Pipe jacking machine control method, system, terminal and storage medium

Abstract

The application discloses a pipe jacking machine control method and system, a terminal and a storage medium, relates to the technical field of device deviation correction control, and comprises the following steps: obtaining tunneling parameters corresponding to the pipe jacking machine, wherein the tunneling parameters comprise deviation correction oil cylinder stroke parameters; inputting the tunneling parameters into a trained attitude deviation prediction model, obtaining a machine head attitude deviation result corresponding to the pipe jacking machine through the attitude deviation prediction model, wherein at least a part of extended training data is included in a training sample set used by the attitude deviation prediction model during pre-training, the extended training data comprises first training samples with missing real attitude deviation labels, and attitude deviation pseudo-labels generated for the first training samples by a trained pseudo-label generation model; and adjusting the stroke difference between the deviation correction oil cylinders of the pipe jacking machine according to the machine head attitude deviation result. In this way, the accuracy of pipe jacking machine deviation correction control can be improved, and the pipe jacking machine control effect can be improved.

Description

Technical Field

[0001] This application relates to the field of equipment correction and control technology, and in particular to a control method, system, terminal and storage medium for a pipe jacking machine. Background Technology

[0002] With the development of science and technology, the use of various mechanical equipment and the correction and control during its use are receiving increasing attention. Pipe jacking construction, due to its advantages such as minimal disturbance to surface traffic, suitability for complex underground environments, and good construction continuity, is widely used in the construction of urban underground pipelines, integrated utility tunnels, and underground passages. During pipe jacking construction, whether the pipe jacking machine can stably advance along the target axis directly affects construction accuracy, structural safety, and the response of the surrounding environment. Therefore, correction and control of pipe jacking machines are also receiving increasing attention.

[0003] Currently, the attitude control of pipe jacking machines typically relies on attitude deviation data provided by the guidance system and the experience-based judgment of on-site operators. The attitude deviation is manually calculated to adjust the parameters of the correction cylinders, thus correcting the machine's attitude. However, the deviation records of the guidance system often rely on manual clicking or reading, resulting in limited data. The problem with existing technology is that manual control depends on the operator's experience, leading to low accuracy. Furthermore, the attitude deviation data collected by the guidance system is sparse and difficult to use to train relevant models for automatic control. Therefore, existing solutions are not conducive to improving the accuracy of pipe jacking machine correction control.

[0004] Therefore, the relevant technologies still need to be improved and developed. Summary of the Invention

[0005] The main purpose of this application is to provide a control method, system, terminal and storage medium for a pipe jacking machine, which aims to solve the technical problems in the related technology where the manual control method for pipe jacking machines has low accuracy and the attitude deviation data collected by the guidance system is sparse, making it difficult to use to train relevant models to achieve automatic control, thus hindering the improvement of the accuracy of the pipe jacking machine's deviation correction control.

[0006] To achieve the above objectives, the first aspect of this application provides a method for controlling a pipe jacking machine, wherein the method includes: Obtain the tunneling parameters corresponding to the pipe jacking machine, including the stroke parameters of the correction cylinder; The aforementioned tunneling parameters are input into the trained attitude deviation prediction model, and the attitude deviation result of the tunnel jacking machine corresponding to the aforementioned tunnel jacking machine is obtained by the aforementioned attitude deviation prediction model. The aforementioned attitude deviation result is used to indicate the deviation between the actual axis of the tunnel jacking machine and the target axis. When the aforementioned attitude deviation prediction model is pre-trained, the training sample set used includes at least a part of the extended training data. The aforementioned extended training data includes a first training sample that lacks a true attitude deviation label, and attitude deviation pseudo labels generated for the aforementioned first training sample by the trained pseudo label generation model. Based on the above head posture deviation results, adjust the stroke difference between the correction cylinders of the above pipe jacking machine; The pseudo-label generation model mentioned above is pre-trained based on the following steps: The continuous second-level tunneling data corresponding to the above-mentioned pipe jacking machine and the sparse real attitude deviation data recorded by the guidance system are obtained. The continuous second-level tunneling data includes tunneling parameters that correspond one-to-one with multiple time points, and the sparse real attitude deviation data includes multiple real attitude deviation values ​​that correspond one-to-one with multiple time points. The number of time points corresponding to the real attitude deviation values ​​is less than the number of time points corresponding to the tunneling parameters. Time alignment and feature construction are performed on the above continuous second-level tunneling data and the above sparse real attitude deviation data to obtain an input feature set and a real label set. The above input feature set includes a first training sample and a second training sample. The time corresponding to the first training sample is missing the real attitude deviation value. The second training sample corresponds one-to-one with the real attitude deviation label in the above real label set. The pseudo-label generation model is trained by using the second training sample as the training input and the real pose deviation label as the training label.

[0007] Optionally, the aforementioned tunneling parameters may also include at least one of the following: advance speed, jacking force of the jacking cylinder, cutterhead parameters, soil chamber parameters, advance status, construction status parameters, and geological information.

[0008] Optionally, the aforementioned geological information includes at least one of the following: stratigraphic type, soil parameters, burial depth information, groundwater information, and geological stratification information corresponding to the construction area where the pipe jacking machine is located.

[0009] Optionally, the above attitude deviation prediction model is pre-trained based on the following steps: The first training sample is input into the trained pseudo-label generation model, and the pseudo-label generation model generates the pose deviation pseudo-label corresponding to the first training sample. The first training sample and the second training sample are used as model inputs, and the pseudo-labels and true attitude deviation labels are used as training labels to train the attitude deviation prediction model.

[0010] Optionally, the pseudo-label generation model mentioned above can be any one of support vector regression model, tree model, or ensemble regression model; The above-mentioned attitude deviation prediction model can be any one of the following: multilayer perceptron model, recurrent neural network model, gated recurrent unit model, or temporal convolutional network model.

[0011] Optionally, the above-mentioned nose attitude deviation results include nose horizontal deviation and nose vertical deviation; The above-mentioned adjustment of the stroke difference between the correction cylinders of the pipe jacking machine based on the above-mentioned head posture deviation results includes: Based on the above-mentioned horizontal deviation of the machine head, adjust the stroke difference between the correction cylinders arranged horizontally in the above-mentioned pipe jacking machine. Based on the aforementioned vertical deviation of the machine head, adjust the stroke difference between the correction cylinders arranged vertically in the jacking machine.

[0012] A second aspect of this application provides a control system for a pipe jacking machine, wherein the system includes: The data acquisition module is used to acquire the tunneling parameters corresponding to the pipe jacking machine, including the stroke parameters of the correction cylinder. The deviation prediction module is used to input the above-mentioned tunneling parameters into the trained attitude deviation prediction model, and to predict the attitude deviation result of the tunnel jacking machine corresponding to the machine head through the attitude deviation prediction model. The attitude deviation result of the machine head is used to indicate the deviation between the actual axis of the machine head and the target axis of the tunnel jacking machine. When the attitude deviation prediction model is pre-trained, the training sample set used includes at least a part of the extended training data. The extended training data includes a first training sample that lacks a real attitude deviation label, and attitude deviation pseudo labels generated for the first training sample by the trained pseudo label generation model. The control module is used to adjust the stroke difference between the correction cylinders of the pipe jacking machine based on the above-mentioned head posture deviation results. The pseudo-label generation model mentioned above is pre-trained based on the following steps: The continuous second-level tunneling data corresponding to the above-mentioned pipe jacking machine and the sparse real attitude deviation data recorded by the guidance system are obtained. The continuous second-level tunneling data includes tunneling parameters that correspond one-to-one with multiple time points, and the sparse real attitude deviation data includes multiple real attitude deviation values ​​that correspond one-to-one with multiple time points. The number of time points corresponding to the real attitude deviation values ​​is less than the number of time points corresponding to the tunneling parameters. Time alignment and feature construction are performed on the above continuous second-level tunneling data and the above sparse real attitude deviation data to obtain an input feature set and a real label set. The above input feature set includes a first training sample and a second training sample. The time corresponding to the first training sample is missing the real attitude deviation value. The second training sample corresponds one-to-one with the real attitude deviation label in the above real label set. The pseudo-label generation model is trained by using the second training sample as the training input and the real pose deviation label as the training label.

[0013] A third aspect of this application provides a terminal, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of any of the above-described pipe jacking machine control methods.

[0014] A fourth aspect of this application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described pipe jacking machine control methods.

[0015] As can be seen from the above, the present application provides a pipe jacking machine control method, which includes: obtaining the tunneling parameters corresponding to the pipe jacking machine, wherein the tunneling parameters include the stroke parameters of the correction cylinders; inputting the tunneling parameters into a trained attitude deviation prediction model, and predicting the head attitude deviation result of the pipe jacking machine through the attitude deviation prediction model, wherein the head attitude deviation result is used to indicate the deviation between the actual axis of the head of the pipe jacking machine and the target axis, wherein the attitude deviation prediction model, during pre-training, uses a training sample set including at least a portion of extended training data, wherein the extended training data includes a first training sample lacking a true attitude deviation label, and attitude deviation pseudo-labels generated for the first training sample by a trained pseudo-label generation model; adjusting the stroke difference between the correction cylinders of the pipe jacking machine according to the head attitude deviation result; wherein the pseudo-label generation model is based on the following steps... First, training is performed: Continuous, second-level tunneling data corresponding to the aforementioned pipe jacking machine and sparse real attitude deviation data recorded by the guidance system are acquired. The continuous, second-level tunneling data includes tunneling parameters corresponding one-to-one with multiple time points, and the sparse real attitude deviation data includes multiple real attitude deviation values ​​corresponding one-to-one with multiple time points. The number of time points corresponding to the real attitude deviation values ​​is less than the number of time points corresponding to the aforementioned tunneling parameters. Time alignment and feature construction are performed on the continuous, second-level tunneling data and the sparse real attitude deviation data to obtain an input feature set and a real label set. The input feature set includes a first training sample and a second training sample. The time point corresponding to the first training sample lacks a real attitude deviation value, and the second training sample corresponds one-to-one with the real attitude deviation labels in the real label set. The second training sample is used as the training input, and the real attitude deviation labels are used as the training labels to train the pseudo-label generation model.

[0016] Thus, during the control of the pipe jacking machine, the attitude deviation is not manually calculated for control. Instead, based on the tunneling parameters, a trained attitude deviation prediction model automatically calculates the corresponding head attitude deviation of the pipe jacking machine. Furthermore, during the training of this attitude deviation prediction model, the training sample set includes at least a portion of extended training data. This extended training data includes a first training sample lacking a true attitude deviation label, and attitude deviation pseudo-labels generated for the first training sample by a trained pseudo-label generation model. The pseudo-label generation model is trained using a second training sample from the sparse true attitude deviation data that corresponds one-to-one with the true attitude deviation label. This allows it to fully learn the initial mapping relationship between tunneling parameters and attitude deviations under limited true labels, thereby generating reliable attitude deviation pseudo-labels for the first training sample lacking a true attitude deviation label. In other words, during the training of this attitude deviation prediction model, the trained pseudo-label model can reasonably expand the attitude deviation data used as labels, thereby reasonably expanding the training samples of the attitude deviation prediction model, improving the training effect of the attitude deviation prediction model, and obtaining a usable attitude deviation prediction model. In this way, the attitude deviation of the pipe jacking machine can be automatically predicted based on the attitude deviation prediction model, thereby adjusting the stroke difference between the correction cylinders of the pipe jacking machine and realizing the control of the pipe jacking machine. This is beneficial to improving the accuracy of the correction control of the pipe jacking machine and enhancing the control effect of the pipe jacking machine. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a pipe jacking machine control method provided in an embodiment of this application; Figure 2 This is a schematic diagram of an online control process provided in an embodiment of this application; Figure 3 This is a schematic diagram of the constituent modules of a pipe jacking machine control system provided in an embodiment of this application; Figure 4 This is a block diagram illustrating the internal structure of a terminal provided in an embodiment of this application. Detailed Implementation

[0019] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail.

[0020] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0021] It should also be understood that the terminology used in this application specification is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this application specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0022] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0023] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to classification to [the described condition or event]."

[0024] 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 a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0025] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0026] Current pipe jacking machine attitude control typically relies on attitude deviation data provided by the guidance system and the experience-based judgment of on-site operators. The attitude of the pipe jacking machine is corrected by manually adjusting the parameters of the correction cylinders. This manual control method depends on the operator's experience, resulting in low accuracy. Furthermore, in actual construction, deviation recording by the guidance system often relies on manual clicking or reading, with a recording frequency far lower than the sampling frequency of the construction equipment's second-level tunneling parameters. This leads to sparse and discrete attitude deviation data. Therefore, the attitude deviation data collected by the guidance system is sparse and difficult to use to train relevant models for automatic control. Relying solely on limited manually clicked deviation data for modeling easily leads to problems such as insufficient training samples, incomplete sample distribution, and limited model generalization ability, making it difficult to meet the requirements for continuous prediction and online control of the pipe jacking machine's head attitude under complex working conditions.

[0027] In contrast, tunneling data generated during pipe jacking construction, such as propulsion speed, jacking force, hydraulic cylinder stroke, and cutterhead operating parameters, are typically continuous, high-frequency, and highly time-series. There is a significant difference in sampling frequency between high-frequency tunneling data and low-frequency deviation labels, and existing methods struggle to fully utilize the attitude evolution information contained in continuous construction data.

[0028] Existing technologies can only utilize a small number of sparse real attitude deviation labels for model training, resulting in a scarcity of training samples and incomplete coverage of working conditions. This leads to poor generalization ability and insufficient continuity of attitude prediction in the constructed control model, making it impossible to accurately capture the dynamic evolution of the pipe jacking machine's head attitude. Furthermore, traditional control methods cannot achieve real-time prediction and dynamic correction of attitude deviations, have low automation levels, and are prone to problems such as excessive head offset and excessive deviation of the construction axis, making it difficult to meet the high-precision control requirements of complex geological formations and long-distance pipe jacking construction.

[0029] To address at least one of the aforementioned technical problems, this application proposes a pipe jacking machine control method. The method includes: acquiring tunneling parameters corresponding to the pipe jacking machine, wherein the tunneling parameters include stroke parameters of the correction cylinders; inputting the tunneling parameters into a trained attitude deviation prediction model, and predicting the head attitude deviation result of the pipe jacking machine through the attitude deviation prediction model, wherein the head attitude deviation result indicates the deviation between the actual axis of the pipe jacking machine's head and the target axis, wherein the attitude deviation prediction model, during pre-training, uses a training sample set including at least a portion of extended training data, the extended training data including a first training sample lacking a true attitude deviation label, and attitude deviation pseudo-labels generated for the first training sample by a trained pseudo-label generation model; adjusting the stroke difference between the correction cylinders of the pipe jacking machine according to the head attitude deviation result; wherein the pseudo-label generation model... The pre-training process is based on the following steps: First, continuous second-level tunneling data corresponding to the aforementioned pipe jacking machine and sparse real attitude deviation data recorded by the guidance system are acquired. The continuous second-level tunneling data includes tunneling parameters corresponding one-to-one with multiple time points, and the sparse real attitude deviation data includes multiple real attitude deviation values ​​corresponding one-to-one with multiple time points. The number of time points corresponding to the real attitude deviation values ​​is less than the number of time points corresponding to the aforementioned tunneling parameters. Second, time alignment and feature construction are performed on the continuous second-level tunneling data and the sparse real attitude deviation data to obtain an input feature set and a real label set. The input feature set includes a first training sample and a second training sample. The time point corresponding to the first training sample lacks a real attitude deviation value, and the second training sample corresponds one-to-one with the real attitude deviation labels in the real label set. Third, the pseudo-label generation model is trained using the second training sample as training input and the real attitude deviation labels as training labels.

[0030] Thus, during the control of the pipe jacking machine, the attitude deviation is not manually calculated for control. Instead, based on the tunneling parameters, a trained attitude deviation prediction model automatically calculates the corresponding head attitude deviation of the pipe jacking machine. Furthermore, during the training of this attitude deviation prediction model, the training sample set includes at least a portion of extended training data. This extended training data includes a first training sample lacking a true attitude deviation label, and attitude deviation pseudo-labels generated for the first training sample by a trained pseudo-label generation model. The pseudo-label generation model is trained using a second training sample from the sparse true attitude deviation data that corresponds one-to-one with the true attitude deviation label. This allows it to fully learn the initial mapping relationship between tunneling parameters and attitude deviations under limited true labels, thereby generating reliable attitude deviation pseudo-labels for the first training sample lacking a true attitude deviation label. In other words, during the training of this attitude deviation prediction model, the trained pseudo-label model can reasonably expand the attitude deviation data used as labels, thereby reasonably expanding the training samples of the attitude deviation prediction model, improving the training effect of the attitude deviation prediction model, and obtaining a usable attitude deviation prediction model. In this way, the attitude deviation of the pipe jacking machine can be automatically predicted based on the attitude deviation prediction model, thereby adjusting the stroke difference between the correction cylinders of the pipe jacking machine and realizing the control of the pipe jacking machine. This is beneficial to improving the accuracy of the correction control of the pipe jacking machine and enhancing the control effect of the pipe jacking machine.

[0031] like Figure 1 As shown in the figure, this application provides a method for controlling a pipe jacking machine. Specifically, the method includes the following steps: Step S100: Obtain the tunneling parameters corresponding to the pipe jacking machine, wherein the tunneling parameters include the stroke parameters of the correction cylinder; Step S200: Input the above-mentioned tunneling parameters into the trained attitude deviation prediction model, and obtain the attitude deviation result of the tunnel jacking machine corresponding to the machine head through the attitude deviation prediction model. The attitude deviation result of the machine head is used to indicate the deviation between the actual axis of the machine head and the target axis of the tunnel jacking machine. When the attitude deviation prediction model is pre-trained, the training sample set used includes at least a part of the extended training data. The extended training data includes a first training sample that is missing the real attitude deviation label, and attitude deviation pseudo labels generated for the first training sample by the trained pseudo label generation model. Step S300: Based on the above head posture deviation results, adjust the stroke difference between the correction cylinders of the above-mentioned pipe jacking machine.

[0032] The pseudo-label generation model mentioned above is pre-trained based on the following steps: The continuous second-level tunneling data corresponding to the above-mentioned pipe jacking machine and the sparse real attitude deviation data recorded by the guidance system are obtained. The continuous second-level tunneling data includes tunneling parameters that correspond one-to-one with multiple time points, and the sparse real attitude deviation data includes multiple real attitude deviation values ​​that correspond one-to-one with multiple time points. The number of time points corresponding to the real attitude deviation values ​​is less than the number of time points corresponding to the tunneling parameters. Time alignment and feature construction are performed on the above continuous second-level tunneling data and the above sparse real attitude deviation data to obtain an input feature set and a real label set. The above input feature set includes a first training sample and a second training sample. The time corresponding to the first training sample is missing the real attitude deviation value. The second training sample corresponds one-to-one with the real attitude deviation label in the above real label set. The pseudo-label generation model is trained by using the second training sample as the training input and the real pose deviation label as the training label.

[0033] The aforementioned tunneling parameters refer to various sensor data collected in real time during the pipe jacking machine construction process, used to characterize the equipment's operating status, control operations, and construction progress, including at least the stroke parameters of the correction cylinders. These correction cylinder stroke parameters represent the current extension length of the correction cylinders, reflecting the execution status of the correction action. The aforementioned target axis refers to the ideal spatial curve (or straight line) along which the pipe jacking machine head should advance from the starting shaft to the receiving shaft, as predetermined in the pipe jacking construction process; i.e., the design axis. It should be noted that the target axis is predetermined by the construction drawings, guidance system, or construction control coordinate system. Subsequent deviations in positive and negative directions and control directions are determined according to the coordinate conventions of the guidance system and the actual arrangement of the correction cylinders.

[0034] In this embodiment, the aforementioned tunneling parameters include at least the stroke parameters of the correction cylinder. In some application scenarios, the aforementioned tunneling parameters also include at least one of the following: propulsion speed, jacking force of the jacking cylinder, cutterhead parameters, soil chamber parameters, propulsion status, construction status parameters, and geological information. Specifically, the aforementioned propulsion speed is used to indicate the speed at which the tunnel head advances forward, i.e., the tunnel head propulsion speed; the aforementioned jacking force of the jacking cylinder is used to indicate the magnitude of the thrust applied by the propulsion system; the aforementioned cutterhead parameters include parameters such as cutterhead rotation speed and torque, used to indicate the cutting status; the aforementioned soil chamber parameters are indirect information reflecting the state of the soil ahead, such as soil chamber pressure; the aforementioned construction status parameters are used to indicate which construction stage is currently in, such as propulsion, stop, or takeover.

[0035] It should be noted that the attitude change of the pipe jacking machine is not only affected by changes in construction parameters, but also closely related to geological conditions. Geological factors such as the stratum type, soil parameters, and groundwater conditions in different construction sections will cause changes in the influence of the correction cylinder parameters on the machine head's attitude deviation. In this embodiment, the aforementioned geological information includes at least one of the following: stratum type, soil parameters, burial depth, groundwater information, and geological stratification information corresponding to the construction area where the pipe jacking machine is located.

[0036] Specifically, the above attitude deviation prediction model is pre-trained based on the following steps: The first training sample is input into the trained pseudo-label generation model, and the pseudo-label generation model generates the pose deviation pseudo-label corresponding to the first training sample. The first training sample and the second training sample are used as model inputs, and the pseudo-labels and true attitude deviation labels are used as training labels to train the attitude deviation prediction model.

[0037] First, continuous, second-level tunneling data and sparse, real-time attitude deviation data of the guidance system are collected during the pipe jacking machine construction process. The continuous, second-level tunneling data consists of tunneling parameters collected every second (or at preset intervals), which may include parameters such as the stroke of the correction cylinder, propulsion speed, jacking cylinder force, cutterhead rotation speed, soil chamber pressure, and geological information. The sparse, real-time attitude deviation data consists of manually recorded horizontal and vertical deviations of the machine head; the recording time for these real-time deviations is much shorter than the time it takes for the tunneling parameters to be collected. In a specific application scenario, the continuous, second-level tunneling data should at least include the correction cylinder stroke data and other time-series parameters that characterize the pipe jacking construction status; the sparse, manually recorded deviation data should at least include the machine head attitude deviation data recorded by the guidance system.

[0038] Multi-source data undergoes time alignment processing, matching tunneling data and deviation data from different sampling frequencies according to a unified time benchmark. Simultaneously, data cleaning operations such as data cleaning, outlier removal, and feature normalization are performed to construct an input feature set and a true label set. Feature construction includes extracting raw features characterizing the pipe jacking construction status from continuous second-level tunneling data, further extracting temporal features characterizing the trend of the jacking head's attitude change, and then combining this with geological information at the corresponding time or construction location to form an input feature set for attitude modeling. Raw features include at least one or more of the following: propulsion speed, jacking cylinder force, cutterhead parameters, soil chamber parameters, propulsion status parameters, construction status parameters, and correction cylinder stroke parameters. Temporal features include at least one or more of the following extracted based on a preset time window: average value, maximum value, minimum value, standard deviation, rate of change, cumulative change, difference features, and lag features.

[0039] The input feature set contains two types of training samples: a first training sample without a true attitude deviation label, and a second training sample matched with a true attitude deviation label. The aforementioned time alignment involves matching continuous second-level tunneling data (including geological information) and sparse attitude deviation data according to a unified time benchmark or a unified construction progress benchmark to form input feature samples corresponding to the true attitude deviation labels. It should be noted that when different data sources have inconsistent sampling frequencies, they can be unified to the same time benchmark through resampling, interpolation, aggregation, or nearest neighbor matching.

[0040] It should be noted that deviation data is used to indicate the deviation between the tunnel boring machine (TBM) and the expected tunneling axis. Geological information can be correlated with construction data according to construction mileage, ring number, or section number to enhance the model's ability to represent attitude change patterns under different geological conditions. When geological information is given in the form of construction section, station number, pipe section number, or jacking mileage, it is mapped to the corresponding time period, jacking interval, or control time based on the current construction location, thereby completing multi-source data matching.

[0041] A pseudo-label generation model was selected, using the second training sample as input and the corresponding real pose deviation labels as supervision labels to complete model training. All first training samples were then input into the trained pseudo-label generation model to generate pseudo-labels for pose deviations corresponding to each missing label sample. During training, the pseudo-label generation model learned the initial relationship between input features and the jacking machine's pose deviations based on a small number of real pose deviation labels.

[0042] A posture deviation prediction model is selected, and the first training sample with pseudo-labels and the second training sample with real labels are used together as the training set to complete the iterative training of the posture deviation prediction model, so as to obtain a posture deviation prediction model that can be predicted online in real time.

[0043] It should be noted that the training termination conditions for the two models mentioned above can be set and adjusted according to actual needs. For example, the number of iterations and / or loss thresholds can be set for the two models respectively, and the training of the models can be judged based on the number of iterations and / or loss thresholds.

[0044] Specifically, the aforementioned pseudo-label generation model can be any one of the following: support vector regression model, tree model, or ensemble regression model. The above-mentioned posture deviation prediction model can be any one of the following: multilayer perceptron model, recurrent neural network model, gated recurrent unit model, or temporal convolutional network model. Other types of models can also be used, such as support vector machine model, tree model, ensemble learning model, etc., without specific limitations.

[0045] In some application scenarios, the above attitude deviation prediction model is used to output the attitude deviation results of the pipe jacking machine head at the current moment or in the future moment, wherein the attitude deviation results at the future moment are predicted based on the input features within the current moment and the preset time window before it.

[0046] Specifically, during training, the attitude deviation prediction model can use training samples (including the first and second training samples) within a preset time window corresponding to the target time as model input. The preset time window is a period of time ending at the target time, and the specific length can be set and adjusted according to actual needs.

[0047] Correspondingly, when using the above attitude deviation model for data prediction, the tunneling parameters within the preset time window corresponding to the time to be predicted can be used as model input to obtain the corresponding prediction results.

[0048] Specifically, for moments with recorded true deviations, the corresponding horizontal and / or vertical nose deviations are used as labels for true attitude deviations, forming a true attitude deviation label set. In some applications, the current input features at the moment corresponding to the label are used as the model input for the training samples at that moment; in other applications, historical time-series features are constructed using continuous tunneling data or correction control parameters within a preset time window prior to the moment corresponding to the label, and these historical time-series features are correlated with the attitude deviation at the moment corresponding to the label, thus forming input feature samples containing historical information.

[0049] In some application scenarios, a wrapper-style feature selection method is employed during the training process of the aforementioned pseudo-label generation model. This pseudo-label generation model can be a supervised regression model, such as support vector regression, random forest regression, gradient boosting tree regression, XGBoost regression, or other models suitable for attitude deviation regression prediction; no specific limitations are imposed here. It should be noted that the purpose of introducing the wrapper-style feature selection method is to: for a multi-source heterogeneous candidate feature set composed of continuous second-level tunneling data, geological information, and deviation control parameters, select feature combinations that are more effective for attitude deviation prediction, thereby improving the accuracy of pseudo-label generation and reducing the adverse effects of noisy and redundant features on the quality of subsequent expanded training samples.

[0050] The aforementioned pseudo-label generation model is used to learn the initial mapping relationship between the input feature set and the attitude deviation of the tunnel jacking machine's head. Specifically, this model is trained on samples with actual deviation labels, and establishes an initial estimation model for the head's attitude deviation through the joint input of continuous second-level tunneling features and geological information. After training, the model is used to predict attitude deviations for moments when no actual deviation data is recorded, generating pseudo-labels for the corresponding moments. This step transforms a large number of second-level moments lacking deviation labels into usable samples with estimated attitude labels, thus forming extended attitude label data that matches the continuous second-level tunneling data.

[0051] In some application scenarios, pseudo-labels can be screened or weighted based on the estimated confidence level, error range, or other reliability indicators of the pseudo-label generation model output, in order to improve the effectiveness of subsequent training samples.

[0052] Specifically, for moments with actual deviation records, the actual attitude deviation label is used; for moments without actual deviation records but whose attitude deviations have been estimated by the pseudo-label generation model, the corresponding pseudo-label is used. This results in training samples covering a more continuous time range.

[0053] In some application scenarios, when training a posture deviation prediction model, a weighted fusion of the first and second training samples can be performed to obtain a corresponding sample set, resulting in better training performance. During weighted fusion, different training weights can be assigned to the first and second training samples respectively. For example, the training weight of the first training sample can be set lower than that of the second training sample, so that the influence of training samples with true labels is greater during model training. This expands the sample size while controlling the adverse effects of pseudo-label errors on subsequent model training, leading to better training results. Based on the above sample fusion operation, the original sparse posture labels are expanded into a more continuous and richer posture sample set, thereby improving the learning ability of the subsequent posture deviation prediction model for continuous construction processes.

[0054] It should be noted that in practical applications, under the same working conditions, the pseudo-label generation model is trained first, followed by the attitude deviation prediction model. When training the attitude deviation prediction model, the first training samples obtained during the pseudo-label generation model training process can be used, or other samples with real attitude deviation labels can be used; no specific limitation is made here. In this embodiment, using the first training samples obtained during the pseudo-label generation model training process enables full utilization of existing data, improving model training efficiency and effectiveness, but this is not considered a specific limitation.

[0055] Specifically, the posture deviation prediction model and the pseudo-label generation model can use multi-source raw data based on the same construction process during training, but their input feature organization, training sample construction, and functional positioning are different.

[0056] The pseudo-label generation model is mainly used in the offline sample expansion stage. It learns the initial relationship between continuous second-level tunneling data, geological information, and the machine head attitude deviation based on samples with real attitude deviation labels, and generates pseudo-labels and their confidence levels for moments lacking attitude labels. The attitude deviation prediction model is mainly used in the online control stage. It predicts and outputs the machine head attitude deviation results for the current and / or future moments based on real-time available input features during construction.

[0057] The input features of the attitude deviation prediction model include real-time tunneling parameters, including at least the stroke parameters of the correction cylinder, and may further include one or more of the following: propulsion speed, jacking force of the jacking cylinder, cutterhead parameters, soil chamber parameters, propulsion status, construction status parameters, and geological information.

[0058] In some application scenarios, the input features of the attitude deviation prediction model also include historical features within a preset time window before the current moment, in order to characterize the temporal dependence of attitude changes on changes in construction status and correction actions.

[0059] During the training of the attitude deviation prediction model, extended training data is also used as model input. The horizontal and / or vertical deviations of the tunnel boring machine (TBM) are used as labels corresponding to the model output. Through training, the attitude deviation prediction model learns the correspondence between continuous second-level tunneling data, geological information, and the stroke parameters of the correction cylinders with the TBM attitude deviation.

[0060] In some application scenarios, the attitude deviation prediction model is used to output the attitude deviation result of the pipe jacking machine head at the current moment. In this case, the input features within a preset time window before the current moment are used as the model input, and the head attitude deviation at the current moment is used as the supervision label for training.

[0061] In other application scenarios, the attitude deviation prediction model is used to output the attitude deviation of the tunnel jacking machine's head at future moments to assist in advance planning and control. In this case, the input features within a preset time window before or at the current moment are used as the model input, and the head attitude deviation at future moments is used as the supervision label for training. Here, the aforementioned future moment corresponds to the moment or control step size after the current moment with a preset lead amount. The specific time difference (or control step size) between the aforementioned future moment and the current moment can be set and adjusted according to actual needs, and is not specifically limited here.

[0062] It should be noted that the data used for training the above pseudo-label generation model and the above posture deviation prediction model can be based on multi-source raw data from the same construction process, but the specific parameters used can also be set and adjusted according to actual needs, and different input features can be constructed according to different task objectives.

[0063] In a preferred embodiment, the pseudo-label generation model and the online attitude deviation prediction model adopt different types of model structures. The former preferably adopts a model suitable for small sample regression, while the latter preferably adopts a model suitable for time series prediction and online control.

[0064] It should be further noted that the aforementioned sparse real attitude deviation data includes the nose horizontal deviation and nose vertical deviation data recorded by the guidance system at discontinuous moments.

[0065] After training the attitude deviation prediction model, the model is deployed in the online control process of pipe jacking construction. Specifically, during construction, the tunneling parameters at the current moment are acquired in real time and input into the trained attitude deviation prediction model to obtain the prediction results of the machine head attitude deviation at the current moment and / or future moments.

[0066] Subsequently, the stroke difference between the correction cylinders is adjusted online based on the head posture deviation result to achieve online correction of the head posture of the pipe jacking machine. Based on the head posture deviation result and a preset control strategy (such as lookup table or calculation via a proportional-integral-derivative controller), the target value of the stroke difference between the correction cylinders of the pipe jacking machine is calculated. Based on the difference between the current actual stroke difference and the target stroke difference, the stroke difference adjustment amount corresponding to each correction cylinder is calculated. A corresponding control command is generated based on the stroke difference adjustment amount and output to the correction cylinder control execution unit to adjust the stroke difference between the correction cylinders, thereby reducing the head posture deviation.

[0067] Specifically, when the prediction results indicate that the horizontal deviation and / or vertical deviation of the tunnel jacking machine head exceeds a preset threshold or shows a deviation trend, the stroke difference between the corresponding correction cylinders is adjusted to bring the tunnel jacking machine head attitude towards the target attitude direction. Specifically, the aforementioned head attitude deviation results include the horizontal deviation and the vertical deviation of the tunnel jacking machine head. The above-mentioned adjustment of the stroke difference between the correction cylinders of the pipe jacking machine based on the above-mentioned head posture deviation results includes: Based on the above-mentioned horizontal deviation of the machine head, adjust the stroke difference between the correction cylinders arranged horizontally in the above-mentioned pipe jacking machine. Based on the aforementioned vertical deviation of the machine head, adjust the stroke difference between the correction cylinders arranged vertically in the jacking machine.

[0068] Specifically, the horizontal deviation of the jacking machine head indicates the horizontal offset of the actual axis of the jacking machine head relative to the target axis, while the vertical deviation of the jacking machine head indicates the vertical offset of the actual axis of the jacking machine head relative to the target axis.

[0069] In some application scenarios, when the horizontal deviation of the tunnel jacking machine head is positive, the stroke difference between the horizontally arranged correction cylinders is adjusted to make the machine head deflect in the direction that reduces the horizontal deviation. When the vertical deviation of the machine head is positive, the stroke difference between the vertically arranged correction cylinders is adjusted to make the machine head deflect in the direction that reduces the vertical deviation. The correspondence between the positive and negative signs of the deviations and the specific adjustment direction of the stroke difference is determined according to the coordinate conventions of the guidance system and the actual arrangement of the correction cylinders in the tunnel jacking machine, and is not specifically limited here. In this way, based on the results of automatic model prediction, the stroke difference between the correction cylinders of the tunnel jacking machine can be automatically adjusted, thereby achieving automatic online correction of the machine head attitude deviation and improving the control effect of the tunnel jacking machine.

[0070] In this embodiment, the core of online control lies in using the relationship between the stroke parameters of the correction cylinder and the attitude deviation of the machine head learned by the trained attitude deviation prediction model to convert the real-time construction parameters into attitude deviation estimation results. Then, based on the results, the stroke difference between the correction cylinders is dynamically adjusted, thereby realizing online control of the attitude of the pipe jacking machine head.

[0071] In some application scenarios, the above method also includes real-time updates to the attitude deviation prediction model. Specifically, the attitude deviation prediction model can be further fine-tuned and updated based on newly added real attitude deviation labels and / or newly added pseudo-attitude deviation labels during construction to improve the adaptability of pipe jacking machine attitude prediction and online control. Furthermore, the upper-level pseudo-label generation model can also be fine-tuned and updated in real time based on newly added real attitude deviation labels, without specific limitations here.

[0072] In this way, not only can the problem of sparse initial true deviation labels be solved, but new effective samples can also be continuously accumulated as construction progresses, thereby gradually improving the attitude prediction accuracy and online control effect.

[0073] In this embodiment of the application, the above-mentioned pipe jacking machine control method is further described in detail based on a specific application scenario. Figure 2 This is a schematic diagram of an online control process provided in an embodiment of this application, such as... Figure 2As shown, when controlling the pipe jacking machine online, real-time tunneling parameters are collected and input into the attitude deviation prediction model for prediction, obtaining the machine head attitude deviation result. Based on this result, the stroke difference between the correction cylinders is adjusted. After completing one adjustment, real-time tunneling parameter collection can continue, or real-time tunneling parameter collection can be performed according to a preset time interval, and the next round of control can be carried out, thereby realizing the online control of the pipe jacking machine.

[0074] Specifically, based on continuous, second-level tunneling data during pipe jacking construction, combined with sparse manual click deviation data recorded by the guidance system and geological information obtained from geological exploration, a pseudo-label generation model for label expansion is trained on a small number of real attitude deviation labels. This model is then used to generate pseudo-labels for attitude deviations at times with missing deviation records. Subsequently, real and pseudo-labels are merged to construct an expanded attitude training sample set. On this basis, an attitude deviation prediction model is trained to learn the mapping relationship between the stroke parameters of the correction cylinders and the attitude deviation of the pipe jacking machine head. Finally, online control of the pipe jacking machine head attitude is achieved by adjusting the stroke difference between the correction cylinders.

[0075] This application provides a pipe jacking machine control method, which includes: acquiring tunneling parameters corresponding to the pipe jacking machine, wherein the tunneling parameters include stroke parameters of the correction cylinders; inputting the tunneling parameters into a trained attitude deviation prediction model, and predicting the head attitude deviation result of the pipe jacking machine through the attitude deviation prediction model, wherein the head attitude deviation result is used to indicate the deviation between the actual axis of the head and the target axis of the pipe jacking machine, wherein the attitude deviation prediction model, during pre-training, uses a training sample set including at least a portion of extended training data, wherein the extended training data includes a first training sample lacking a true attitude deviation label, and attitude deviation pseudo-labels generated for the first training sample by a trained pseudo-label generation model; adjusting the stroke difference between the correction cylinders of the pipe jacking machine according to the head attitude deviation result; wherein the pseudo-label generation model is pre-trained based on the following steps. Exercise: Obtain continuous second-level tunneling data corresponding to the aforementioned pipe jacking machine and sparse real attitude deviation data recorded by the guidance system. The continuous second-level tunneling data includes tunneling parameters corresponding one-to-one with multiple time points, and the sparse real attitude deviation data includes multiple real attitude deviation values ​​corresponding one-to-one with multiple time points. The number of time points corresponding to the real attitude deviation values ​​is less than the number of time points corresponding to the aforementioned tunneling parameters. Perform time alignment and feature construction on the continuous second-level tunneling data and the aforementioned sparse real attitude deviation data to obtain an input feature set and a real label set. The input feature set includes a first training sample and a second training sample. The time point corresponding to the first training sample lacks a real attitude deviation value, and the second training sample corresponds one-to-one with the real attitude deviation labels in the real label set. Use the second training sample as training input and the real attitude deviation labels as training labels to train the aforementioned pseudo-label generation model.

[0076] Thus, during the control of the pipe jacking machine, the attitude deviation is not manually calculated for control. Instead, based on the tunneling parameters, a trained attitude deviation prediction model automatically calculates the corresponding head attitude deviation of the pipe jacking machine. Furthermore, during the training of this attitude deviation prediction model, the training sample set includes at least a portion of extended training data. This extended training data includes a first training sample lacking a true attitude deviation label, and attitude deviation pseudo-labels generated for the first training sample by a trained pseudo-label generation model. The pseudo-label generation model is trained using a second training sample from the sparse true attitude deviation data that corresponds one-to-one with the true attitude deviation label. This allows it to fully learn the initial mapping relationship between tunneling parameters and attitude deviations under limited true labels, thereby generating reliable attitude deviation pseudo-labels for the first training sample lacking a true attitude deviation label. In other words, during the training of this attitude deviation prediction model, the trained pseudo-label model can reasonably expand the attitude deviation data used as labels, thereby reasonably expanding the training samples of the attitude deviation prediction model, improving the training effect of the attitude deviation prediction model, and obtaining a usable attitude deviation prediction model. In this way, the attitude deviation of the pipe jacking machine can be automatically predicted based on the attitude deviation prediction model, thereby adjusting the stroke difference between the correction cylinders of the pipe jacking machine and realizing the control of the pipe jacking machine. This is beneficial to improving the accuracy of the correction control of the pipe jacking machine and enhancing the control effect of the pipe jacking machine.

[0077] Specifically, based on the pipe jacking machine control method provided in this application, the complementarity between continuous second-level tunneling data and sparse manual click deviation data can be utilized. A pseudo-label generation model can generate pseudo-labels for missing data points, thereby alleviating the problem of insufficient samples caused by the scarcity of real labels. By fusing real labels and pseudo-labels to expand the training samples of the attitude deviation prediction model, the learning ability of the attitude deviation prediction model for continuous construction processes can be improved, thereby enhancing the continuity and accuracy of the pipe jacking machine head attitude deviation prediction.

[0078] Meanwhile, in this embodiment of the application, incorporating geological information into the attitude prediction modeling process can enhance the model's ability to characterize the nose attitude change patterns under different geological conditions and improve the adaptability of the attitude deviation prediction model under complex working conditions.

[0079] Thus, by learning the mapping relationship between the stroke parameters of the correction cylinder and the attitude deviation of the machine head based on the attitude deviation prediction model, and by adjusting the stroke difference between the correction cylinders online to control the attitude of the pipe jacking machine, it is possible to reduce the dependence on manual experience adjustment and improve the level of automation control in pipe jacking construction.

[0080] like Figure 3As shown in the figure, corresponding to the above-described pipe jacking machine control method, this application embodiment also provides a pipe jacking machine control system, which includes: The data acquisition module 310 is used to acquire the tunneling parameters corresponding to the pipe jacking machine, wherein the tunneling parameters include the stroke parameters of the correction cylinder; The deviation prediction module 320 is used to input the above-mentioned tunneling parameters into the trained attitude deviation prediction model, and to predict the attitude deviation result of the tunnel jacking machine corresponding to the machine head through the attitude deviation prediction model. The attitude deviation result of the machine head is used to indicate the deviation between the actual axis of the machine head and the target axis of the tunnel jacking machine. When the attitude deviation prediction model is pre-trained, the training sample set used includes at least a part of the extended training data. The extended training data includes a first training sample that lacks a real attitude deviation label, and attitude deviation pseudo labels generated for the first training sample by the trained pseudo label generation model. The control module 330 is used to adjust the stroke difference between the correction cylinders of the pipe jacking machine according to the above-mentioned head posture deviation results.

[0081] The pseudo-label generation model mentioned above is pre-trained based on the following steps: The continuous second-level tunneling data corresponding to the above-mentioned pipe jacking machine and the sparse real attitude deviation data recorded by the guidance system are obtained. The continuous second-level tunneling data includes tunneling parameters that correspond one-to-one with multiple time points, and the sparse real attitude deviation data includes multiple real attitude deviation values ​​that correspond one-to-one with multiple time points. The number of time points corresponding to the real attitude deviation values ​​is less than the number of time points corresponding to the tunneling parameters. Time alignment and feature construction are performed on the above continuous second-level tunneling data and the above sparse real attitude deviation data to obtain an input feature set and a real label set. The above input feature set includes a first training sample and a second training sample. The time corresponding to the first training sample is missing the real attitude deviation value. The second training sample corresponds one-to-one with the real attitude deviation label in the above real label set. The pseudo-label generation model is trained by using the second training sample as the training input and the real pose deviation label as the training label.

[0082] The attitude deviation prediction model does not directly drive the correction cylinder; its output can be used as the input of the control module 330 to further determine the stroke difference adjustment amount.

[0083] Thus, during the control of the pipe jacking machine, the attitude deviation is not manually calculated for control. Instead, based on the tunneling parameters, a trained attitude deviation prediction model automatically calculates the corresponding head attitude deviation of the pipe jacking machine. Furthermore, during the training of this attitude deviation prediction model, the training sample set includes at least a portion of extended training data. This extended training data includes a first training sample lacking a true attitude deviation label, and attitude deviation pseudo-labels generated for the first training sample by a trained pseudo-label generation model. The pseudo-label generation model is trained using a second training sample from the sparse true attitude deviation data that corresponds one-to-one with the true attitude deviation label. This allows it to fully learn the initial mapping relationship between tunneling parameters and attitude deviations under limited true labels, thereby generating reliable attitude deviation pseudo-labels for the first training sample lacking a true attitude deviation label. In other words, during the training of this attitude deviation prediction model, the trained pseudo-label model can reasonably expand the attitude deviation data used as labels, thereby reasonably expanding the training samples of the attitude deviation prediction model, improving the training effect of the attitude deviation prediction model, and obtaining a usable attitude deviation prediction model. In this way, the attitude deviation of the pipe jacking machine can be automatically predicted based on the attitude deviation prediction model, thereby adjusting the stroke difference between the correction cylinders of the pipe jacking machine and realizing the control of the pipe jacking machine. This is beneficial to improving the accuracy of the correction control of the pipe jacking machine and enhancing the control effect of the pipe jacking machine.

[0084] It should be noted that the specific structure and implementation of the above-mentioned pipe jacking machine control system and its various modules or units can be referred to the corresponding descriptions in the above method embodiments, and will not be repeated here.

[0085] It should be noted that the division of the various modules of the above-mentioned pipe jacking machine control system is not unique and is not intended as a specific limitation.

[0086] Based on the above embodiments, this application also provides a terminal, the principle block diagram of which can be as follows: Figure 4 As shown. The aforementioned terminal includes a processor, memory, network interface, and display screen connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps of any of the aforementioned pipe jacking machine control methods. The display screen can be a liquid crystal display (LCD) or an e-ink display.

[0087] Those skilled in the art will understand that Figure 4The block diagram shown is only a partial structural diagram related to the solution of this application and does not constitute a limitation on the terminal on which the solution of this application is applied. The specific terminal may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.

[0088] In one embodiment, a terminal is provided, the terminal including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of any of the pipe jacking machine control methods provided in the embodiments of this application.

[0089] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the pipe jacking machine control methods provided in this application.

[0090] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0091] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the above device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0092] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0093] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0094] In the embodiments provided in this application, it should be understood that the disclosed systems / terminal devices and methods can be implemented in other ways. For example, the system / terminal device embodiments described above are merely illustrative. For instance, the division of modules or units described above is merely a logical functional division, and in actual implementation, it can be divided in other ways. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0095] If the integrated modules / units described above are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, and software distribution media, etc. It should be noted that the content included in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0096] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions are not in essence a departure from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A control method for a pipe jacking machine, characterized in that, The method includes: Obtain the tunneling parameters corresponding to the pipe jacking machine, wherein the tunneling parameters include the stroke parameters of the correction cylinder; The tunneling parameters are input into a trained attitude deviation prediction model, and the attitude deviation result of the tunnel jacking machine is predicted by the attitude deviation prediction model. The attitude deviation result is used to indicate the deviation between the actual axis of the tunnel jacking machine and the target axis. When the attitude deviation prediction model is pre-trained, the training sample set used includes at least a part of extended training data. The extended training data includes a first training sample that is missing a true attitude deviation label, and attitude deviation pseudo labels generated for the first training sample by a trained pseudo label generation model. Based on the head posture deviation results, adjust the stroke difference between the correction cylinders of the pipe jacking machine; The pseudo-label generation model is pre-trained based on the following steps: The continuous second-level tunneling data corresponding to the pipe jacking machine and the sparse real attitude deviation data recorded by the guidance system are obtained. The continuous second-level tunneling data includes tunneling parameters that correspond one-to-one with multiple time points. The sparse real attitude deviation data includes multiple real attitude deviation values ​​that correspond one-to-one with multiple time points. The number of time points corresponding to the real attitude deviation values ​​is less than the number of time points corresponding to the tunneling parameters. Time alignment and feature construction are performed on the continuous second-level tunneling data and the sparse real attitude deviation data to obtain an input feature set and a real label set. The input feature set includes a first training sample and a second training sample. The time corresponding to the first training sample is missing the real attitude deviation value. The second training sample corresponds one-to-one with the real attitude deviation label in the real label set. The pseudo-label generation model is trained by using the second training sample as the training input and the real pose deviation label as the training label.

2. The pipe jacking machine control method according to claim 1, characterized in that, The tunneling parameters also include at least one of the following: advance speed, jacking force of the jacking cylinder, cutterhead parameters, soil chamber parameters, advance status, construction status parameters, and geological information.

3. The pipe jacking machine control method according to claim 2, characterized in that, The geological information includes at least one of the following: stratum type, soil parameters, burial depth information, groundwater information, and geological stratification information corresponding to the construction area where the pipe jacking machine is located.

4. The pipe jacking machine control method according to claim 1, characterized in that, The attitude deviation prediction model is pre-trained based on the following steps: The first training sample is input into the trained pseudo-label generation model, and the pseudo-label generation model generates the pose deviation pseudo-label corresponding to the first training sample. The first training sample and the second training sample are used as model inputs, and the pseudo-label of the posture deviation and the real posture deviation label are used as training labels to train the posture deviation prediction model.

5. The pipe jacking machine control method according to any one of claims 1 to 4, characterized in that, The pseudo-label generation model can be any one of the support vector regression model, tree model, or ensemble regression model. The attitude deviation prediction model is any one of the following: multilayer perceptron model, recurrent neural network model, gated recurrent unit model, or temporal convolutional network model.

6. The pipe jacking machine control method according to any one of claims 1 to 4, characterized in that, The nose attitude deviation results include the nose horizontal deviation and the nose vertical deviation; The step of adjusting the stroke difference between the correction cylinders of the pipe jacking machine based on the head posture deviation result includes: Adjust the stroke difference between the horizontally arranged correction cylinders of the pipe jacking machine according to the horizontal deviation of the machine head; Based on the vertical deviation of the machine head, adjust the stroke difference between the correction cylinders arranged vertically in the jacking machine.

7. A pipe jacking machine control system, characterized in that, The system includes: The data acquisition module is used to acquire the tunneling parameters corresponding to the pipe jacking machine, wherein the tunneling parameters include the stroke parameters of the correction cylinder; The deviation prediction module is used to input the tunneling parameters into a trained attitude deviation prediction model, and to predict the attitude deviation result of the tunnel jacking machine corresponding to the machine head through the attitude deviation prediction model. The attitude deviation result of the machine head is used to indicate the deviation between the actual axis of the tunnel jacking machine and the target axis. When the attitude deviation prediction model is pre-trained, the training sample set used includes at least a part of the extended training data. The extended training data includes a first training sample that is missing the real attitude deviation label, and attitude deviation pseudo labels generated for the first training sample by a trained pseudo label generation model. The control module is used to adjust the stroke difference between the correction cylinders of the pipe jacking machine according to the head posture deviation result; The pseudo-label generation model is pre-trained based on the following steps: The continuous second-level tunneling data corresponding to the pipe jacking machine and the sparse real attitude deviation data recorded by the guidance system are obtained. The continuous second-level tunneling data includes tunneling parameters that correspond one-to-one with multiple time points. The sparse real attitude deviation data includes multiple real attitude deviation values ​​that correspond one-to-one with multiple time points. The number of time points corresponding to the real attitude deviation values ​​is less than the number of time points corresponding to the tunneling parameters. Time alignment and feature construction are performed on the continuous second-level tunneling data and the sparse real attitude deviation data to obtain an input feature set and a real label set. The input feature set includes a first training sample and a second training sample. The time corresponding to the first training sample is missing the real attitude deviation value. The second training sample corresponds one-to-one with the real attitude deviation label in the real label set. The pseudo-label generation model is trained by using the second training sample as the training input and the real pose deviation label as the training label.

8. A terminal, characterized in that, The terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When executed by the processor, the computer program implements the steps of the pipe jacking machine control method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the pipe jacking machine control method as described in any one of claims 1 to 6.

Images