TBM tunneling control parameter intelligent optimization decision method based on deep transfer learning

Abstract

The application discloses a TBM tunneling control parameter intelligent optimization decision method based on deep migration learning. The method comprises the following steps: acquiring TBM multi-source heterogeneous data to construct a TBM data set; training a TBM tunneling load parameter real-time prediction model based on the TBM data set to obtain a trained TBM tunneling load parameter prediction model; constructing a tunneling benefit function; obtaining optimal tunneling control parameters based on the tunneling benefit function and the tunneling load parameter real-time prediction model; adding the optimal tunneling control parameters into the TBM data set for updating, training a tunneling control parameter real-time optimization decision model to obtain a trained tunneling control parameter real-time optimization decision model; acquiring TBM multi-source heterogeneous data to be detected, and identifying the TBM multi-source heterogeneous data to be detected based on the tunneling control parameter real-time optimization decision model to obtain optimal tunneling control parameters. The application can avoid a complex optimization process in the tunneling process, and has the advantages of good real-time performance, high accuracy and the like.

Description

Technical Field

[0001] This application relates to the field of underground engineering TBM tunneling technology, and in particular to a TBM tunneling control parameter intelligent optimization decision-making method, system, storage medium and electronic equipment based on deep transfer learning. Background Technology

[0002] Tunnel Boring Machines (TBMs) are widely used in the construction of deep and long tunnels due to their advantages such as high tunneling efficiency, good tunnel quality, minimal disturbance to the surrounding rock, and environmental friendliness. Unlike drill-and-blast methods, the tunneling cost and efficiency of TBMs are more sensitive to geological formations and are easily affected by adverse geological conditions. Therefore, during the tunneling process, TBMs need to continuously adjust their tunneling control parameters (i.e., the operating parameters selected by the operator, such as cutterhead speed N and advance speed AR) according to changes in geological conditions to keep them within a reasonable range.

[0003] In actual tunneling, TBMs mostly rely on the driver's subjective experience, resulting in poor matching between tunneling parameters and geological conditions, and even potential decision-making errors. This leads to problems such as low tunneling efficiency, high energy consumption, and severe cutterhead wear. Furthermore, the various TBM optimization decision-making methods and systems proposed to date require complex optimization processes or do not fully utilize rock-machine feedback information in practical applications, making it difficult to simultaneously ensure the timeliness and accuracy of optimization decisions. Summary of the Invention

[0004] This application provides a method, system, storage medium, and electronic device for intelligent optimization decision-making of TBM tunneling control parameters based on deep transfer learning. It can avoid the complex optimization process during tunneling and has the advantages of good real-time performance and high accuracy.

[0005] This application provides an intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning, including: Acquire TBM multi-source heterogeneous data and construct a TBM dataset; The real-time prediction model for TBM tunneling load parameters was trained based on the TBM dataset to obtain the trained TBM tunneling load parameter prediction model. Construct the tunneling benefit function; Based on the aforementioned tunneling benefit function and real-time prediction model for tunneling load parameters, the tunneling control parameters for each tunneling cycle are optimized to obtain the optimal tunneling control parameters. The optimal tunneling control parameters are added to the TBM dataset to obtain an updated TBM dataset. The real-time optimization decision model for tunneling control parameters is trained based on the updated TBM dataset to obtain a trained real-time optimization decision model for tunneling control parameters with the optimal control parameters as the output. Acquire multi-source heterogeneous data of the TBM to be detected, and use the real-time optimization decision model based on the tunneling control parameters to identify the multi-source heterogeneous data of the TBM to be detected, so as to obtain the optimal tunneling control parameters.

[0006] Furthermore, in the aforementioned intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning, the multi-source heterogeneous data includes TBM mechanical parameters, electro-hydraulic parameter data, surrounding rock data obtained from geological exploration, cutterhead vibration signals, and rock debris images.

[0007] Furthermore, the aforementioned intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning requires preprocessing of the TBM dataset after its construction: Non-working data in the TBM mechanical parameters and electro-hydraulic parameters are removed to form TBM tunneling parameter data in units of cycle segments; Identify data anomalies and frame drops in the TBM tunneling parameters, correct locally weakly anomalous data, and remove globally strong anomalous data; Each cycle segment is divided into an empty push segment, an ascending segment, a stable segment, and a descending segment to obtain preprocessed historical cycle segment data; The TBM multi-source heterogeneous data other than the mechanical parameters and electro-hydraulic parameters are subjected to data noise reduction, missing and abnormal data supplementation, and invalid data removal.

[0008] Furthermore, the above-mentioned intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning includes a real-time prediction model for TBM tunneling load parameters comprising a multi-branch input deep neural network for processing different types of TBM multi-source heterogeneous data, including an input module, a multi-branch feature extraction module, a feature fusion module, a multi-branch prediction module, and an output module. The processing steps of the real-time prediction model for tunneling load parameters include: The preprocessed TBM multi-source heterogeneous data is divided into three categories: real-time updated ascending segment data, historical cycle segment data, and tunneling control parameters, and then input into the input module respectively. The multi-branch feature extraction module extracts deep features of TBM multi-source heterogeneous data in different forms. The feature fusion module fuses the deep features of each branch to obtain fused features. The multi-branch prediction module predicts the fused features to obtain the predicted tunneling load parameters, which are then output through the output module.

[0009] Furthermore, in the above-mentioned intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning, the tunneling benefit function is:

[0010]

[0011] in, F , T , N , AR These are the average values ​​of total propulsion force, cutterhead torque, cutterhead rotation speed, and propulsion speed during the stabilization phase. For the tunneling benefit function, SE S represents the tunneling specific energy, which is the energy consumption of a TBM excavating a unit volume of rock, and S represents the cross-sectional area of ​​the tunnel face. CW The cutterhead wear coefficient represents the amount of wear on the cutterhead per unit volume of rock during TBM tunneling. , , The weighting coefficient is between 0 and 1.

[0012] Furthermore, in the above-mentioned intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning, the multi-branch feature extraction module can be a temporal convolutional neural network, a long short-term memory network, or a fully connected neural network, and the feature fusion module and the multi-branch prediction module are fully connected neural networks.

[0013] Furthermore, the above-mentioned intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning, wherein, The training of the real-time optimization decision model for the tunneling control parameters based on the updated TBM dataset includes: The real-time prediction model of TBM tunneling load parameters used to predict TBM tunneling load parameters is defined as the source model, and the real-time optimization decision model of tunneling control parameters is defined as the target model. The structure and weights of the multi-branch feature extraction modules of the source model used to process the real-time updated rising segment data and the multi-branch feature extraction modules used to process the historical cyclic segment data are transferred to the target model. The structures of the feature fusion module and multi-branch prediction module of the source model are transferred to the target model, and the weights are randomly initialized. The network weights of the multi-branch feature extraction module and the feature fusion module are fine-tuned using L2 regularization, and the branch prediction module is retrained.

[0014] This application also provides an intelligent optimization decision-making system for TBM tunneling control parameters based on deep transfer learning, including: The acquisition module is used to acquire TBM multi-source heterogeneous data and construct a TBM dataset; the TBM dataset includes TBM multi-source heterogeneous data and tunneling load parameter data. The TBM tunneling load parameter prediction module trains the real-time prediction model of TBM tunneling load parameters based on the TBM dataset, and obtains the trained TBM tunneling load parameter prediction model. The tunneling benefit function construction module is used to construct the tunneling benefit function; The tunneling control parameter optimization module is used to optimize the tunneling control parameters of each tunneling cycle segment based on the tunneling benefit function and the real-time prediction model of tunneling load parameters, so as to obtain the optimal tunneling control parameters. The training module is used to add the optimal tunneling control parameters into the TBM dataset to obtain an updated TBM dataset, and to train the real-time optimization decision model of tunneling control parameters based on the updated TBM dataset to obtain a trained real-time optimization decision model of tunneling control parameters with the optimal control parameters as the output. The tunneling control parameter optimization decision module is used to acquire multi-source heterogeneous data of the TBM to be detected, and to identify the multi-source heterogeneous data of the TBM to be detected based on the real-time optimization decision model of the tunneling control parameters, so as to obtain the optimal tunneling control parameters.

[0015] This application also provides a computer-readable storage medium storing multiple instructions adapted for loading by a processor to execute any of the above-described intelligent optimization decision-making methods for TBM tunneling control parameters based on deep transfer learning.

[0016] This application also provides an electronic device, including a processor and a memory, wherein the processor is electrically connected to the memory, the memory is used to store instructions and data, and the processor is used in the steps of the intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning described in any of the above claims.

[0017] This application provides a TBM tunneling control parameter intelligent optimization decision-making method, system, storage medium, and electronic device based on deep transfer learning. This application constructs a real-time prediction model for TBM tunneling load parameters based on collected multi-source heterogeneous data, mines deep features in the data, constructs a tunneling benefit function, and finds the optimal tunneling operation parameters for each cycle segment based on the real-time prediction model and optimization algorithm, which serves as a supplement to the dataset. Based on a pre-trained and fine-tuned deep transfer learning strategy and the supplemented dataset, an intelligent optimization decision-making model for TBM tunneling control parameters is constructed. This model can provide the optimal tunneling operation parameters in real time during the trial tunneling phase, with the results directly predicted by the model, avoiding the complex optimization process during tunneling. It has advantages such as good real-time performance and high accuracy, ensuring efficient, safe, and low-cost TBM tunneling, and improving the intelligence level of TBM tunneling control, even achieving complete automation of tunneling control. Attached Figure Description

[0018] The technical solution and other beneficial effects of this application will become apparent from the following detailed description of specific embodiments in conjunction with the accompanying drawings.

[0019] Figure 1 A flowchart illustrating the intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning, provided in this application embodiment.

[0020] Figure 2 This is a schematic diagram of the structure of the real-time prediction model for tunneling load parameters provided in the embodiments of this application.

[0021] Figure 3 The TBM tunneling load parameter prediction results are provided for the embodiments of this application.

[0022] Figure 4 This is a schematic diagram showing the average tunneling efficiency function value and the maximum tunneling efficiency function value of all individuals in each generation during the optimization process provided in the embodiments of this application.

[0023] Figure 5 The diagram provided for the embodiments of this application shows the final optimization result.

[0024] Figure 6 This is a diagram illustrating the architecture of an intelligent optimization decision-making model for TBM tunneling control parameters based on deep transfer learning, as provided in an embodiment of this application.

[0025] Figure 7 This is a schematic diagram of the structure of the intelligent optimization decision-making system for TBM tunneling control parameters based on deep transfer learning, provided in an embodiment of this application.

[0026] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] This application provides a method, system, storage medium, and electronic device for intelligent optimization decision-making of TBM tunneling control parameters based on deep transfer learning. The intelligent optimization decision-making system for TBM tunneling control parameters provided in this application can be integrated into an electronic device, such as a terminal or server. The terminal can include a tablet computer, laptop computer, personal computer (PC), microprocessor box, or other devices.

[0029] Please see Figure 1 , Figure 1 The flowchart illustrates a TBM tunneling control parameter intelligent optimization decision-making method based on deep transfer learning, provided in this application embodiment. This method, applied in electronic devices, includes the following steps: S1. Obtain TBM multi-source heterogeneous data and construct the TBM dataset.

[0030] Specifically, the multi-source heterogeneous data includes TBM tunneling machinery and electro-hydraulic parameter data, and includes, but is not limited to, at least one of the following: surrounding rock data obtained from geological surveys, cutterhead vibration signals, rock debris images, radar imaging data, surrounding rock displacement data, microseismic monitoring data, and other TBM tunnel construction monitoring data.

[0031] Furthermore, after constructing the TBM dataset, it is necessary to preprocess the TBM dataset: S11, remove non-working data from the TBM mechanical and electro-hydraulic parameters to form TBM tunneling parameter data in units of cycle segments.

[0032]

[0033]

[0034] in, , , , The first i The total thrust, cutterhead torque, cutterhead rotation speed, and feed speed at any given moment. express i The working status of the TBM at all times, if If the value is 0, the TBM is considered to be in a shutdown state, and the entire data at that moment is discarded as non-working data.

[0035] S12 identifies data anomalies and frame drops in the TBM tunneling parameters, corrects locally weakly anomalous data, and removes globally strongly anomalous data.

[0036] S13, divide each cycle segment into an empty push segment, an ascending segment, a stable segment, and a descending segment to obtain preprocessed historical cycle segment data.

[0037] S14 performs data noise reduction, missing and abnormal data supplementation, and invalid data removal operations on TBM multi-source heterogeneous data other than TBM mechanical parameters and electro-hydraulic parameters.

[0038] S2, Based on the TBM dataset, a real-time prediction model for TBM tunneling load parameters is trained to obtain a trained TBM tunneling load parameter prediction model.

[0039] In one embodiment, the real-time prediction model for TBM tunneling load parameters includes a multi-branch input deep neural network for processing different types of TBM multi-source heterogeneous data, including an input module, a multi-branch feature extraction module, a feature fusion module, a multi-branch prediction module, and an output module. The processing steps of the real-time prediction model for tunneling load parameters include: S21, the preprocessed TBM multi-source heterogeneous data is divided into three categories: real-time updated rising section data, historical cycle section data, and tunneling control parameters, and then input into the input module respectively; S22, deep features of TBM multi-source heterogeneous data in different forms are extracted through the multi-branch feature extraction module; S23, The deep features of each branch are fused through the feature fusion module to obtain fused features; S24 uses a multi-branch prediction module to predict the fused features, obtains the predicted tunneling load parameters, and outputs them through the output module.

[0040] In one embodiment, the multi-branch feature extraction module can be a temporal convolutional neural network, a long short-term memory network, or a fully connected neural network, and the feature fusion module and the multi-branch prediction module can be fully connected neural networks.

[0041] Specifically, the architecture of the multi-branch feature extraction module in a multi-branch input deep neural network depends on the data format of each input branch. For time-series data (such as TBM tunneling parameter data in the ascending section, historical TBM tunneling parameter data in the circulating section, surrounding rock displacement data, etc.), a neural network structure capable of processing time-series data is adopted, including but not limited to recurrent neural networks (RNNs), convolutional neural networks, temporal convolutional neural networks, Transformer networks, and their variants and combinations. For non-time-series data (such as the average advance speed in the stable section, cutterhead wear, etc.), a fully connected neural network is adopted. For image data or field data (such as tunnel image data, radar imaging data, etc.), a convolutional neural network or its variants are adopted.

[0042] For a given set of multi-source heterogeneous data, multiple branches of input data can be generated. Different multi-source data can also be merged according to their temporal, spatial, and data structure similarities to form the same branch of input data. The multi-branch feature extraction module includes, but is not limited to, at least one of the following: ascending segment TBM parameter data, historical multi-cycle segment TBM parameter data and rock mass information data, current radar imaging data, and other TBM tunnel construction monitoring and acquisition data.

[0043] Specifically in this example, Figure 2 This is a schematic diagram of the structure of the real-time prediction model for tunneling load parameters provided in the embodiments of this application, as shown below. Figure 2 As shown, the model inputs include predicted cycle loading segment data, historical cycle data (including TBM tunneling parameter data and rock mass information data), and tunneling control parameters (total thrust and cutterhead torque). The output is the tunneling load parameters (total thrust and cutterhead torque). The input and output characteristics of each branch of the model are shown in Table 1.

[0044] Table 1. Model Input and Output Features

[0045] Each branch of the multi-branch feature extraction module employs a Temporal Convolutional Neural Network (TCN), a Long Short-Term Memory Network (LSTM), and a Fully Connected Neural Network (FCNN), respectively. The remaining modules all use FCNN neural networks. The specific network structures are as follows: Figure 2 The parameters of each neural network layer are shown in Table 2. The hyperparameters for model training were determined using grid search, and the specific parameters are shown in Table 3.

[0046] Table 2 Parameters of each neural network layer in the model

[0047] Table 3. Optimization of Hyperparameters for Neural Network Training

[0048] During training, 80% of the data is randomly selected as the training set, 10% as the validation set, and 10% as the test set. Figure 3 The TBM tunneling load parameter prediction results are provided for the embodiments of this application.

[0049] S3, construct the tunneling benefit function.

[0050] The tunneling benefit function, constructed by comprehensively considering safety, cost, and efficiency, is as follows:

[0051]

[0052] in, F, T , N , AR These are the average values ​​of total propulsion force, cutterhead torque, cutterhead rotation speed, and propulsion speed during the stabilization phase. For the tunneling benefit function, SE S represents the tunneling specific energy, which is the energy consumption of a TBM excavating a unit volume of rock, and S represents the cross-sectional area of ​​the tunnel face. CW The cutterhead wear coefficient represents the amount of wear on the cutterhead per unit volume of rock during TBM tunneling. , , The weighting coefficient is between 0 and 1.

[0053] Specifically in this example, α 1. α 2. α All three values ​​are taken as 1 / 3, and the feasible range for propulsion speed and cutterhead torque is taken as ±20% of the actual operating parameters during tunneling. F , T , N , AR All values ​​do not exceed the machine's rated parameters. Specifically, in this example, CW is taken in the following form (in this example...). Take 2.5 × 10 -6 ):

[0054] S4. Based on the tunneling benefit function and the real-time prediction model of tunneling load parameters, the tunneling control parameters of each tunneling cycle are optimized to obtain the optimal tunneling control parameters.

[0055] Based on a real-time prediction model of tunneling load parameters, an optimization algorithm (such as a genetic algorithm) is used to optimize the tunneling control parameters for each tunneling cycle. The optimal advance speed and optimal cutterhead speed obtained from the optimization of each cycle are recorded and expanded into the dataset. The optimization algorithm selects tunneling control parameter data (N, AR) within the feasible region using a certain strategy. Based on the real-time prediction model of tunneling load parameters (N, AR), it predicts the tunneling load parameters (F, T) under these tunneling control parameters (N, AR), thereby obtaining the tunneling benefit function value. Through multiple searches using a certain strategy, the optimal tunneling control parameters are found.

[0056] In this specific example, the optimization algorithm uses a genetic algorithm. The genetic algorithm has a population size of 600, a generation count of 100, a mutation probability of 0.5, a crossover probability of 0.5, and uses a tournament selection operator with two individuals. This operation is performed on each iteration. Figure 4 This is a schematic diagram illustrating the average tunneling efficiency function value and the maximum tunneling efficiency function value of all individuals in each generation during the genetic algorithm optimization process provided in this application embodiment. Figure 5The diagram provided for the embodiments of this application shows the final optimization result.

[0057] S5. Add the optimal tunneling control parameters into the TBM dataset to obtain an updated TBM dataset. Train the real-time optimization decision model for tunneling control parameters based on the updated TBM dataset to obtain a trained real-time optimization decision model for tunneling control parameters with the optimal control parameters as the output.

[0058] Based on the pre-training-fine-tuning deep transfer learning strategy, a real-time optimization decision model for TBM tunneling control parameters is constructed by transferring some networks and weights from the constructed TBM tunneling load parameter real-time prediction model. The model is then trained, tested, and validated using an expanded TBM multi-source heterogeneous dataset. Figure 6 This is a diagram illustrating the architecture of an intelligent optimization decision-making model for TBM tunneling control parameters based on deep transfer learning, as provided in an embodiment of this application.

[0059] In this embodiment, the real-time optimization decision model for tunneling control parameters has the same inputs as the real-time prediction model for tunneling load parameters, except that it does not use tunneling control parameters as model inputs. The output is the optimal tunneling control parameters (obtained from S4). The real-time prediction parameters for near-cooperation parameters are defined as the source model, and the real-time optimization decision model for tunneling control parameters is defined as the target model. The pre-training-fine-tuning deep transfer learning strategy for each part of the model is as follows: S51 defines the real-time prediction model of TBM tunneling load parameters used to predict TBM tunneling load parameters as the source model, and the real-time optimization decision model of tunneling control parameters obtained after training as the target model.

[0060] S52, the structure and weights of the multi-branch feature extraction modules of the source model, which are used to process real-time updated rising segment data and to process historical cyclic segment data, are transferred to the target model.

[0061] S53, the structure of the feature fusion module and multi-branch prediction module of the source model is transferred to the target model, and the weights are randomly initialized.

[0062] The structure of a neural network model is its topology, i.e., how many layers it has, how many neurons in each layer, and the weights, which are the connections between neurons and are a set of numbers. In this embodiment, some weights are transferred, some are not, and the entire structure is transferred.

[0063] S54 uses L2 regularization to fine-tune the network weights of the multi-branch feature extraction module and the feature fusion module, and retrains the branch prediction module.

[0064] The formula for the L2 regularization term is as follows:

[0065]

[0066] in, To add regularization to the loss function, Loss The original loss function, y and These are the actual value and the predicted value of the feature, respectively. R ( ω ) is an L2 regularization term. This serves as the starting point for model weights, i.e., the source model weights. The current weights of the model, and Increased bias leads to a larger loss function value in the model; therefore, L2 regularization can limit the deviation between the model weights and their starting point, enabling fine-tuning of the model. Specifically, in this example… The training hyperparameters of the model are set to 0.01, and the model training hyperparameters are the same as those of the real-time prediction model of tunneling load parameters (Table 1).

[0067] S6: Acquire multi-source heterogeneous data of the TBM to be detected, and use the decision model to identify the multi-source heterogeneous data of the TBM to be detected in real time based on the tunneling control parameters to obtain the optimal tunneling control parameters.

[0068] Based on the method described in the above embodiments, this embodiment will further describe it from the perspective of a TBM tunneling control parameter intelligent optimization decision system based on deep transfer learning. This TBM tunneling control parameter intelligent optimization decision system based on deep transfer learning can be implemented as an independent entity or integrated into an electronic device. The electronic device can be a terminal, server, or other device. The terminal can include a tablet computer, laptop computer, personal computer (PC), microprocessor box, or other devices.

[0069] Please see Figure 7 , Figure 7 This application provides a detailed description of a deep transfer learning-based intelligent optimization decision-making system for TBM tunneling control parameters, applicable to electronic devices. This system may include: The acquisition module is used to acquire TBM multi-source heterogeneous data and construct the TBM dataset; The TBM tunneling load parameter prediction module trains the real-time prediction model of TBM tunneling load parameters based on the TBM dataset, and obtains the trained TBM tunneling load parameter prediction model. The tunneling benefit function construction module is used to construct the tunneling benefit function; The tunneling control parameter optimization module is used to optimize the tunneling control parameters of each tunneling cycle segment based on the tunneling benefit function and the real-time prediction model of tunneling load parameters, so as to obtain the optimal tunneling control parameters. The training module is used to add the optimal tunneling control parameters into the TBM dataset to obtain an updated TBM dataset, and to train the real-time optimization decision model of tunneling control parameters based on the updated TBM dataset to obtain a trained real-time optimization decision model of tunneling control parameters with the optimal control parameters as the output. The tunneling control parameter optimization decision module is used to acquire multi-source heterogeneous data of the TBM to be detected, and to identify the multi-source heterogeneous data of the TBM to be detected based on the real-time optimization decision model of the tunneling control parameters, so as to obtain the optimal tunneling control parameters.

[0070] In specific implementation, the above modules and / or units can be implemented as independent entities, or they can be arbitrarily combined and implemented as the same or several entities. For the specific implementation of the above modules and / or units, please refer to the previous method embodiments. For the specific beneficial effects that can be achieved, please also refer to the beneficial effects in the previous method embodiments, which will not be repeated here.

[0071] In addition, this application also provides an electronic device, which may be a computer, tablet computer, or other similar device. This electronic device can implement the steps of any embodiment of the intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning provided in this application. Therefore, it can achieve the beneficial effects that any of the intelligent optimization decision-making methods for TBM tunneling control parameters based on deep transfer learning provided in this invention can achieve, as detailed in the preceding embodiments, and will not be repeated here.

[0072] Figure 8 A specific structural block diagram of an electronic device provided in an embodiment of the present invention is shown. This electronic device can be used to implement the intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning provided in the above embodiments. The electronic device 500 can be a terminal, server, or other device. The terminal can include a tablet computer, laptop computer, personal computer (PC), microprocessor box, or other devices.

[0073] RF circuit 510 is used to receive and transmit electromagnetic waves, converting electromagnetic waves into electrical signals and vice versa, thereby enabling communication with communication networks or other devices. RF circuit 510 may include various existing circuit elements used to perform these functions, such as antennas, radio frequency transceivers, digital signal processors, encryption / decryption chips, subscriber identity modules (SIM cards), memory, etc. RF circuit 510 can communicate with various networks such as the Internet, corporate intranets, and wireless networks, or communicate with other devices via wireless networks. The aforementioned wireless networks may include cellular telephone networks, wireless local area networks (WLANs), or metropolitan area networks (MANs). The aforementioned wireless networks may use various communication standards, protocols, and technologies, including but not limited to Global System for Mobile Communication (GSM), Enhanced Data GSM Environment (EDGE), Wideband Code Division Multiple Access (WCDMA), Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Wireless Fidelity (Wi-Fi) (such as IEEE 802.11a, IEEE 802.11b, IEEE 802.11g, and / or IEEE 802.11n), Voice over Internet Protocol (VoIP), Worldwide Interoperability for Microwave Access (Wi-Max), other protocols for email, instant messaging, and short messages, and any other suitable communication protocols, including those that have not yet been developed.

[0074] The memory 520 can be used to store software programs and modules, such as the program instructions / modules corresponding to those in the above embodiments. The processor 580 executes various functional applications and data processing by running the software programs and modules stored in the memory 520, such as taking pictures with the front-facing camera, processing the captured images, and switching the display colors of the content displayed on the screen. The memory 520 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 520 may further include memory remotely located relative to the processor 580, and these remote memories can be connected to the electronic device 500 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0075] The input unit 530 can be used to receive input numeric or character information, and to generate a keyboard and mouse related to user settings and function control. Display unit 540 can be used to display information input by the user or information provided to the user, as well as various graphical user interfaces, which can be composed of graphics, text, icons, video, and any combination thereof. Display unit 540 may include display panel 541, which may optionally be configured in the form of LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), or other similar forms.

[0076] Audio circuitry 560, speaker 561, and microphone 562 provide an audio interface between the user and electronic device 500. Audio circuitry 560 converts received audio data into electrical signals and transmits them to speaker 561, where speaker 561 converts them into sound signals for output. Conversely, microphone 562 converts collected sound signals into electrical signals, which are then received by audio circuitry 560, converted back into audio data, and processed by processor 580. The audio data is then transmitted via RF circuitry 510 to, for example, another terminal, or output to memory 520 for further processing. Audio circuitry 560 may also include an earphone jack to facilitate communication between external headphones and electronic device 500.

[0077] Electronic device 500, through transmission module 570 (e.g., Wi-Fi module), can help users receive requests, send information, etc., providing users with wireless broadband internet access. Although transmission module 570 is shown in the figure, it is understood that it is not an essential component of electronic device 500 and can be omitted as needed without changing the essence of the invention.

[0078] The processor 580 is the control center of the electronic device 500. It connects to various parts of the phone via various interfaces and lines, and performs various functions and processes data of the electronic device 500 by running or executing software programs and / or modules stored in the memory 520, and by calling data stored in the memory 520, thereby providing overall monitoring of the electronic device. Optionally, the processor 580 may include one or more processing cores; in some embodiments, the processor 580 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 580.

[0079] Electronic device 500 also includes a power supply 590 (such as a battery) that supplies power to various components. In some embodiments, the power supply may be logically connected to processor 580 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 590 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0080] Although not shown, the electronic device 500 also includes cameras (such as front-facing cameras and rear-facing cameras), Bluetooth modules, etc., which will not be described in detail here. Specifically, in this embodiment, the display unit of the electronic device is a touch screen display, and the mobile terminal also includes a memory and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by one or more processors. One or more programs contain instructions for performing the following operations: Acquire TBM multi-source heterogeneous data and construct a TBM dataset; The real-time prediction model for TBM tunneling load parameters was trained based on the TBM dataset to obtain the trained TBM tunneling load parameter prediction model. Construct the tunneling benefit function; Based on the aforementioned tunneling benefit function and real-time prediction model for tunneling load parameters, the tunneling control parameters for each tunneling cycle are optimized to obtain the optimal tunneling control parameters. The optimal tunneling control parameters are added to the TBM dataset to obtain an updated TBM dataset. The real-time optimization decision model for tunneling control parameters is trained based on the updated TBM dataset to obtain a trained real-time optimization decision model for tunneling control parameters with the optimal control parameters as the output. Acquire multi-source heterogeneous data of the TBM to be detected, and use the real-time optimization decision model based on the tunneling control parameters to identify the multi-source heterogeneous data of the TBM to be detected, so as to obtain the optimal tunneling control parameters.

[0081] In practice, the above modules can be implemented as independent entities or combined in any way to be implemented as the same or several entities. For the specific implementation of the above modules, please refer to the previous method implementation examples, which will not be repeated here.

[0082] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor. Therefore, embodiments of the present invention provide a storage medium storing multiple instructions that can be loaded by a processor to execute the steps of any embodiment of the intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning provided by the present invention.

[0083] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0084] Since the instructions stored in the storage medium can execute the steps in any embodiment of the intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning provided in the embodiments of the present invention, the beneficial effects that any intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning provided in the embodiments of the present invention can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0085] The foregoing has provided a detailed description of a TBM tunneling control parameter intelligent optimization decision-making method, system, storage medium, and electronic device based on deep transfer learning, as provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for intelligent optimization and decision-making of TBM tunneling control parameters based on deep transfer learning, characterized in that, The method includes: Acquire TBM multi-source heterogeneous data and construct a TBM dataset; The real-time prediction model for TBM tunneling load parameters is trained based on the TBM dataset to obtain the trained TBM tunneling load parameter prediction model. The real-time prediction model for TBM tunneling load parameters includes a multi-branch input deep neural network for processing different types of TBM multi-source heterogeneous data, including an input module, a multi-branch feature extraction module, a feature fusion module, a multi-branch prediction module, and an output module. The processing steps of the real-time prediction model for tunneling load parameters include: dividing the preprocessed TBM multi-source heterogeneous data into three categories: real-time updated ascending segment data, historical cycle segment data, and tunneling control parameters, and inputting them into the input module respectively; extracting deep features of different forms of TBM multi-source heterogeneous data through the multi-branch feature extraction module; fusing the deep features of each branch through the feature fusion module to obtain fused features; predicting the fused features through the multi-branch prediction module to obtain the predicted tunneling load parameters, and outputting them through the output module. Construct the tunneling benefit function; Based on the aforementioned tunneling benefit function and real-time prediction model for tunneling load parameters, the tunneling control parameters for each tunneling cycle are optimized to obtain the optimal tunneling control parameters. The optimal tunneling control parameters are added to the TBM dataset to obtain an updated TBM dataset. Based on the updated TBM dataset, a real-time optimization decision model for tunneling control parameters is trained to obtain a trained real-time optimization decision model for tunneling control parameters that outputs the optimal control parameters. The training of the real-time optimization decision model for tunneling control parameters based on the updated TBM dataset includes: defining a real-time prediction model for TBM tunneling load parameters (used to predict TBM tunneling load parameters) as the source model and defining the real-time optimization decision model for tunneling control parameters as the target model; transferring the structure and weights of the multi-branch feature extraction module (used to process the real-time updated rising segment data and the historical cyclic segment data) of the source model to the target model; transferring the structure of the feature fusion module and the multi-branch prediction module of the source model to the target model and randomly initializing the weights; fine-tuning the network weights of the multi-branch feature extraction module and the feature fusion module using L2 regularization and retraining the branch prediction module. Acquire multi-source heterogeneous data of the TBM to be detected, and use the real-time optimization decision model based on the tunneling control parameters to identify the multi-source heterogeneous data of the TBM to be detected, so as to obtain the optimal tunneling control parameters.

2. The intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning according to claim 1, characterized in that, The multi-source heterogeneous data includes TBM mechanical parameters, electro-hydraulic parameter data, surrounding rock data obtained from geological exploration, cutterhead vibration signals, and rock debris images.

3. The intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning according to claim 2, characterized in that, After constructing the TBM dataset, it is necessary to preprocess the TBM dataset: Non-working data in the TBM mechanical parameters and electro-hydraulic parameters are removed to form TBM tunneling parameter data in units of cycle segments; Identify data anomalies and frame drops in the TBM tunneling parameters, correct locally weakly anomalous data, and remove globally strong anomalous data; Each cycle segment is divided into an empty push segment, an ascending segment, a stable segment, and a descending segment to obtain preprocessed historical cycle segment data; The TBM multi-source heterogeneous data other than the mechanical parameters and electro-hydraulic parameters are subjected to data noise reduction, missing and abnormal data supplementation, and invalid data removal.

4. The intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning according to claim 1, characterized in that, The tunneling benefit function is: in, F , T , N , AR These are the average values ​​of total propulsion force, cutterhead torque, cutterhead rotation speed, and propulsion speed during the stabilization phase. For the tunneling benefit function, SE S represents the tunneling specific energy, which is the energy consumption of a TBM excavating a unit volume of rock, and S represents the cross-sectional area of ​​the tunnel face. CW The cutterhead wear coefficient represents the amount of wear on the cutterhead per unit volume of rock during TBM tunneling. , , The weighting coefficient is between 0 and 1.

5. The intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning according to claim 1, characterized in that, The multi-branch feature extraction module is a temporal convolutional neural network, a long short-term memory network, or a fully connected neural network, and the feature fusion module and the multi-branch prediction module are fully connected neural networks.

6. A deep transfer learning-based intelligent optimization decision-making system for TBM tunneling control parameters, wherein the deep transfer learning-based intelligent optimization decision-making system for TBM tunneling control parameters is used to implement the deep transfer learning-based intelligent optimization decision-making method for TBM tunneling control parameters as described in claim 1, characterized in that, include: The acquisition module is used to acquire TBM multi-source heterogeneous data and construct the TBM dataset; The TBM tunneling load parameter prediction module trains the real-time prediction model of TBM tunneling load parameters based on the TBM dataset, and obtains the trained TBM tunneling load parameter prediction model. The tunneling benefit function construction module is used to construct the tunneling benefit function; The tunneling control parameter optimization module is used to optimize the tunneling control parameters of each tunneling cycle segment based on the tunneling benefit function and the real-time prediction model of tunneling load parameters, so as to obtain the optimal tunneling control parameters. The training module is used to add the optimal tunneling control parameters into the TBM dataset to obtain an updated TBM dataset, and to train the real-time optimization decision model of tunneling control parameters based on the updated TBM dataset to obtain a trained real-time optimization decision model of tunneling control parameters with the optimal control parameters as the output. The tunneling control parameter optimization decision module is used to acquire multi-source heterogeneous data of the TBM to be detected, and to identify the multi-source heterogeneous data of the TBM to be detected based on the real-time optimization decision model of the tunneling control parameters, so as to obtain the optimal tunneling control parameters.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor to execute the intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning as described in any one of claims 1 to 5.

8. An electronic device, characterized in that, The device includes a processor and a memory, the processor being electrically connected to the memory, the memory being used to store instructions and data, and the processor being used to execute the steps in the intelligent optimization decision-making method for TBM tunneling control parameters based on deep transfer learning as described in any one of claims 1 to 5.

Images