

[0062] P cls (i)=σ(W cls ·f i )+b cls ;

[0063] Where P cls (i) is the classification probability of the i-th category, W cls is the classification weight, f i is the image feature vector, σ is the sigmoid activation function, and b cls is the bias term.

[0064] The correction process of the XGBOOST model uses the following expression:

[0065]

[0066] where P XGBoost (i) is the modified classification probability of the i-th class, α k is the weight of each feature, f k (x) is the contribution of the k-th feature on the sample x, and N is the total number of features.

[0067] In this way, the dependence on absolute classification results is reduced, and the flexibility and accuracy of classification are improved.

[0068] Step 2: Prediction of Slag Modification Parameters

[0069] (2.1) Historical data input

[0070] The relevant parameters recorded by the sensors during shield construction (such as cutter torque (X1), average thrust speed (X2), average soil pressure (X3), bentonite flow (X4), foam liquid flow (X5), foam air flow (X6), and tunneling stability parameters (Y2, calculated manually) are input into the LSTM network to form time series data for training the prediction model. The data length is set to 30 time steps (about 30 seconds). These input parameters are obtained through communication with the shield driver, fully considering the actual requirements of calculation time and performance, etc.

[0071] (2.2) LSTM network prediction

[0072] The LSTM network extracts features from historical data and predicts the probability of different types of slag within the next 30 seconds. The slag type probability output by the LSTM is used to reflect the probability of occurrence of different types of slag, thereby providing a basis for subsequent optimization of slag modification parameters.

[0073] The relevant formulas of the LSTM network are as follows:

[0074] The formula of the LSTM network is:

[0075] f t =σ(W f ·[h t-1 ,x t ]+b f );

[0076] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0077]

[0078]

[0079] ot = σ(W o · [h t-1 , x t ] + b o ) ;

[0080] h t = o t · tanh(C t ) ;

[0081] where f t , i t , are forget gate, input gate and candidate memory state respectively, C t is memory cell state, h t is the output at the current time, h t-1 is the hidden state at the previous time step, x t is the input at the current time step, σ is the Sigmoid activation function, W f , b f are the weights and biases of the forget gate, W i , b i are the weights and biases of the input gate, W C , b C are the weights and biases of the candidate memory state, W o , b o are the weights and biases of the output gate.

[0082] (2.3) Linear regression and genetic algorithm optimization

[0083] Based on the relationship of the input data (X1,2,3,4,5,6) in the past 30 seconds, the weights and intercepts of the features are obtained through the linear regression model. The input feature weights and intercepts of the linear regression are derived from historical data to describe the parameter relationship in the past time period.

[0084] The calculation formula of the linear regression is:

[0085]

[0086] where w is the weight of linear regression, b is the bias term, X scaled is the normalized feature column, is the target column;

[0087] Then, the future 30 seconds of slag type probability value predicted by LSTM is combined with genetic algorithm for backstepping optimization. The goal of genetic algorithm is to find a feature combination that makes the predicted value calculated by these features as close as possible to the future 30 seconds of slag type probability value predicted by our LSTM, so as to obtain a more accurate (X1,2,3,4,5,6).

[0088] The genetic algorithm fitness function is:

[0089]

[0090] where individual is an individual in the genetic algorithm, representing a feature value vector; w and b are the weights and bias term of the linear regression model; is the target value predicted by the current individual (processed slag probability); penalty is the penalty term, used to handle the penalty when the feature value exceeds the predetermined range.

[0091] Step 3: Optimization of slag improvement parameters

[0092] (3.1) Optimization goal setting

[0093] This embodiment adopts particle swarm optimization (PSO) algorithm to optimize the slag improvement effect and tunneling stability during the tunneling process of the shield machine. The optimization goals include the following two main goals:

[0094] 1. Maximize the probability of normal slag type: By adjusting the slag improvement parameters, the type of slag encountered during the operation of the shield machine tends to be "normal", thereby improving the tunneling efficiency and stability.

[0095] 2. Maximize the stability of tunneling: This index is based on the sum of the standard deviations of X1 to X6 within a certain period of time, reflecting the stability of the tunneling process of the shield machine. The smaller the standard deviation, the smaller the fluctuation in the tunneling process, and the better the stability.

[0096] As shown in Figure 4 , the optimization objective function considers both the slag improvement effect and the tunneling stability, and the goal is to adjust the bentonite flow, foam liquid flow and foam air flow to achieve the best slag improvement effect and improve the tunneling stability.

[0097] (3.2) Particle initialization

[0098] In the particle swarm, each particle represents a combination of slag improvement parameters (X4, X5, X6), where:

[0099] • X4: bentonite flow

[0100] • X5: foam liquid flow

[0101] • X6: foam air flow

[0102] The initialization of the particle is randomly generated within the set parameter range, which is determined based on historical experience values and actual engineering requirements. For example:

[0103] • The range of bentonite flow rate (X4): set according to the delivery capacity of the bentonite system interface and the demand for soil improvement;

[0104] • The range of foam liquid flow rate (X5) and foam air flow rate (X6): set a reasonable flow rate range based on the proportion of foam liquid and compressed air mixed through the foam generator to form a foam mixture.

[0105] These fluids enter the cutterhead injection port through a common pipeline and participate in the soil improvement process. After the initialization of the particle position and velocity, the optimization process will be updated according to the fitness function in the subsequent optimization process to find the optimal solution.

[0106] (3.3) Particle update and convergence

[0107] In the iteration process of the particle swarm optimization (PSO) algorithm, particles dynamically adjust their positions and velocities according to the predetermined fitness function. After each iteration, the position and velocity of the particle are updated according to the feedback information of the fitness function, so as to calculate the new fitness value.

[0108] The PSO algorithm fitness calculation formula is:

[0109]

[0110] Where Z is the value of the fitness function, i.e. the value of the objective function; c0 is the bias term; c i is the coefficient of the linear term, representing the coefficients of bentonite flow rate, foam liquid flow rate and foam air flow rate; c ij is the coefficient of the quadratic term, representing the cross terms of x i and x j ; x i is the i-th component of the particle position vector, corresponding to the bentonite flow rate, foam liquid flow rate or foam air flow rate.

[0111] This fitness function takes into account the following two key factors:

[0112] (a) Soil type probability error: This factor measures the deviation between the PSO model prediction result and the LSTM prediction value through the soil type probability predicted by the LSTM model. Error evaluation usually uses root mean square error (RMSE) or other error measurement methods, and the optimization goal is to minimize the prediction error to improve the accurate prediction of soil type.

[0113] (b) Tunneling stability parameter: This index reflects the stability of the tunneling process by calculating the sum of the standard deviations of the bentonite flow (X1), the foam liquid flow (X2), the foam air flow (X3), and other related parameters. The smaller the standard deviation, the smaller the fluctuation in the tunneling process, and the better the stability of the system. This stability index is used as an evaluation criterion in the PSO optimization process, and the optimization goal is to maximize the stability of the tunneling process.

[0114] During the iteration process of the PSO algorithm, the particles continuously adjust their positions and velocities, gradually approaching the optimal solution, to optimize the combination of bentonite flow (X4), foam liquid flow (X5), and foam air flow (X6), thereby achieving dual optimization of the spoil improvement effect and tunneling stability.

[0115] In this scheme, the optimization process not only considers the spoil improvement effect in the next 30 seconds, but also optimizes the spoil improvement parameters in the past 30 seconds. By using two PSO algorithms respectively, the best spoil improvement parameters for the next 30 seconds and the past 30 seconds are obtained. The two optimization results will be added according to the fitness weight, and a smaller fitness will obtain a larger weight, thereby ensuring the overall balance and improvement of the optimization goal. Through this mechanism, historical data and future prediction can be effectively combined to obtain more accurate and stable spoil improvement parameters.

[0116] (c) Convergence: After multiple iterations, the particle swarm can quickly converge within 10 iterations, with a significant reduction in RMSE and effective improvement of the optimization goal.

[0117] Step 4: System integration and feedback

[0118] (4.1) Integrated system

[0119] The above modules are connected to the shield machine control system through a data bus, and real-time classification results and parameter prediction values are received and fed back to the tunneling control logic. The system is implemented using C++, ensuring that the reasoning speed meets the real-time requirements of the shield.

[0120] (4.2) Dynamic adjustment

[0121] According to the current spoil classification and predicted parameters, the injection flow of bentonite and foam agent is automatically adjusted, and the adjustment effect is monitored in real time through sensors to form a closed-loop control.

[0122] The present application also discloses an intelligent shield spoil improvement system based on machine learning, which comprises the following modules:

[0123] The collection parameter module collects the muck image and related sensor parameters in the shield tunneling process; the muck image real-time classification module classifies the muck image by using an improved YOLOV8 network, corrects the classification randomness by an XGBOOST model, and obtains an accurate muck type probability;

[0124] The muck improvement parameter prediction module classifies the muck image by using an improved YOLOV8 network, adjusts the model output probability to correct the classification randomness by analyzing the feature importance in the muck image classification through an XGBOOST model, and obtains an accurate muck type probability;

[0125] The prediction parameter module inputs the sensor parameter time sequence in the tunneling process into an LSTM network, combines the muck type probability, and combines linear regression and genetic algorithm to predict the muck improvement parameter required in the future tunneling process;

[0126] The parameter optimization module adopts a particle swarm optimization (PSO) algorithm, takes the muck type probability and the tunneling stability as objective functions, and solves the optimal solution of the bentonite flow, the foam liquid flow and the foam air flow;

[0127] The real-time adjustment module feeds back the parameters optimized by the parameter optimization module to the shield machine control system, realizes real-time monitoring and dynamic adjustment in the tunneling process.

[0128] The application further discloses an electronic device, which comprises:

[0129] One or more processors;

[0130] A memory for storing one or more programs;

[0131] When the one or more programs are executed by the one or more processors, the one or more processors implement the machine learning-based intelligent shield muck improvement method.

[0132] And a computer readable storage medium, which stores computer instructions, characterized in that the instructions are executed by a processor to implement the steps of the machine learning-based intelligent shield muck improvement method.

[0133] It should be noted that, in the present application, the relational terms such as "first" and "second", and the like, are used solely to distinguish one entity or action from another, without necessarily requiring or implying any actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0134] The foregoing is merely illustrative of the principles of the application and various modifications can be made by those skilled in the art without departing from the spirit and scope of the application. The above embodiments are illustrative, and not restrictive, of the scope of the application.

Claims

1. A method for improving a shield muck based on machine learning, characterized by, It comprises the following steps: (1) Collecting the image of muck and related sensor parameters in the process of shield tunneling; (2) Using the improved YOLOV8 network to classify the muck image, and using the XGBOOST model to analyze the feature importance in the muck image classification, adjust the model output probability to correct the classification randomness, and obtain the accurate muck type probability; (3) Using the LSTM network to input the time sequence of sensor parameters in the process of tunneling, combining the muck type probability, and combining linear regression and genetic algorithm to predict the muck improvement parameters needed in the future tunneling process; (4) Using the particle swarm optimization PSO algorithm to take the muck type probability and the tunneling stability as the objective function, and solving the optimal solution of the bentonite flow, the foam liquid flow and the foam air flow; (5) The parameters optimized in step (4) are fed back to the shield machine control system to realize real-time monitoring and dynamic adjustment in the process of tunneling.

2. The method of claim 1, wherein, The YOLOV8 network is based on the improved VanillaNet structure, which improves the accuracy of muck classification to more than 98% through feature extraction, enhancement and classification prediction.

3. The method of claim 1, wherein, The output calculation formula of YOLOV8 in step (2) is: P cls (i) = σ(W cls ·f i )+b cls ; where P cls (i) is the classification probability of the i-th class, W cls is the classification weight, f i is the image feature vector, σ is the sigmoid activation function, and b cls is the bias term.

4. The method of claim 1, wherein, The correction process of XGBOOST model in step (2) uses the following expression: where P XGBoost (i) is the corrected class probability of the i-th class, a k is the weight of each feature, f k (x) is the contribution of the k-th feature on the sample x, and N is the total number of features.

5. The method of claim 1, wherein, The formula of LSTM network in step (3) is: f t = σ(W f · [h t-1 , x t ]+ b f ); i t = σ(W i · [h t-1 , x t ]+ b i ); o t = σ(W o · [h t-1 , x t ]+ b o ); h t = o t tanh(C t ); wherein f t , i t , are the forget gate, the input gate and the candidate memory state, respectively, C t is the memory cell state, h t is the output at the current time step, h t-1 is the hidden state at the previous time step, x t is the input at the current time step, σ is the Sigmoid activation function, W f , b f are the weights and biases of the forget gate, respectively, W i , b i are the weights and biases of the input gate, respectively, W C , b C are the weights and biases of the candidate memory state, respectively, and W o , b o are the weights and biases of the output gate, respectively. The calculation formula of linear regression is: where w is the weight of the linear regression, b is the bias term, X scaled is the normalized feature column, is the target column; The fitness function of genetic algorithm is: Wherein, individual is an individual in the genetic algorithm, indicating a characteristic value vector; w and b are the weight and bias term of the linear regression model; is the target value predicted by the current individual, that is, the processed probability of the slag; penalty is a penalty term for processing the penalty when the characteristic value exceeds the predetermined range.

6. The method of claim 1, wherein, The fitness calculation formula of PSO algorithm in step (4) is: where Z is the value of the fitness function, i.e., the value of the objective function; c0is a bias term; c i is the coefficient of the linear term, representing the coefficients of the bentonite flow rate, the foam liquid flow rate, and the foam air flow rate; c ij is the coefficient of the quadratic term, representing the cross terms of x i and x j ; x i is the i-th component of the particle position vector, corresponding to the bentonite flow rate, the foam liquid flow rate, or the foam air flow rate. 7.A machine learning based intelligent shield muck improvement system, characterized in that, It comprises the following modules: Parameter acquisition module: collecting the image of muck and related sensor parameters in the process of shield tunneling; Real-time muck image classification module: using the improved YOLOV8 network to classify the muck image, and using the XGBOOST model to correct the classification randomness to obtain the accurate muck type probability; Muck improvement parameter prediction module: using the improved YOLOV8 network to classify the muck image, and using the XGBOOST model to analyze the feature importance in the muck image classification, adjust the model output probability to correct the classification randomness, and obtain the accurate muck type probability; Prediction parameter module: using the LSTM network to input the time sequence of sensor parameters in the process of tunneling, combining the muck type probability, and combining linear regression and genetic algorithm to predict the muck improvement parameters needed in the future tunneling process; Parameter optimization module: using the particle swarm optimization PSO algorithm to take the muck type probability and the tunneling stability as the objective function, and solving the optimal solution of the bentonite flow, the foam liquid flow and the foam air flow; Real-time adjustment module: the parameters optimized by the parameter optimization module are fed back to the shield machine control system to realize real-time monitoring and dynamic adjustment in the process of tunneling.

8. An electronic device, comprising: It comprises: One or more processors; Memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1-6.

9. A computer readable storage medium having stored thereon computer instructions, wherein, The instructions are executed by the processor to implement the steps of the method of any one of claims 1-6.

Images