A shield tunneling machine digital cockpit method and system based on artificial intelligence

Abstract

The application belongs to the technical field of artificial intelligence, and discloses a shield machine digital cockpit method and system based on artificial intelligence, which comprises: adaptive three-dimensional digital model establishment: parameterized modeling is adopted, so that the model can adaptively match shield equipment of any size; shield machine tunneling parameter prediction model: based on conditional generative adversarial neural network, aiming at data types, a prediction model between related tunneling parameters is established by combining various neural networks; shield machine decision optimization algorithm: based on NSGA-III multi-objective optimization algorithm, an optimal decision solution is given in the tunneling process; digital twin cockpit interactive interface construction: in order to enable the operator to effectively utilize the digital twin cockpit, an intuitive and friendly interactive interface is provided; the interface allows the operator to view and analyze the prediction results, adjust and optimize the parameters, and observe the actual influence of these adjustments on the three-dimensional digital model.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of artificial intelligence, and particularly relates to a shield machine digital cockpit method and system based on artificial intelligence. BACKGROUND

[0002] The shield machine is a key equipment for modern underground engineering construction, and plays an important role in the construction of infrastructure such as subways, tunnels and pipelines. However, the traditional shield machine construction decision-making is heavily dependent on human experience, and due to human uncertainty, problems and safety hazards are easily caused. In the complex underground environment, according to the experience of the driver, inappropriate driving parameters may be used in the excavation process, resulting in low drilling efficiency, serious disc cutter wear, and even the risk of groundwater leakage and tunnel collapse. In addition, the delay and uncertainty of information exchange between the control room and the construction site affect the accuracy and timeliness of decision-making. Considering the many uncontrollable factors in construction, such as geological conditions and groundwater level, relying on traditional manual decision-making will increase the risk of errors and accidents. How to propose correct and reasonable decision-making schemes and timely feedback to improve the safety and operating efficiency of the shield machine is an important problem to be solved.

[0003] The integration of digital technology in the shield machine significantly improves construction efficiency and construction quality, providing accurate and real-time data to strengthen project management and decision-making. Digital twin technology uses data from numerous sensors to achieve real-time monitoring and thorough analysis of the construction process, ensuring accurate and detailed information support. It combines these data with intelligent algorithms to predict results and optimize operations, thereby creating a digital mirror of the physical reality. In addition, digital twin technology constitutes the basic logic behind the digital twin cockpit, which is a practical expression of the digital twin principle. As an innovative solution, the digital twin cockpit uses digital and intelligent systems to improve the operating experience of machinery. It integrates data collection, data processing and data analysis technologies to bring internal and external information together into a unified control interface. The essence of the digital twin cockpit is to develop equipment into an intelligent system through real-time monitoring, information processing and decision optimization to improve efficiency and safety. This technology involves digital modeling and simulation of various data samples and operating states of the actual equipment, ensuring seamless connection between the virtual and physical domains. Therefore, through the digital twin cockpit, operators can access comprehensive data and state information, including speed, position, power and mechanical conditions, allowing them to fully control and optimize shield machine excavation decisions.

[0004] Through the above analysis, the problems and defects of the prior art are: in order to improve the tunneling safety and efficiency of the shield machine, how to build a three-dimensional digital model corresponding to the shield machine entity, how to build a robust machine learning method to realize decision control, and how to design an interactive interface to realize good human-machine connection are the key problems of establishing a digital twin cockpit. SUMMARY

[0005] In view of the problems existing in the prior art, the present application provides a shield machine digital cockpit method and system based on artificial intelligence, aiming at the high uncertainty problem of mainly relying on manual decision in the shield machine construction process, introducing the key idea of digital twin, combining advanced machine learning methods, realizing accurate prediction of related parameters in the shield machine tunneling process, and giving corresponding optimization guidance, reducing the danger in the construction process, and improving the tunneling efficiency.

[0006] The present application is implemented in this way, a shield machine digital cockpit method based on artificial intelligence, comprising:

[0007] S1, adaptive three-dimensional digital model establishment: parameterized modeling is adopted, so that the model can adaptively match shield equipment of any size;

[0008] S2, shield machine tunneling parameter prediction model: based on conditional generative adversarial neural network, for data type, combined with various neural networks (such as: deep neural network, convolutional neural network and residual neural network), a prediction model between related tunneling parameters is established;

[0009] S3, shield machine decision optimization algorithm: based on NSGA-III multi-objective optimization algorithm, the optimal decision solution is given in the tunneling process;

[0010] S4, digital twin cockpit interactive interface construction: in order to enable the operator to effectively utilize the digital twin cockpit, an intuitive and friendly interactive interface is provided; the interface allows the operator to view and analyze the prediction results, adjust the optimization parameters, and observe the actual impact of these adjustments on the three-dimensional digital model.

[0011] Furthermore, S1 specifically includes three main systems based on the functions of the tunnel boring machine (TBM) components: an excavation system, a support system, and an earthmoving system. Modeling is performed using Solidworks software, focusing on the main components of these three systems: the cutterhead model, the hydraulic cylinder model, and the segment assembly machine model. During parametric modeling, constraints and dependencies between component features are established according to the design intent, ensuring that every modification to the core parameters results in model changes consistent with predefined constraints and specifications. All three main component models use the cutterhead diameter as the core parameter, linking component dimensions to the cutterhead diameter through reasonable quantitative relationships. Quantitative relationships are used as the modeling logic, and built-in equation functionality is used to create global variables based on the core parameters. These global variable values, or functions containing global variable values, are then used to calibrate component dimensions. Therefore, the dimensions of all model components are represented as equations related to the core parameters, allowing for the modification of only the core parameter values ​​to obtain the corresponding digital model based on the actual dimensions of the TBM, achieving adaptive model matching.

[0012] Furthermore, S2 specifically includes: In order to realistically reflect the tunnel boring machine excavation scenario, it is necessary to construct machine learning models of actual operating parameters and decision parameters, and embed them into the digital twin cockpit.

[0013] Furthermore, S3 specifically includes: after establishing a prediction model between tunneling parameters, the optimization algorithm can optimize the decision parameters and integrate the optimization results into the cockpit, with the specific optimization target value given based on engineering experience.

[0014] Furthermore, in S4, the interface consists of two main parts: the main interface and sub-interfaces. The main interface displays the dynamic tunneling posture and some basic information of the tunnel boring machine during construction. At the bottom of the main interface are six input buttons for the sub-interfaces. The sub-interfaces handle functions such as data analysis, displaying optimization results, and querying logs. The cockpit interface is created using the Unity engine.

[0015] Another object of the present invention is to provide an AI-based digital cockpit system for tunnel boring machines that implements the aforementioned AI-based digital cockpit method, comprising:

[0016] Adaptive 3D Digital Model Building Module: Used for parametric modeling, enabling the model to adaptively match any size tunnel boring machine;

[0017] The tunnel boring machine excavation parameter prediction model establishment module is used to establish a prediction model between relevant tunneling parameters based on conditional generative adversarial neural networks and various neural networks (such as deep neural networks, convolutional neural networks, and residual neural networks) for different data types.

[0018] Tunnel Boring Machine Decision Optimization Module: Used to provide the optimal decision solution during tunneling based on the NSGA-III multi-objective optimization algorithm;

[0019] Digital Twin Cockpit Interface: This interface provides an intuitive and user-friendly interface to enable operators to effectively utilize the digital twin cockpit. It allows operators to view and analyze prediction results, adjust and optimize parameters, and observe the actual impact of these adjustments on the 3D digital model.

[0020] Another object of the present invention is to provide a computer device including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the aforementioned artificial intelligence-based digital cockpit method for tunnel boring machines.

[0021] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the aforementioned AI-based digital cockpit method for tunnel boring machines.

[0022] Another objective of this invention is to provide an information data processing terminal for implementing the aforementioned AI-based digital cockpit system for tunnel boring machines.

[0023] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0024] First, this invention proposes a method for creating a digital twin cockpit for tunnel boring machines (TBMs) based on machine learning and digital twin technologies. By collecting sensor data from the TBM, digitally simulating the TBM's operating scenarios, and using advanced machine learning models, it achieves accurate prediction of relevant parameters during the TBM's tunneling process and provides corresponding optimization guidance, thereby reducing the dangers of the construction process and improving tunneling efficiency.

[0025] Based on the case study, the main beneficial results are summarized as follows: (1) The proposed machine learning model can provide a reliable estimate of the tunneling parameters of the tunnel boring machine, and its R... 2RMSE and MAE were 0.955, 2.102 and 1.177, respectively. (2) Utilizing the robustness of the prediction model, the NSGA-III method showed a significant enhancement in the TBM excavation attitude under five different thrust conditions, with key parameters improved by 17.80%, 55.84% and 7.58%, respectively. (3) The machine learning method proposed in this invention was juxtaposed with various baseline methods and rigorously validated through a series of different evaluation metrics. The prediction accuracy of the CGAN model was improved by 1.9% to 13%. NSGA-III showed a 1 to 2-fold improvement in the spatial distribution of the optimization solution set. (4) The digital twin cockpit provides a visual representation of the shield machine operation information and optimization results, providing suggestions for the safety management and efficiency of tunneling.

[0026] Secondly, the integration of digital technology into tunnel boring machines (TBMs) can significantly improve construction efficiency and quality, providing accurate and real-time data for enhanced project management and decision-making. Digital twins, as a typical digital technology, offer advantages such as operational process visualization, real-time status monitoring, and future risk warnings. As a new application of digital twin technology in the TBM field, the TBM digital cockpit integrates the advantages of traditional TBM equipment and digital twin technology. The digital cockpit integrates data collection, processing, and analysis technologies, converging internal and external information into a single control interface. The essence of the digital cockpit is to develop the TBM into an intelligent system, improving efficiency and safety through real-time monitoring, information processing, and decision optimization. This technology involves digitally modeling and simulating various data samples and operational states of the actual equipment, ensuring a seamless connection between the virtual and physical domains. Through the digital cockpit, operators can access comprehensive data and status information, including speed, position, power, and mechanical conditions, enabling them to fully control and optimize driving decisions. In the long term, the combination of digital twin technology and machine learning makes remote monitoring and management of TBMs possible. Operators can remotely access real-time models and analysis results to understand the status of the tunnel boring machine and make necessary adjustments in a timely manner, without needing to be present on-site.

[0027] The main contribution of this invention is a novel intelligent control method that combines digital twins, machine learning, and the NSGA-III algorithm for the optimization of TBM excavation operations.

[0028] Second, the key research questions of the technical solution of this invention are: (1) how to solve the problems of low reliability of manual decision-making and untimely decision response in traditional shield tunneling construction? (2) how to improve the problems of insufficient real-time performance, inability to optimize and visualize online in real time in past artificial intelligence decision-making methods? (3) how to solve the problem of the inability to interact and control information bidirectionally between the operator and the TBM? To solve these problems, this study implements a digital cockpit architecture, which is divided into two parts: cockpit interface (front end) and data processing center (back end). The digital cockpit interface, based on digital twin technology, can receive and schedule data information in real time and display the TBM status. The data processing center integrates modeling, prediction and optimization algorithms, realizes parameter processing and provides optimal performance.

[0029] The innovations of this method are: (1) It proposes a comprehensive control method for tunnel boring machine (TBM) excavation, combining machine learning, genetic algorithms, and other methods to establish a comprehensive TBM construction control method covering data acquisition, prediction, and optimization decision-making throughout the entire construction cycle. (2) It designs a digital cockpit that effectively integrates relevant algorithms and provides real-time display of the TBM construction status and optimal decision-making capabilities. This invention changes the previous decision-making process that relied on human experience, overcomes the limitations of manual decision-making, and improves the intelligence level of TBM equipment. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the overall research framework of the research method provided in the embodiments of the present invention;

[0031] Figure 2 This is a schematic diagram of a three-dimensional digital model of the main components of a tunnel boring machine provided in an embodiment of the present invention;

[0032] Figure 3 This is a schematic diagram of the architecture of a complex neural network provided in an embodiment of the present invention;

[0033] Figure 4 This is a schematic diagram of the prediction results of the machine learning model provided in an embodiment of the present invention;

[0034] Figure 5 This is a schematic diagram of Pareto sets (total thrust / kN) under different operating conditions provided in the embodiments of the present invention; wherein, (a): 68007100; (b): 71007400; (c): 74007700; (d): 77008100; (e): 81007400;

[0035] Figure 6 This is a schematic diagram of the digital twin cockpit visualization interface provided in an embodiment of the present invention;

[0036] Figure 7This is a flowchart of the AI-based digital cockpit method for tunnel boring machines provided in this embodiment of the invention.

[0037] Figure 8 This is a structural diagram of the AI-based digital cockpit system for tunnel boring machines provided in an embodiment of the present invention;

[0038] Figure 9 This is a virtual demonstration diagram of shield machine parameter estimation provided in an embodiment of the present invention;

[0039] Figure 10 This is a schematic diagram of the error indices of different prediction methods provided in the embodiments of the present invention;

[0040] Figure 11 This is a schematic diagram of the loss function values ​​of different prediction methods provided in the embodiments of the present invention. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0042] To address the problems existing in the prior art, the present invention provides a method and system for a digital cockpit for tunnel boring machines based on artificial intelligence. The present invention will be described in detail below with reference to the accompanying drawings.

[0043] Reference Figure 1 As shown in the figure, this embodiment provides a digital cockpit method for tunnel boring machines based on artificial intelligence, with the example being a water diversion project in Northeast China. The main steps include:

[0044] Step 1: Description of the adaptive 3D digital model establishment method;

[0045] This invention utilizes a two-dimensional sketch associated with a solid model for parametric design. The three main components—the cutter head, the hydraulic cylinder, and the wafer loader—all use the cutter head diameter as the core parameter. Figure 2 The three-dimensional models of the main components of the tunnel boring machine are displayed. The dimensions of other components are all based on the diameter of the cutterhead, achieving the purpose of adaptive size matching.

[0046] When designing a model using SolidWorks software, the positional relationships between various detailed components must first be clearly defined and constrained. Then, solid models are generated through extrusion and cut operations. The operation parameters used in the extrusion and cut operations to form solid features are either fixed values ​​or related to the core parameter, the cutter head diameter. The ratio of other detailed component dimensions to the core parameter cutter head diameter is calculated, and then the proportional value is used as the driving logic. Using the software's built-in equation function, global variables are created by multiplying the proportional value by the core parameter. The dimensions of relevant components are then calibrated using the global variable values ​​or functions containing global variable values. Furthermore, when positional constraints exist, the model dimensions can be changed by modifying the core parameter values. Table 1 shows the parametric logic for some components.

[0047] Table 1. Parametric Formulas for Model Components

[0048]

[0049]

[0050] Step 2: Description of the tunnel boring machine excavation parameter prediction model;

[0051] Based on the dataset obtained from actual cases and the requirements of actual engineering, this invention selected target parameters such as shield tail deviation, main hydraulic tank temperature, and control oil circuit pressure. The decision variables selected in this study are shown in Table 2.

[0052] Table 2 Introduction to Tunneling Parameters

[0053]

[0054] In this case, a Conditional Generative Adversarial Neural Network (CGAN) structure combining multiple neural networks is used to obtain the mapping relationship between different parameters. The generator architecture defines two input layers: a vector containing 6 floating-point numbers and a noise vector containing 1 floating-point number, which are concatenated as the total input. The dimension of the random noise should be the same as the output dimension to maintain the correspondence between the noise and the generator output. The input vector is processed through deep neural network layers, each with 64 neurons containing an activation function and a kernel initializer. The output layer contains a neuron using a linear activation function with initialized weights of a random normal distribution. The input vector is concatenated to the output layer. An attention mechanism is responsible for dynamically adjusting the focus of attention. CNN networks and ResNet extract features to form the complete generator architecture. The discriminator architecture is similar, except that the last layer of the neural network used for the classification task uses a sigmoid activation function to output the confidence level of the classification, and the initialized weights are randomly uniformly distributed. The CGAN architecture uses the Adam optimizer to adjust different learning rates for different parameters to achieve convergence more efficiently. The activation function is "elu", the random seed is 1985, and the model loss function is binary cross-entropy. During iterative training, the discriminator network and generator network are trained alternately, with half of the training data being real and half being generated data, and the corresponding loss function values ​​are stored. The proposed complex neural network structure is as follows: Figure 3 As shown.

[0055] The CGAN model was trained on a dataset of 2400 samples, with 70% used as the training set and 30% as the test set. The dataset was derived from real-world engineering cases to ensure the robustness of the training process. Standard preprocessing steps were performed before training, including missing value interpolation and outlier removal. A five-fold crossover method was used to optimize the hyperparameters. Key hyperparameters included the learning rate, batch size, and number of iterations. Specific model parameter settings are shown in Table 5. The improved CGAN model achieved a prediction accuracy of 0.94. Figure 4 As shown, the R^2 values ​​of the mining parameter prediction model relative to the three training objectives are 0.9497, 0.9657 and 0.9511, respectively, indicating that it has good sample fitting ability.

[0056] Table 3 Machine Learning Model Parameter Information

[0057]

[0058] Step 3: Description of the tunnel boring machine decision optimization algorithm;

[0059] After establishing predictive models for decision variables and target parameters, it is necessary to optimize the operating parameters of the tunnel boring machine (TBM) to achieve better operating conditions. To differentiate the working states of the TBM under different thrusts, five construction conditions were divided into five intervals of 300 kN based on the total thrust range. A series of dynamic parameters were selected as decision variables, and optimization algorithms were used to optimize the parameters under different thrust conditions. During the optimization process, the value of each variable needs to conform to the actual engineering situation and be controlled within a reasonable range. Based on engineering experience, reasonable value ranges for TBM tunneling parameters are given, as shown in Table 4. The optimization algorithm requires an objective function; in this case, the objective function is set as the difference between the three target parameters and their ideal values.

[0060] Table 4 Range of tunneling parameters

[0061]

[0062] The optimization algorithm used in this case is the NSGA-III algorithm, which has strong adaptability and can effectively find a set of balanced non-dominated solutions in practical problems. For Pareto solutions, the ideal point is directly obtained through the TOPSIS method. Since there are five working conditions during the tunnel boring machine construction process, five optimization schemes will be generated, such as... Figure 5 As shown in Table 5, to illustrate the reliability of the optimization method, this paper compares the changes in the target parameters before and after optimization.

[0063] Table 5 Optimization rate under different working conditions

[0064]

[0065]

[0066] Step 4: Construction of the digital twin cockpit interface;

[0067] The digital twin cockpit framework proposed in this invention comprises four main components: (1) Visualization and interactive interface: The system's visualization interface presents real-time parameter values ​​and optimization suggestions during the excavation process. Furthermore, this interface establishes a mapping between the actual tunnel boring machine (TBM) and the virtual machine in the virtual space. After the TBM starts running, the virtual model displayed on the main interface will synchronize with the physical entity, displaying the action responses during the excavation process. (2) Tunneling parameter database: This database mainly stores data collected by various sensors during the TBM's operation. The other two components rely on this database to access relevant data. (3) Prediction and optimization algorithm: This component is the core of the digital twin cockpit, utilizing intelligent algorithms to predict and optimize excavation parameters. The optimization results are then transmitted to the visualization interface for display. Operators can adjust their excavation strategies based on the optimization results. (4) Detailed information: This component can be obtained in the detailed information section, where detailed information and data about the model can be observed. This information displays an overview of the main interface model information of the digital twin cockpit and marks information such as the TBM's outer diameter and total length. Figure 6 The visual interface of the digital twin cockpit was showcased.

[0068] The digital twin cockpit can reflect the real-time status of the tunnel boring machine (TBM), predict potential problems, and provide improvement measures. The 3D model provides a visual and interactive platform for the entire digital twin system. Users can observe the results of machine learning model predictions and the impact of parameter adjustments after NSGA-III optimization on the model. The 3D model can display the real-time status and operation of the TBM, and show the potential impact of the optimization process. The machine learning model, as the core prediction tool, is used to establish the mapping relationship between parameters. Through training, the machine learning model can learn the mapping relationship of the TBM under different operating conditions. This includes understanding how to adjust operating parameters under different conditions such as total thrust, cylinder pressure, and temperature. NSGA-III is used to adjust tunneling parameters within a multi-objective optimization framework. This algorithm finds the optimal trade-off solution among multiple objectives through the principle of genetic algorithms. To maintain the accuracy and practicality of the digital twin model, real-time data from sensors needs to be continuously integrated into the model. This includes data collected from sensors, as well as other relevant information generated during operation. The machine learning model can use this data for real-time predictions, and the NSGA-III algorithm can adjust the optimization strategy accordingly.

[0069] Data input is implemented using C# scripts. Sensors continuously transmit real-time values ​​of the mining parameters to the database. The system reads the database via the script and displays the real-time data in the corresponding boxes. Then, the data enters the intelligent algorithm, where optimization suggestions are obtained through the algorithm interface, and the adjusted mining parameter values ​​are stored in the database. The system then retrieves the optimized mining parameter values ​​from the database again.

[0070] The tunnel boring machine (TBM) excavation process is simulated using animation components in Unity. These components set the actions of the model components, demonstrating actions such as the rotation of the cutterhead for excavation, the hydraulic cylinder-driven movement of the tunnel segments, and the rotation of the segment assembler in the desired direction after a new segment is acquired. By continuing the hydraulic cylinder-driven movement, the tunnel segments are installed in the designated positions, completing the tunnel excavation process.

[0071] like Figure 10 As shown, the AI-based digital cockpit method for tunnel boring machines includes:

[0072] Adaptive 3D digital model establishment: Parametric modeling is adopted, enabling the model to adaptively match shield tunneling equipment of any size;

[0073] Tunnel boring machine tunneling parameter prediction model: Based on conditional generative adversarial neural network, and combined with various neural networks (such as deep neural network, convolutional neural network and residual neural network) for different data types, a prediction model between relevant tunneling parameters was established.

[0074] Shield tunneling machine decision optimization algorithm: Based on the NSGA-III multi-objective optimization algorithm, the optimal decision solution is provided during the tunneling process;

[0075] Digital Twin Cockpit Interface Construction: To enable operators to effectively utilize the digital twin cockpit, an intuitive and user-friendly interface is provided. This interface allows operators to view and analyze prediction results, adjust and optimize parameters, and observe the actual impact of these adjustments on the 3D digital model.

[0076] S1 specifically comprises three main systems based on the functions of the tunnel boring machine (TBM) components: excavation system, support system, and earthwork removal system. Modeling utilizes Solidworks software, focusing on the main components of these three systems: the cutterhead model, the hydraulic cylinder model, and the segment assembly machine model. During parametric modeling, constraints and dependencies between component features are established according to the design intent, ensuring that every modification to the core parameters results in model changes consistent with predefined constraints and specifications. All three main component models use the cutterhead diameter as the core parameter, linking component dimensions to the cutterhead diameter through reasonable quantitative relationships. Quantitative relationships are used as the modeling logic, with built-in equation functionality used to create global variables based on the core parameters. These global variable values, or functions containing global variable values, are then used to calibrate component dimensions. Therefore, the dimensions of all model components are represented as equations related to the core parameters, allowing for the modification of only the core parameter values ​​to obtain the corresponding digital model based on the actual dimensions of the TBM, achieving adaptive model matching.

[0077] S2 specifically includes: In order to realistically reflect the tunnel boring machine excavation scenario, it is necessary to build machine learning models of actual operating parameters and decision parameters, and embed them into the digital twin cockpit.

[0078] S3 specifically includes: After establishing a prediction model between tunneling parameters, the optimization algorithm can optimize the decision parameters and integrate the optimization results into the cockpit. The specific optimization target value is given based on engineering experience.

[0079] In S4, the interface consists of two main parts: the main interface and sub-interfaces. The main interface displays the dynamic tunneling posture and some basic information of the tunnel boring machine during construction. At the bottom of the main interface are six input buttons for the sub-interfaces. The sub-interfaces handle functions such as data analysis, displaying optimization results, and querying logs. The cockpit interface is created using the Unity engine.

[0080] like Figure 11 As shown, the AI-based digital cockpit system for tunnel boring machines provided in this embodiment of the invention includes:

[0081] Adaptive 3D Digital Model Building Module: Used for parametric modeling, enabling the model to adaptively match any size tunnel boring machine;

[0082] The tunnel boring machine excavation parameter prediction model establishment module is used to establish a prediction model between relevant tunneling parameters based on conditional generative adversarial neural networks and various neural networks (such as deep neural networks, convolutional neural networks, and residual neural networks) for different data types.

[0083] Tunnel Boring Machine Decision Optimization Module: Used to provide the optimal decision solution during tunneling based on the NSGAIII multi-objective optimization algorithm;

[0084] Digital Twin Cockpit Interface: This interface provides an intuitive and user-friendly interface to enable operators to effectively utilize the digital twin cockpit. It allows operators to view and analyze prediction results, adjust and optimize parameters, and observe the actual impact of these adjustments on the 3D digital model.

[0085] An application embodiment of the present invention provides a computer device, which includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of an artificial intelligence-based digital cockpit method for tunnel boring machines.

[0086] An application embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of an artificial intelligence-based digital cockpit method for tunnel boring machines.

[0087] An application embodiment of the present invention provides an information data processing terminal, which is used to implement an artificial intelligence-based digital cockpit system for tunnel boring machines.

[0088] like Figure 9 The diagram shown is a virtual demonstration of shield machine parameter estimation provided in an embodiment of the present invention. This technical solution relies on a shield machine for implementation. First, a three-dimensional digital model is established based on the actual dimensions of the shield machine. Then, the integrated digital decision module performs real-time parameter prediction and optimization based on data collected by the shield machine's body sensors. Figure 10 The diagram illustrates the error indices of different prediction methods provided in this embodiment of the invention. Prediction and optimization results are displayed on the interactive interface. As the project progresses, the system can continuously learn and optimize its operational strategies to adapt to the ever-changing engineering environment and requirements. Figure 11 The diagram shows the loss function values ​​of different prediction methods provided in the embodiments of the present invention. The operator's control effect on the tunnel boring machine will also be reflected in the three-dimensional digital model.

[0089] Table 6 Error Indicators for Different Prediction Methods

[0090]

[0091] (1) The improved CGAN model achieves a prediction accuracy of 0.94. For example... Figure 4 As shown, the R² values ​​of the parameter prediction model relative to the three training objectives are 0.9497, 0.9657, and 0.9511, indicating good sample fitting ability. The model generated root mean square errors of 4.021, 1.073, and 1.211 for the three prediction objectives, while the mean absolute errors were as low as 2.783, 0.287, and 0.355, respectively. Although there are some differences in the magnitude of these results, all are small compared to the scale of the original data. Overall, both the training and test sets produced better prediction results, indicating that the model did not suffer from overfitting or underfitting. Furthermore, due to the adversarial training mechanism in CGAN, it typically generates more resilient models, especially when dealing with noisy data and outliers. In the digital twin cockpit, the predicted values ​​and the actual recorded historical data are shown... Figure 9 As shown. In summary, the improved CGAN offers higher efficiency and lower computational cost, making it suitable for engineering scenarios and subsequent parameter prediction.

[0092] (2) The NSGA-III algorithm achieved a multi-objective optimization average rate of 27% under a total thrust of 6800-8300, demonstrating strong effectiveness and robustness. (Refer to the research...) Figure 5Table 5 clearly illustrates the improvements achieved by these objectives. The most significant improvement is in the temperature regulation of the main hydraulic cylinder. The average optimization rates for the three objectives are 17.80%, 55.84%, and 7.58%, respectively. These results not only highlight the effectiveness of the NSGA-III algorithm in improving specific operating parameters but also its overall impact on the efficiency and durability of the TBM system. NSGA-III demonstrates consistent performance improvements across different thrust scenarios, proving its versatility and adaptability. Table 5 reinforces this by showing the consistent effectiveness of the algorithm in enhancing all three objective parameters under different conditions. The maximum optimization rates for these objectives are impressive, at 31.19%, 71.24%, and 10.34%, respectively. These results not only highlight the algorithm's high adaptability but also its potential to significantly extend TBM service life and reduce maintenance requirements. Overall, NSGA-III can better mitigate adverse effects such as wear and overload in TBM operation under various scenarios.

[0093] (3) Compared to baseline machine learning methods, the improved CGAN achieves an accuracy improvement of 1.9% to 13%. The improved CGAN exhibits better predictive performance and faster convergence efficiency. Figure 10 As shown in Table 6, the R values ​​of the improved CGAN, CGANCNN, BP, RF, ANN, and SVR are... 2 The scores were 0.955, 0.84, 0.890, 0.935, 0.937, and 0.937, respectively. Compared with other methods, the improved CGAN achieved prediction accuracy improvements of 7.3%, 13%, 2.1%, 1.9%, and 1.9%, respectively. Other accuracy regression metrics such as RMSE and MAE also showed similar performance. Figure 11 As shown, the improved CGAN, CGANCNN, SVR, BP, and ANN converge in the 9th, 10th, 22nd, 19th, and 31st generations, respectively, indicating that the improved CGAN can obtain results quickly. Meanwhile, from... Figure 6 As can be seen, the improved CGAN convergence curve encloses the smallest region, indicating that it can achieve good results in scenarios with fewer training iterations. When the number of iterations is 20, the loss values ​​of the improved CGAN, SVR, BP, and ANN are 0.013, 0.038, 0.018, and 0.057, respectively. This shows that the improved CGAN has higher prediction accuracy. The results demonstrate that this method outperforms other machine learning algorithms in all aspects of this study.

[0094] (4) The intelligent decision-making module of the digital twin cockpit provides reliable guidance for the control of mining parameters during TBM operation. As a demonstration of the digital twin cockpit, the digital twin driving model is able to suggest appropriate control values ​​(x1 to x6) for mining parameters using the NSGA-III algorithm. The machine learning model with high prediction accuracy predicts the performance changes of the TBM based on the suggested values. Compared with the original values, the performance estimated by the digital twin cockpit is generally improved in all three objectives (y1 to y3). The digital twin cockpit achieves virtual-to-physical control by creating a virtual model to simulate the actual operation of the TBM. This control can be achieved through routine adjustment or remote control. In routine adjustment, the TBM operator can manually adjust key operating parameters according to the suggested values ​​provided by the digital twin model to optimize the performance of the TBM. On the other hand, remote control can directly transmit the operating parameters provided by the digital twin model to the TBM through signal transmission to achieve real-time adjustment and optimization. Therefore, actual TBM operation can be guided by a digital cockpit with intelligent algorithms.

[0095] This invention provides a parametric modeling and core parameter application method: In the process of modeling tunnel boring machine components, the cutterhead diameter is set as the core parameter, and the dimensions of all key component models—the cutterhead model, hydraulic cylinder model, and segment assembly machine model—are defined through their quantitative relationship with the cutterhead diameter. This parametric modeling method allows the model to adaptively adjust according to changes in the core parameter, thereby achieving a highly flexible and adaptable design scheme.

[0096] This invention provides a method for constructing constraint and dependency relationships between component features: by establishing constraint and dependency relationships between component features during the design phase, it ensures that each modification of the model is consistent with predefined constraints and specifications, thereby increasing the accuracy and reliability of the design.

[0097] This invention provides a built-in equation function and global variable application**: using the built-in equation function of Solidworks software to create global variables based on core parameters, and using these global variables or functions containing global variable values ​​to calibrate part dimensions. This not only optimizes the design process but also enhances the modifiability and scalability of the model, enabling designers to quickly and accurately adjust model dimensions according to actual needs.

[0098] This invention provides an adaptive model matching mechanism: the aforementioned innovations collectively achieve an adaptive model matching mechanism, which allows for the acquisition of a corresponding digital model based on the actual dimensions of the tunnel boring machine (TBM) by modifying only one core parameter—the cutterhead diameter. This adaptive matching significantly improves the efficiency of model design, providing greater flexibility and the ability to quickly respond to actual changes in TBM design.

[0099] These innovations break through the traditional modeling methods for tunnel boring machine components. By introducing parametric design and adaptive matching mechanisms, they significantly improve the efficiency and adaptability of the design, while also providing a reference for the design of similar complex mechanical equipment.

[0100] Application Example 1: Urban Underground Tunnel Engineering

[0101] In urban underground tunnel projects, tunnel boring machines (TBMs) are crucial construction equipment. Using AI-based digital cockpits for TBMs can significantly improve construction efficiency and safety.

[0102] S1: Adaptive 3D Digital Model Establishment

[0103] We create customized 3D digital models of tunnel boring machines (TBMs) for urban underground tunnel projects, automatically adjusting model parameters according to tunnel dimensions to ensure the model accurately reflects the structure and function of the actual TBM.

[0104] S2: Prediction Model for Tunnel Boring Machine Excavation Parameters

[0105] Historical geological and construction data are collected as inputs, and a condition-generating adversarial network is used to predict the geological conditions that may be encountered during tunneling, such as soil thickness, rock hardness, and key parameters such as tunneling speed and torque.

[0106] S3: Shield Machine Decision Optimization Algorithm

[0107] Based on the predicted tunneling parameters and actual construction objectives (such as cost, time, safety, etc.), the NSGA-III algorithm is used to calculate the optimal tunneling strategy, including the speed and thrust of the tunnel boring machine.

[0108] S4: Construction of Digital Twin Cockpit Interaction Interface

[0109] Develop an interactive interface that allows engineers and operators to monitor the tunneling status in real time, view the predicted results of tunneling parameters, make decision adjustments, and intuitively observe the impact of these adjustments on the tunnel boring machine and the tunnel project.

[0110] Application Example 2: Foundation Construction of a Cross-Sea Bridge

[0111] When carrying out the foundation construction project of the cross-sea bridge, the tunnel boring machine needs to operate under complex and ever-changing seabed geological conditions, which places higher demands on the control and adjustment of the tunnel boring machine.

[0112] S1: Adaptive 3D Digital Model Establishment

[0113] For specific seabed geological structures, an adaptive three-dimensional digital model of the tunnel boring machine is established, and the model parameters are automatically adjusted according to different construction sections to match the actual situation.

[0114] S2: Prediction Model for Tunnel Boring Machine Excavation Parameters

[0115] By combining seabed geological exploration data and construction data from similar previous projects, a conditional generative adversarial network is used to simulate tunneling parameters under different seabed geological conditions, such as wear, torque, and tunneling speed.

[0116] S3: Shield Machine Decision Optimization Algorithm

[0117] By utilizing the NSGA-III algorithm to comprehensively consider multiple objectives such as construction efficiency, equipment wear, and safety, an optimal shield tunneling machine operation strategy is proposed to ensure both high efficiency and safety during the construction process.

[0118] S4: Construction of Digital Twin Cockpit Interaction Interface

[0119] Design an interactive digital twin cockpit interface that allows operators to adjust the tunneling strategy of the tunnel boring machine based on real-time geological data and prediction results. At the same time, the 3D model displayed on the interface can help operators intuitively understand the impact of tunneling strategy adjustments on the construction progress.

[0120] These two application examples demonstrate the practical application of the AI-based digital cockpit method for tunnel boring machines in various complex engineering projects, showcasing the significant advantages of this technical solution in improving construction efficiency, ensuring construction safety, and optimizing the decision-making process.

[0121] It should be noted that embodiments of the present invention can be implemented using hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented using hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or using software executed by various types of processors, or using a combination of the above-described hardware circuitry and software, such as firmware.

[0122] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An artificial intelligence based method for a digital cockpit of a tunneling machine, characterized in that, Comprise: S1, adaptive three-dimensional digital model establishment: adopt parameterized modeling, so that the model can adaptively match any size of shield equipment; S2, shield tunneling parameter prediction model: based on conditional generative adversarial neural network, for data type, combined with multiple neural networks, a prediction model between related tunneling parameters is established; S3, shield machine decision optimization algorithm: based on NSGA-III multi-objective optimization algorithm, the optimal decision solution is given in the process of tunneling; S4, digital twin cockpit interactive interface construction: in order to enable operators to effectively use the digital twin cockpit, the interactive interface of the digital twin cockpit is provided; the interface allows operators to view and analyze prediction results, adjust optimization parameters, and observe the actual impact of these adjustments on the three-dimensional digital model; S1 specifically comprises: Adopting parameterized modeling technology, setting the cutterhead diameter as the core parameter, and defining the size of the cutterhead model, hydraulic cylinder model and segment assembly machine model according to the core parameter through quantitative relationship, so that the model can adaptively adjust according to the change of the core parameter; In the design stage, the constraint relationship and dependency relationship between the features of the parts are constructed to ensure that the model modification is consistent with the predefined constraints and specifications; Using the built-in equation function of Solidworks software, global variables are created according to the core parameters, and these global variables or functions containing global variable values are used to calibrate the size of the parts; Adaptive model matching mechanism is realized, and the corresponding digital model can be obtained based on the actual size of the shield machine by modifying only the core parameter cutterhead diameter.

2. The artificial intelligence-based tunneling machine digital cockpit method of claim 1, wherein, S2 specifically comprises: Build machine learning model of actual operation parameters and decision parameters, and embed it into digital twin cockpit.

3. The artificial intelligence-based tunneling machine digital cockpit method of claim 1, wherein, S3 specifically comprises: after establishing the prediction model between the tunneling parameters, the optimization algorithm optimizes the decision parameters, and integrates the optimization results into the cockpit, and the specific optimization target value is given according to engineering experience.

4. The artificial intelligence-based tunneling machine digital cockpit method of claim 1, wherein, In S4, the interface is composed of two parts: main interface and sub-interface; the main interface background displays the dynamic tunneling posture and some basic information in the shield construction process, and there are six sub-interface input buttons at the bottom of the main interface; the sub-interface undertakes data analysis, optimization result display and log query function; the cockpit interactive interface is created using Unity engine.

5. An artificial intelligence-based digital cockpit system for a tunneling machine, which implements the artificial intelligence-based digital cockpit method according to any one of claims 1 to 4, characterized in that, Comprise: Adaptive three-dimensional digital model establishment module: for adopting parameterized modeling, so that the model can adaptively match any size of shield equipment; Shield tunneling parameter prediction model establishment module: for based on conditional generative adversarial neural network, for data type, combined with multiple neural networks, a prediction model between related tunneling parameters is established; Shield machine decision optimization module: for based on NSGAIII multi-objective optimization algorithm, the optimal decision solution is given in the process of tunneling; Digital twin cockpit interactive interface: for in order to enable operators to effectively use the digital twin cockpit, the interactive interface of the digital twin cockpit is provided; the interface allows operators to view and analyze prediction results, adjust optimization parameters, and observe the actual impact of these adjustments on the three-dimensional digital model. 6.A computer device, comprising a memory and a processor, the memory storing a computer program, the computer program being executed by the processor to cause the processor to perform the steps of the method for a digital cockpit of a shield tunneling machine based on artificial intelligence according to any one of claims 1 to 4. 7.A computer readable storage medium storing a computer program, the computer program being executed by a processor to cause the processor to perform the steps of the method for a digital cockpit of a shield tunneling machine based on artificial intelligence according to any one of claims 1 to 4. 8.An information data processing terminal for implementing the system for a digital cockpit of a shield tunneling machine based on artificial intelligence according to claim 5.

Images