Tunneling machine deviation correction and direction control method, device, electronic equipment and storage medium

Abstract

The present application discloses a method for controlling deviation correction and steering of a roadheader, the method comprising: obtaining historical excavation data of the roadheader, and using the historical excavation data of the roadheader to train a pre-trained model for deviation correction and steering parameters of the roadheader; using target excavation data to transfer learning the pre-trained model for deviation correction and steering parameters of the roadheader, and obtaining a prediction model for deviation correction and steering parameters of the roadheader; obtaining real-time excavation data of the current project, and inputting the real-time excavation data into the prediction model for deviation correction and steering parameters of the roadheader, and obtaining a prediction value for deviation correction and steering parameters; performing parameter correction on the prediction value for deviation correction and steering parameters according to the steering posture of the roadheader and the maximum movement of the edge cutter, and using the corrected prediction value for deviation correction and steering parameters to control the deviation correction and steering of the roadheader, the present application can improve the accuracy and efficiency of deviation correction and steering control of the roadheader. The present application also discloses a control device for deviation correction and steering of a roadheader, a storage medium, and an electronic device, which have the above beneficial effects.

Description

Technical Field

[0001] The present application relates to the technical field of roadheader control, and in particular to a roadheader deviation correction and direction control method, device, electronic equipment and storage medium. Background Art

[0002] A tunnel boring machine (such as a full-face hard rock tunnel boring machine (TBM)) is a large-scale construction equipment that integrates mechanical, electrical, hydraulic, optical, and gas systems. It can be used for construction processes such as tunneling, support, and slag discharge and can perform continuous operations. It has the advantages of fast tunneling speed, environmental protection, and high comprehensive benefits. Its application in tunnel projects such as railways, hydropower, transportation, mining, and municipal engineering has grown rapidly.

[0003] To ensure the quality of tunnel construction, the tunnel boring machine must excavate strictly according to the set target axis. However, in the actual construction process, the tunnel boring machine will inevitably be affected by its own weight, geological conditions, human factors, etc., resulting in changes in the excavation posture, deviation from the target axis, position deviation and angle deviation, and it is necessary to correct the deviation and adjust the direction to return it to the target trajectory.

[0004] Existing tunnel boring machine construction mostly relies on manual control. The tunnel boring machine is turned left and right by adjusting the support shoe cylinder to move left and right, and the tunnel boring machine is turned up and down by adjusting the torque cylinder to extend and retract at the same time. The operating accuracy is greatly affected by the personnel. Improper operation can easily cause a series of problems such as damage to the cutterhead, tool, main drive and uneven tunnel wall, making it difficult to guarantee the construction quality.

[0005] How to improve the deviation correction and direction adjustment control accuracy and efficiency of the roadheader is a technical problem that technical personnel in this field currently need to solve. Summary of the invention

[0006] The purpose of the present application is to provide a method for controlling the deviation and direction of a tunnel boring machine, a device for controlling the deviation and direction of a tunnel boring machine, a storage medium and an electronic device, which can improve the accuracy and efficiency of the deviation and direction of the tunnel boring machine.

[0007] In order to solve the above technical problems, the present application provides a roadheader deviation correction and steering control method, the roadheader deviation correction and steering control method comprising:

[0008] Acquire historical excavation data of the roadheader, and use the historical excavation data of the roadheader to train a pre-training model of deviation correction and direction adjustment parameters of the roadheader;

[0009] Obtain target excavation data of the current project, and use the target excavation data to perform transfer learning on the pre-trained model of the deviation correction and steering parameters of the tunnel boring machine to obtain a prediction model for the deviation correction and steering parameters of the tunnel boring machine;

[0010] Acquire real-time excavation data of the current project, and input the real-time excavation data into the prediction model of the deviation correction and direction adjustment parameters of the tunnel boring machine to obtain the prediction value of the deviation correction and direction adjustment parameters;

[0011] The deviation correction and steering parameter prediction value is corrected according to the steering posture of the tunnel boring machine and the maximum movement of the edge cutter, and the deviation correction and steering control of the tunnel boring machine is carried out using the corrected deviation correction and steering parameter prediction value.

[0012] Optionally, using the historical excavation data of the roadheader to train a pre-trained model of deviation correction and direction adjustment parameters of the roadheader includes:

[0013] Cleaning the historical excavation data of the roadheader and extracting feature data;

[0014] Establishing a feature data group according to the feature data, and arranging the feature data group in chronological order;

[0015] Constructing the pre-training model of the deviation correction and steering parameters of the roadheader based on the deep learning neural network; wherein the pre-training model of the deviation correction and steering parameters of the roadheader comprises an input layer, an LSTM layer, a fully connected layer and an output layer;

[0016] The feature data group arranged in chronological order is used to train the tunnel boring machine deviation correction and steering parameter pre-training model.

[0017] Optionally, the real-time excavation data is input into the prediction model of the deviation correction and steering parameter of the tunnel boring machine to obtain the prediction value of the deviation correction and steering parameter, including:

[0018] The real-time tunneling data is input into the tunneling machine deviation correction and steering parameter prediction model to obtain the predicted value of the gripper cylinder action stroke and the predicted value of the torque cylinder action stroke; wherein the predicted value of the deviation correction and steering parameter includes the predicted value of the gripper cylinder action stroke and the predicted value of the torque cylinder action stroke.

[0019] Optionally, the parameter correction of the deviation correction and steering parameter prediction value according to the steering posture of the tunnel boring machine and the maximum edge cutter movement includes:

[0020] Correcting the predicted value of the action stroke of the gripper cylinder according to the horizontal orientation posture of the roadheader and the maximum movement amount of the edge cutter;

[0021] The predicted value of the torque cylinder action stroke is corrected according to the vertical adjustment posture of the tunnel boring machine and the maximum edge cutter movement.

[0022] Optionally, the predicted value of the action stroke of the gripper cylinder is corrected according to the horizontal orientation posture of the roadheader and the maximum edge cutter movement, including:

[0023] Calculating a horizontal deviation correction angle according to the horizontal orientation posture of the roadheader;

[0024] Calculating the maximum horizontal deviation correction angle of a single adjustment according to the maximum edge cutter movement amount;

[0025] Determining whether the horizontal correction angle is greater than the maximum horizontal correction angle;

[0026] If so, the predicted value of the action stroke of the shoe support cylinder is corrected using the maximum horizontal correction angle.

[0027] Optionally, the predicted value of the torque cylinder action stroke is corrected according to the vertical adjustment posture of the tunnel boring machine and the maximum edge cutter movement, including:

[0028] Calculating a vertical deviation correction angle according to the vertical orientation posture of the tunnel boring machine;

[0029] Calculating the maximum vertical deviation correction angle of a single adjustment according to the maximum edge cutter movement amount;

[0030] Determining whether the vertical correction angle is greater than the maximum vertical correction angle;

[0031] If so, the maximum vertical correction angle is used to correct the predicted value of the torque cylinder action stroke.

[0032] Optionally, the method of using the corrected deviation correction and steering parameter prediction value to perform deviation correction and steering control on the roadheader includes:

[0033] Calculate the horizontal and vertical propulsion strokes based on the minimum turning radius of the roadheader and the corrected deviation correction and direction adjustment parameter prediction values;

[0034] Determining a total propulsion stroke according to the horizontal propulsion stroke and the vertical propulsion stroke;

[0035] The corrected deviation correction and steering parameter prediction value and the total propulsion stroke are used to perform deviation correction and steering control on the roadheader.

[0036] The present application also provides a deviation correction and direction adjustment control device for a roadheader, the device comprising:

[0037] A historical data training module is used to obtain the historical excavation data of the roadheader, and use the historical excavation data of the roadheader to train a pre-training model for the deviation correction and direction adjustment parameters of the roadheader;

[0038] A transfer learning module is used to obtain target excavation data of the current project, and use the target excavation data to perform transfer learning on the pre-trained model of the deviation correction and steering parameters of the tunnel boring machine to obtain a prediction model of the deviation correction and steering parameters of the tunnel boring machine;

[0039] A prediction value determination module is used to obtain real-time excavation data of the current project, and input the real-time excavation data into the prediction model of the deviation correction and direction adjustment parameters of the tunnel boring machine to obtain the prediction value of the deviation correction and direction adjustment parameters;

[0040] The deviation correction control module is used to perform parameter correction on the deviation correction and direction adjustment parameter prediction value according to the heading posture of the tunnel boring machine and the maximum edge cutter movement, and use the corrected deviation correction and direction adjustment parameter prediction value to perform deviation correction and direction adjustment control on the tunnel boring machine.

[0041] The present application also provides a storage medium on which a computer program is stored, and when the computer program is executed, the steps of the above-mentioned tunnel boring machine deviation correction and direction adjustment control method are implemented.

[0042] The present application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, the steps of the above-mentioned tunnel boring machine deviation correction and direction adjustment control method are implemented.

[0043] The present application provides a method for controlling a deviation correction and direction turning of a roadheader, comprising: obtaining historical excavation data of the roadheader, and using the historical excavation data of the roadheader to train a pre-trained model for deviation correction and direction turning parameters of the roadheader; obtaining target excavation data of the current project, and using the target excavation data to perform transfer learning on the pre-trained model for deviation correction and direction turning parameters of the roadheader to obtain a prediction model for deviation correction and direction turning parameters of the roadheader; obtaining real-time excavation data of the current project, and inputting the real-time excavation data into the prediction model for deviation correction and direction turning parameters of the roadheader to obtain prediction values ​​of deviation correction and direction turning parameters; performing parameter correction on the prediction values ​​of deviation correction and direction turning parameters according to the steering posture of the roadheader and the maximum movement of the edge cutter, and using the corrected prediction values ​​of deviation correction and direction turning parameters to control the deviation correction and direction turning of the roadheader.

[0044] The present application uses the historical excavation data of the tunnel boring machine to train the pre-trained model of the tunnel boring machine's deviation correction and steering parameters, and fine-tunes the pre-trained model of the tunnel boring machine's deviation correction and steering parameters through transfer learning and the target excavation data of the current project, thereby constructing a tunnel boring machine's deviation correction and steering parameter prediction model that is more suitable for the current new project. Inputting real-time excavation data into the tunnel boring machine's deviation correction and steering parameter prediction model can obtain the deviation correction and steering parameter prediction value. The present application uses the tunnel boring machine's steering posture and the maximum edge cutter movement to perform parameter correction on the deviation correction and steering parameter prediction value, and then uses the corrected deviation correction and steering parameter prediction value to perform tunnel boring machine deviation correction and steering control. The above process can realize automatic deviation correction and adjustment of the tunnel boring machine without manual participation, and can improve the accuracy and efficiency of the tunnel boring machine's deviation correction and steering control. The present application also provides a tunnel boring machine deviation correction and steering control device, a storage medium and an electronic device, which have the above-mentioned beneficial effects and are not repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 A flow chart of a method for controlling deviation correction and direction adjustment of a roadheader provided in an embodiment of the present application;

[0047] Figure 2 An automatic deviation correction and direction adjustment control flow chart provided in an embodiment of the present application;

[0048] Figure 3 A schematic diagram of a TBM deviation correction and steering parameter pre-training model provided in an embodiment of the present application;

[0049] Figure 4 A schematic diagram of a transfer learning process provided in an embodiment of the present application;

[0050] Figure 5 A schematic diagram of a horizontal adjustment posture provided in an embodiment of the present application;

[0051] Figure 6 A schematic diagram of a vertical adjustment posture provided in an embodiment of the present application;

[0052] Figure 7 A schematic diagram of the horizontal and vertical directions of a TBM turning provided in an embodiment of the present application;

[0053] Figure 8 A schematic diagram of input and output of deviation correction and direction adjustment control parameters for a roadheader provided in an embodiment of the present application;

[0054] Fig. 9 A schematic structural diagram of a deviation correction and direction adjustment control device for a tunnel boring machine provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0056] See below Figure 1 , Figure 1 A flow chart of a method for controlling deviation correction and direction adjustment of a tunnel boring machine provided in an embodiment of the present application.

[0057] Specific steps may include:

[0058] S101: Acquire historical excavation data of a roadheader, and use the historical excavation data of the roadheader to train a pre-training model for deviation correction and direction adjustment parameters of the roadheader;

[0059] The present embodiment is applied to a tunnel boring machine control device, and the tunnel boring machine may be a full-face hard rock tunnel boring machine (TBM). In this step, the historical tunnel boring machine data may be obtained according to the tunnel boring machine operation log, and the historical tunnel boring machine data may include various tunnel boring machine parameters in a historical period, such as thrust, thrust speed, cutter head speed, cutter head torque, horizontal deviation, vertical deviation, heading angle, and pitch angle.

[0060] Specifically, the present embodiment can train the tunnel boring machine correction and steering parameter pre-training model in the following manner: perform data cleaning on the tunnel boring machine's historical tunneling data and extract feature data; establish a feature data group based on the feature data, and arrange the feature data group in chronological order; construct the tunnel boring machine correction and steering parameter pre-training model based on a deep learning neural network; wherein the tunnel boring machine correction and steering parameter pre-training model includes an input layer, an LSTM layer, a fully connected layer and an output layer; and use the feature data group arranged in chronological order to train the tunnel boring machine correction and steering parameter pre-training model.

[0061] S102: Acquire target excavation data of the current project, and use the target excavation data to perform transfer learning on the pre-trained model of the deviation correction and steering parameters of the tunnel boring machine to obtain a prediction model of the deviation correction and steering parameters of the tunnel boring machine;

[0062] Among them, after obtaining the pre-trained model of the tunnel boring machine's deviation correction and steering parameters, this embodiment can establish a tunnel boring machine's deviation correction and steering parameter prediction model based on transfer learning. In order to make the tunnel boring machine's deviation correction and steering parameter prediction model more suitable for the current project, this embodiment uses the target tunneling data to transfer learning the tunnel boring machine's deviation correction and steering parameter pre-trained model, so as to fine-tune the tunnel boring machine's deviation correction and steering parameter pre-trained model and obtain the tunnel boring machine's deviation correction and steering parameter prediction model. The above method can improve the accuracy of model prediction and solve the problem of model universality between different projects.

[0063] S103: Acquire real-time excavation data of the current project, and input the real-time excavation data into the prediction model of deviation correction and steering parameters of the roadheader to obtain prediction values ​​of deviation correction and steering parameters;

[0064] Among them, this step is based on the tunnel boring machine correction and steering parameter prediction model that has been obtained. This step obtains the real-time tunneling data of the current project (i.e., the current tunneling data). After the real-time tunneling data is input into the tunnel boring machine correction and steering parameter prediction model, the tunnel boring machine correction and steering parameter prediction model can output the corresponding correction and steering parameter prediction values.

[0065] Specifically, this embodiment can input the real-time excavation data into the tunnel boring machine correction and steering parameter prediction model to obtain the support shoe cylinder action stroke prediction value and the torque cylinder action stroke prediction value; the above-mentioned correction and steering parameter prediction value includes the support shoe cylinder action stroke prediction value and the torque cylinder action stroke prediction value.

[0066] S104: Correcting the deviation correction and steering parameter prediction value according to the steering posture of the roadheader and the maximum movement of the edge cutter, and controlling the deviation correction and steering of the roadheader using the corrected deviation correction and steering parameter prediction value.

[0067] Among them, in order to improve the control accuracy, this embodiment determines whether the predicted value of the correction and steering parameter needs to be corrected; if so, the predicted value of the correction and steering parameter is corrected according to the steering posture of the tunnel boring machine and the maximum edge cutter movement, and then the corrected predicted value of the correction and steering parameter is used to control the tunnel boring machine's correction and steering; if not, the predicted value of the correction and steering parameter calculated by S103 is used to control the tunnel boring machine's correction and steering.

[0068] Specifically, the process of using the corrected correction and steering parameter prediction values ​​to control the tunnel boring machine in this embodiment includes: calculating the horizontal propulsion stroke and the vertical propulsion stroke according to the minimum turning radius of the tunnel boring machine and the corrected correction and steering parameter prediction values; determining the total propulsion stroke according to the horizontal propulsion stroke and the vertical propulsion stroke; and using the corrected correction and steering parameter prediction values ​​and the total propulsion stroke to control the tunnel boring machine.

[0069] This embodiment uses the historical excavation data of the tunnel boring machine to train the tunnel boring machine correction and steering parameter pre-training model, and fine-tunes the tunnel boring machine correction and steering parameter pre-training model through transfer learning and the target excavation data of the current project, thereby constructing a tunnel boring machine correction and steering parameter prediction model that is more suitable for the current new project. Inputting real-time excavation data into the tunnel boring machine correction and steering parameter prediction model can obtain correction and steering parameter prediction values. This embodiment uses the tunnel boring machine steering posture and maximum edge cutter movement to perform parameter correction on the correction and steering parameter prediction values, and then uses the corrected correction and steering parameter prediction values ​​to perform tunnel boring machine correction and steering control. The above process can realize automatic correction and adjustment of the tunnel boring machine without human intervention, and can improve the accuracy and efficiency of the tunnel boring machine's correction and steering control.

[0070] As for Figure 1Further introduction to the corresponding embodiment, this embodiment can correct the predicted value of the shoe cylinder action stroke according to the horizontal steering posture of the tunnel boring machine and the maximum edge cutter movement. The specific process includes: calculating the horizontal correction angle according to the horizontal steering posture of the tunnel boring machine; calculating the maximum horizontal correction angle of a single adjustment according to the maximum edge cutter movement; judging whether the horizontal correction angle is greater than the maximum horizontal correction angle; if so, using the maximum horizontal correction angle to correct the predicted value of the shoe cylinder action stroke, so as to use the corrected predicted value of the shoe cylinder action stroke to perform tunnel boring machine correction and steering control; if not, using the uncorrected predicted value of the shoe cylinder action stroke to perform tunnel boring machine correction and steering control.

[0071] This embodiment can also correct the torque cylinder action stroke prediction value according to the vertical adjustment posture of the tunnel boring machine and the maximum edge cutter movement. The specific process includes: calculating the vertical correction angle according to the vertical adjustment posture of the tunnel boring machine; calculating the maximum vertical correction angle of a single adjustment according to the maximum edge cutter movement; judging whether the vertical correction angle is greater than the maximum vertical correction angle; if so, using the maximum vertical correction angle to correct the torque cylinder action stroke prediction value, so as to use the corrected torque cylinder action stroke prediction value to perform tunnel boring machine correction and direction control; if not, using the uncorrected torque cylinder action stroke prediction value to perform tunnel boring machine correction and direction control.

[0072] The process described in the above embodiment is explained below through an embodiment of correcting and steering control of a full-face hard rock tunnel boring machine TBM in actual application. This embodiment provides a TBM automatic correcting and steering control scheme. This scheme establishes a TBM correcting and steering parameter prediction model based on deep transfer learning. The model inputs are tunneling parameters such as thrust, thrust speed, cutter head speed, cutter head torque, horizontal deviation, vertical deviation, heading angle and pitch angle. The output is the predicted value of the action stroke of the support shoe cylinder and the predicted value of the action stroke of the torque cylinder used to control the TBM steering. This scheme corrects the predicted value of the steering parameter by analyzing the TBM steering posture model and combining the maximum edge cutter movement, and then plans the TBM correction trajectory based on the minimum turning radius, and controls the TBM to complete automatic correcting and steering.

[0073] See also Figure 2 , Figure 2An automatic deviation correction and steering control flow chart is provided in an embodiment of the present application, and the process includes: a historical data training stage, a deviation correction and steering parameter prediction stage, a deviation correction and steering parameter correction stage, and a deviation correction trajectory planning stage. In the historical data training stage, the TBM historical excavation data is used to perform deep learning neural network training to obtain a TBM deviation correction and steering parameter pre-training model. In the deviation correction and steering parameter prediction stage, the TBM real-time excavation data is obtained, and a TBM deviation correction and steering parameter prediction model is constructed through transfer learning to obtain the deviation correction and steering parameter prediction values: the support shoe cylinder action stroke △L and the torque cylinder action stroke △d. In the deviation correction and steering parameter correction stage, the steering correction angle θ and The correction values ​​of the deviation correction parameters △L and △d are obtained by using the large edge cutter movement constraint. In the deviation correction trajectory planning stage, the TBM deviation correction trajectory is planned based on the minimum turning radius to obtain the propulsion stroke S, and the TBM deviation correction and steering action is controlled according to the parameter correction values ​​△L and △d. It is judged whether the step steering is completed. If not, the TBM deviation correction and steering action is continued to be controlled. If so, the deviation correction and steering parameter prediction stage is entered.

[0074] This embodiment proposes a complete set of automatic control methods for TBM deviation correction and steering, including steering parameter prediction, steering parameter prediction value correction, trajectory planning, etc., which are highly practical and feasible. A TBM deviation correction and steering parameter prediction model is constructed based on the deep transfer learning method. The TBM historical excavation data is first trained by a deep learning neural network to obtain a TBM deviation correction and steering parameter pre-training model. Then, the TBM deviation correction and steering pre-training model is fine-tuned through transfer learning and a small amount of new project sample data, thereby constructing a TBM deviation correction and steering parameter prediction model that is more suitable for the current new project. The prediction accuracy is higher and the model is more targeted. This embodiment corrects the deviation correction and steering parameter prediction value by analyzing the TBM steering posture and based on the maximum edge cutter movement constraint, ensuring that the final steering control parameters are reliable and avoiding damage to the cutterhead and main drive and uneven tunnel walls. The deviation correction trajectory planning of this embodiment takes into account the horizontal and vertical directions, meets the minimum turning radius of the TBM and adopts a step-by-step steering method. The deviation correction and steering strategy is more in line with the TBM control law and is more practical. This embodiment may include the following steps:

[0075] Step 1: Build a TBM deviation correction and steering parameter pre-training model based on a deep learning neural network.

[0076] (1) In this step, the massive amount of historical TBM excavation data can be cleaned by using big data analysis methods to remove invalid data, useless fields, and duplicate data, and the following characteristic data can be screened out: thrust force F, thrust speed V, cutter head speed n, cutter head torque T, horizontal deviation e1, vertical deviation e2, heading angle α, pitch angle β, support shoe cylinder action stroke △L, and torque cylinder action stroke △d.

[0077] (2) Arrange the filtered feature data and establish the feature data set X = (F, V, n, T, e1, e2, α, β) T and Y = (△L, △d) T , and arrange the feature data groups in chronological order, that is, X t =(F t 、V t 、n t 、T t 、e1 t 、e2 t , α t , β t ) T , Y t =(△L t , △d t ) T , t is the number of feature data sets, t=1,2,3...,t.

[0078] (3) Based on deep learning neural network, a pre-training model for TBM deviation correction and steering parameters is designed. Figure 3 As shown in the figure, the TBM deviation correction and steering parameter pre-training model includes input layer, LSTM layer, fully connected layer and output layer. Set the number of LSTM layers to m, the number of fully connected layers to 2 (the number of layers can be modified), and the model input is: X t , X t-1 , X t-2 , …, X t-z ; z+1 is the number of nodes in the input layer, and the output of the TBM correction and orientation parameter pre-training model is: is the predicted value of the action stroke of the support shoe cylinder, It is the predicted value of the torque cylinder action stroke.

[0079] (4) The feature data set X t =(F t 、V t 、n t 、T t 、e1 t 、e2 t , α t , β t ) T and Y t =(△L t , △d t ) T , t=1,2,3...,t, are respectively used as the input samples and output samples of the TBM deviation correction and steering parameter pre-training model to train the model. According to the network structure of the TBM deviation correction and steering parameter pre-training model, it can be known that:

[0080] 1) The output of the LSTM layer is:

[0081]

[0082]

[0083] In the above formula is the output of the m-th layer LSTM network at time t, Sig(x) is the sigmoid activation function, and tanh(x) is the Tanh activation function. is the output of the mth layer cell at time t-1.

[0084] Among them, Wi t (m) ,Wo t (m) 、Wc t (m) 、Wf t (m) are the weight matrices of the input gate, output gate, Cell unit and forget gate of the m-th layer LSTM network at the current time t; Ui t (m) , Uo t (m) , Uc t (m) , Uf t (m) Respectively represent the weight matrices of the input gate, output gate, Cell unit and forget gate of the m-th layer LSTM network at the previous (t-1) moment; bi t (m) 、bo t (m) , bc t (m) 、bf t (m) The bias vector corresponding to the input gate, output gate, Cell unit and forget gate of the m-th layer LSTM network.

[0085]

[0086]

[0087] 2) The output of the fully connected layer is:

[0088]

[0089] g2=tanh(W2g1+b2).

[0090] Among them, g1 is the output of the first layer of the fully connected layer, W1 and b1 are the weight matrix and threshold between the output of the LSTM layer and the first layer of the fully connected layer respectively; g2 is the output of the second layer of the fully connected layer, W2 and b2 are the weight matrix and threshold between the first layer and the second layer of the fully connected layer respectively.

[0091] 3) The output of the model output layer is:

[0092] Among them, W3 and b3 are the weight matrix and threshold between the second layer of the fully connected layer and the output layer respectively; ReLU(x) is the output layer activation function; ReLU(x)=max(0,x).

[0093] 4) Input sample X t Input into the TBM deviation correction and steering pre-training model, the model output is Export the model Compared with the actual output sample Y t The error between is used as the objective function of model update, and the weight matrix of the model is updated through the back propagation algorithm, that is:

[0094]

[0095]

[0096] In the above formula, E is the objective function of the model update, J is the total objective function of the pre-trained model, and the weight update matrix is:

[0097]

[0098] In the above formula, η is the learning rate or learning operator. When ΔW≤ε, ε is greater than zero and small enough, the weight matrix update is stopped and the current optimal weight matrix is ​​obtained.

[0099] In this embodiment, step 4) can be repeated. When the training of t groups of input and output samples is completed, the training of the TBM deviation correction and steering parameter pre-training model is completed.

[0100] Step 2: Establish a TBM deviation correction and steering parameter prediction model based on transfer learning.

[0101] (1) Obtain a small amount of excavation data for the current new project and organize it to obtain input sample X q =(F q 、V q 、n q 、T q 、e1 q 、e2 q , α q , β q ) T And the output sample Yq =(△L q , △d q ) T , q is the sample size, q=1,2,3...,q.

[0102] (2) Based on the transfer learning method, the pre-trained model of TBM deviation correction and steering parameters is fine-tuned using the input and output samples of the new project. That is, only the weight matrix of the training fully connected layer is updated and other layers are frozen. The new model obtained after training is the TBM deviation correction and steering parameter prediction model suitable for the current new project. Figure 4 , Figure 4 A schematic diagram of a transfer learning process provided in an embodiment of the present application.

[0103] The updated network connection weight matrix of the fully connected layer of the TBM deviation correction and steering parameter prediction model is:

[0104]

[0105] (3) When TBM excavation is carried out in a new project, the real-time excavation parameters X = (F, V, n, T, e1, e2, α, β) T By inputting them into the TBM deviation correction and steering parameter prediction model, the current TBM deviation correction and steering parameter prediction values ​​△L and △d can be obtained.

[0106] Step 3: Correct the predicted values ​​of the deviation correction and steering parameters based on the TBM steering posture analysis and the maximum edge cutter movement.

[0107] (1) Analyze the horizontal orientation of the TBM, such as Figure 5 As shown in the horizontal steering posture diagram, when the predicted value △L is known, the corresponding horizontal correction angle θ can be obtained as: L is the length of the main beam.

[0108] (2) Consider the maximum edge cutter movement a max The maximum horizontal correction angle θ that can be adjusted at a time is max for: r is the cutter head radius.

[0109] (3) Determine whether θ is greater than θ max ;

[0110] If θ≤θ max , then the stroke of the support shoe cylinder is the predicted value △L, and the horizontal correction angle is θ corresponding to △L; if θ>θ max , then take the horizontal correction angle θ=θ max , correct the predicted value of the action stroke of the support shoe cylinder △L, that is, at this time:

[0111] (4) Analyze the vertical orientation of the TBM, such as Figure 6 From the vertical steering posture diagram shown in FIG. 1 , it can be seen that when the predicted value △d is known, the corresponding vertical correction angle can be obtained: for:

[0112]

[0113] In the above formula, β is the current pitch angle of the TBM, and d is the extended stroke of the torque cylinder corresponding to the current pitch angle β;

[0114] (5) Consider the maximum edge cutter movement a max The maximum vertical correction angle that can be adjusted at a time is for:

[0115] (6)Judgment Is it greater than

[0116] like Then the stroke of the support shoe cylinder is the predicted value △d, and the vertical correction angle is the value corresponding to △d. like Then take the horizontal correction angle Correct the predicted value of the torque cylinder stroke △d, that is, at this time:

[0117] Step 4: Plan the deviation correction trajectory based on the minimum turning radius and obtain the propulsion stroke S after the adjustment.

[0118] (1) Plan the deviation correction trajectory based on the minimum turning radius R of the TBM. The schematic diagram of the TBM turning is as follows: Figure 7 As shown, Figure 7 The diagram is a schematic diagram of the horizontal and vertical directions of the TBM when turning. B represents the center position of the cutterhead before trajectory planning in the horizontal direction, A represents the center position of the cutterhead after trajectory planning in the horizontal direction, D represents the end position of the main beam before correction in the horizontal direction, and C represents the end position of the main beam after correction in the horizontal direction. The same is true in the vertical direction. B' represents the center position of the cutterhead before trajectory planning in the vertical direction, A' represents the center position of the cutterhead after trajectory planning in the vertical direction, D' represents the end position of the main beam before correction in the vertical direction, and C' represents the end position of the main beam after correction in the vertical direction.

[0119] (2) Based on Figure 7 From the geometric relationship in , we can obtain the TBM's horizontal and vertical travel distances X and Y respectively:

[0120]

[0121]

[0122] (3) Calculate the backward propulsion stroke S as:

[0123] (4) △L, △d and S are used as TBM deviation correction and adjustment control values ​​to control the TBM to perform deviation correction and adjustment actions.

[0124] This embodiment proposes a TBM automatic deviation correction and steering control method, which reduces manual operations and improves the automation level and construction efficiency. On the basis of the TBM deviation correction and steering parameter pre-training model constructed by deep learning neural network, the TBM deviation correction and steering parameter pre-training model is fine-tuned by transfer learning method, thereby obtaining a TBM deviation correction and steering parameter prediction model that is more suitable for the current new project, further improving the accuracy of model prediction, and at the same time solving the problem of model universality between different projects. This embodiment corrects the deviation correction and steering parameter prediction value based on the TBM steering posture and the maximum edge cutter movement, thereby protecting the cutterhead, main drive, etc. from damage, extending the service life of the TBM, and at the same time ensuring the smoothness of the construction tunnel wall and improving the construction quality. This embodiment considers both the horizontal and vertical directions in the deviation correction trajectory planning, and plans the excavation process of the TBM during a single steering based on the deviation correction and steering parameter prediction correction value and the minimum turning radius. The step-by-step steering method is adopted, which complies with the TBM deviation correction and steering control law, and improves the practicality and reliability of the TBM automatic deviation correction and steering control method. Please refer to Figure 8 , Figure 8 A schematic diagram of input and output of a tunnel boring machine deviation correction and steering control parameter provided in an embodiment of the present application is firstly to establish a TBM deviation correction and steering parameter prediction model based on a deep transfer neural network. The model inputs are tunneling parameters such as thrust force F, thrust speed V, cutter head speed n, cutter head torque T, horizontal deviation, vertical deviation, heading angle and pitch angle, and the outputs are the predicted values ​​of the action stroke of the support shoe cylinder and the predicted values ​​of the action stroke of the torque cylinder used to control the TBM steering. The above-mentioned horizontal deviation and vertical deviation are determined according to the target axis. The steering parameter prediction value is corrected by analyzing the TBM steering posture model and combining the maximum edge cutter movement. The TBM deviation correction trajectory is planned in combination with the predicted parameter correction value and the minimum turning radius, so as to obtain the final control parameters (i.e., control values) for controlling the TBM deviation correction and steering. The final control parameters are input into the TBM controller so that the TBM deviation correction and steering system can adjust the TBM posture.

[0125] See also Fig. 9 , Fig. 9 A schematic diagram of the structure of a deviation correction and direction adjustment control device for a roadheader provided in an embodiment of the present application;

[0126] The device may include:

[0127] The historical data training module 901 is used to obtain the historical excavation data of the roadheader, and use the historical excavation data of the roadheader to train the pre-training model of the roadheader deviation correction and direction adjustment parameters;

[0128] The transfer learning module 902 is used to obtain the target excavation data of the current project, and use the target excavation data to perform transfer learning on the pre-trained model of the deviation correction and steering parameters of the tunnel boring machine to obtain a prediction model of the deviation correction and steering parameters of the tunnel boring machine;

[0129] The prediction value determination module 903 is used to obtain the real-time excavation data of the current project, and input the real-time excavation data into the prediction model of the deviation correction and steering parameter of the tunnel boring machine to obtain the prediction value of the deviation correction and steering parameter;

[0130] The deviation correction control module 904 is used to perform parameter correction on the deviation correction and steering parameter prediction value according to the steering posture of the tunnel boring machine and the maximum edge cutter movement, and use the corrected deviation correction and steering parameter prediction value to perform deviation correction and steering control on the tunnel boring machine.

[0131] This embodiment uses the historical excavation data of the tunnel boring machine to train the tunnel boring machine correction and steering parameter pre-training model, and fine-tunes the tunnel boring machine correction and steering parameter pre-training model through transfer learning and the target excavation data of the current project, thereby constructing a tunnel boring machine correction and steering parameter prediction model that is more suitable for the current new project. Inputting real-time excavation data into the tunnel boring machine correction and steering parameter prediction model can obtain correction and steering parameter prediction values. This embodiment uses the tunnel boring machine steering posture and maximum edge cutter movement to perform parameter correction on the correction and steering parameter prediction values, and then uses the corrected correction and steering parameter prediction values ​​to perform tunnel boring machine correction and steering control. The above process can realize automatic correction and adjustment of the tunnel boring machine without human intervention, and can improve the accuracy and efficiency of the tunnel boring machine's correction and steering control.

[0132] Furthermore, the historical data training module 901 is used to clean the historical excavation data of the tunnel boring machine and extract feature data; it is also used to establish a feature data group based on the feature data, and arrange the feature data group in chronological order; it is also used to construct the tunnel boring machine correction and steering parameter pre-training model based on the deep learning neural network; wherein the tunnel boring machine correction and steering parameter pre-training model includes an input layer, an LSTM layer, a fully connected layer and an output layer; it is also used to train the tunnel boring machine correction and steering parameter pre-training model using the feature data group arranged in chronological order.

[0133] Furthermore, the prediction value determination module 903 is used to input the real-time excavation data into the tunnel boring machine correction and steering parameter prediction model to obtain the support shoe cylinder action stroke prediction value and the torque cylinder action stroke prediction value; wherein, the correction and steering parameter prediction value includes the support shoe cylinder action stroke prediction value and the torque cylinder action stroke prediction value.

[0134] Furthermore, the deviation correction control module 904 is used to correct the predicted value of the action stroke of the support shoe cylinder according to the horizontal adjustment posture of the tunnel boring machine and the maximum edge cutter movement; and is also used to correct the predicted value of the action stroke of the torque cylinder according to the vertical adjustment posture of the tunnel boring machine and the maximum edge cutter movement.

[0135] Furthermore, the deviation correction control module 904 includes:

[0136] The first correction unit is used to calculate the horizontal correction angle according to the horizontal adjustment posture of the tunnel boring machine; it is also used to calculate the maximum horizontal correction angle of a single adjustment according to the maximum edge cutter movement; it is also used to determine whether the horizontal correction angle is greater than the maximum horizontal correction angle; if so, the maximum horizontal correction angle is used to correct the predicted value of the action stroke of the support shoe cylinder.

[0137] Furthermore, the deviation correction control module 904 includes:

[0138] The first correction unit is used to calculate the vertical correction angle according to the vertical adjustment posture of the tunnel boring machine; it is also used to calculate the maximum vertical correction angle of a single adjustment according to the maximum edge cutter movement; it is also used to determine whether the vertical correction angle is greater than the maximum vertical correction angle; if so, the maximum vertical correction angle is used to correct the predicted value of the torque cylinder action stroke.

[0139] Furthermore, the deviation correction control module 904 is used to calculate the horizontal propulsion stroke and the vertical propulsion stroke based on the minimum turning radius of the tunnel boring machine and the corrected deviation correction and steering parameter prediction values; it is also used to determine the total propulsion stroke based on the horizontal propulsion stroke and the vertical propulsion stroke; it is also used to perform deviation correction and steering control of the tunnel boring machine using the corrected deviation correction and steering parameter prediction values ​​and the total propulsion stroke.

[0140] Since the embodiments of the apparatus part correspond to the embodiments of the method part, please refer to the description of the embodiments of the method part for the embodiments of the apparatus part, which will not be repeated here.

[0141] The present application also provides a storage medium on which a computer program is stored, and when the computer program is executed, the steps provided in the above embodiment can be implemented. The storage medium may include: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0142] The present application also provides an electronic device, which may include a memory and a processor, wherein a computer program is stored in the memory, and when the processor calls the computer program in the memory, the steps provided in the above embodiment may be implemented. Of course, the electronic device may also include various network interfaces, power supplies and other components.

[0143] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of this application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of this application.

[0144] It should also be noted that, in this specification, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "comprising" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

Claims

1. A method for controlling deviation correction and direction adjustment of a roadheader, characterized in that: include: Acquire historical excavation data of the roadheader, and use the historical excavation data of the roadheader to train a pre-training model of deviation correction and direction adjustment parameters of the roadheader; Obtain target excavation data of the current project, and use the target excavation data to perform transfer learning on the pre-trained model of the deviation correction and steering parameters of the tunnel boring machine to obtain a prediction model for the deviation correction and steering parameters of the tunnel boring machine; Acquire real-time excavation data of the current project, and input the real-time excavation data into the prediction model of the deviation correction and direction adjustment parameters of the tunnel boring machine to obtain the prediction value of the deviation correction and direction adjustment parameters; The deviation correction and steering parameter prediction value is modified according to the steering posture of the roadheader and the maximum movement of the edge cutter, and the deviation correction and steering control of the roadheader is performed using the modified deviation correction and steering parameter prediction value; The real-time excavation data is input into the prediction model of the deviation correction and steering parameter of the tunnel boring machine to obtain the prediction value of the deviation correction and steering parameter, including: Input the real-time excavation data into the prediction model of the deviation correction and steering parameters of the tunnel boring machine to obtain the predicted value of the action stroke of the gripper cylinder and the predicted value of the action stroke of the torque cylinder; wherein the predicted value of the deviation correction and steering parameters includes the predicted value of the action stroke of the gripper cylinder and the predicted value of the action stroke of the torque cylinder; The step of performing parameter correction on the predicted value of the deviation correction and steering parameter according to the steering posture of the roadheader and the maximum edge cutter movement comprises: Correcting the predicted value of the action stroke of the gripper cylinder according to the horizontal orientation posture of the roadheader and the maximum movement amount of the edge cutter; The predicted value of the torque cylinder action stroke is corrected according to the vertical adjustment posture of the tunnel boring machine and the maximum edge cutter movement.

2. The method for controlling deviation correction and direction adjustment of a roadheader according to claim 1, characterized in that: Using the historical excavation data of the roadheader to train a pre-training model for the deviation correction and direction adjustment parameters of the roadheader includes: Cleaning the historical excavation data of the roadheader and extracting feature data; Establishing a feature data group according to the feature data, and arranging the feature data group in chronological order; Constructing the pre-training model of the deviation correction and steering parameters of the roadheader based on the deep learning neural network; wherein the pre-training model of the deviation correction and steering parameters of the roadheader comprises an input layer, an LSTM layer, a fully connected layer and an output layer; The feature data group arranged in chronological order is used to train the tunnel boring machine deviation correction and steering parameter pre-training model.

3. The method for controlling deviation correction and direction adjustment of a roadheader according to claim 1, characterized in that: The predicted value of the action stroke of the gripper cylinder is corrected according to the horizontal orientation posture of the roadheader and the maximum edge cutter movement, including: Calculating a horizontal deviation correction angle according to the horizontal orientation posture of the roadheader; Calculating the maximum horizontal deviation correction angle of a single adjustment according to the maximum edge cutter movement amount; Determining whether the horizontal correction angle is greater than the maximum horizontal correction angle; If so, the predicted value of the action stroke of the shoe support cylinder is corrected using the maximum horizontal correction angle.

4. The method for controlling deviation correction and direction adjustment of a roadheader according to claim 1, characterized in that: The method of correcting the predicted value of the torque cylinder action stroke according to the vertical adjustment posture of the roadheader and the maximum edge cutter movement amount includes: Calculating a vertical deviation correction angle according to the vertical orientation posture of the tunnel boring machine; Calculating the maximum vertical deviation correction angle of a single adjustment according to the maximum edge cutter movement amount; Determining whether the vertical correction angle is greater than the maximum vertical correction angle; If so, the maximum vertical correction angle is used to correct the predicted value of the torque cylinder action stroke.

5. The method for controlling deviation correction and direction adjustment of a roadheader according to any one of claims 1 to 4, characterized in that: The method of using the corrected deviation correction and steering parameter prediction value to control the roadheader deviation correction and steering comprises: Calculating the horizontal propulsion stroke and the vertical propulsion stroke according to the minimum turning radius of the roadheader and the corrected deviation correction and direction adjustment parameter prediction value; Determining a total propulsion stroke according to the horizontal propulsion stroke and the vertical propulsion stroke; The corrected deviation correction and steering parameter prediction value and the total propulsion stroke are used to perform deviation correction and steering control on the roadheader.

6. A deviation correction and direction control device for a roadheader, characterized in that: include: A historical data training module is used to obtain the historical excavation data of the roadheader, and use the historical excavation data of the roadheader to train a pre-training model for the deviation correction and direction adjustment parameters of the roadheader; A transfer learning module is used to obtain target excavation data of the current project, and use the target excavation data to perform transfer learning on the pre-trained model of the deviation correction and steering parameters of the tunnel boring machine to obtain a prediction model of the deviation correction and steering parameters of the tunnel boring machine; A prediction value determination module is used to obtain real-time excavation data of the current project, and input the real-time excavation data into the prediction model of the deviation correction and direction adjustment parameters of the tunnel boring machine to obtain the prediction value of the deviation correction and direction adjustment parameters; A deviation correction control module is used to perform parameter correction on the deviation correction and direction adjustment parameter prediction value according to the heading posture of the roadheader and the maximum edge cutter movement, and perform deviation correction and direction adjustment control on the roadheader using the corrected deviation correction and direction adjustment parameter prediction value; The prediction value determination module is used to input the real-time excavation data into the prediction model of the deviation correction and steering parameters of the tunnel boring machine to obtain the prediction value of the action stroke of the gripper cylinder and the prediction value of the action stroke of the torque cylinder; wherein the prediction value of the deviation correction and steering parameters includes the prediction value of the action stroke of the gripper cylinder and the prediction value of the action stroke of the torque cylinder; Among them, the correction control module is used to correct the predicted value of the shoe cylinder's motion stroke according to the horizontal adjustment posture of the tunnel boring machine and the maximum edge cutter movement; it is also used to correct the predicted value of the torque cylinder's motion stroke according to the vertical adjustment posture of the tunnel boring machine and the maximum edge cutter movement.

7. An electronic device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, the steps of the deviation correction and direction adjustment control method of the tunnel boring machine as claimed in any one of claims 1 to 5 are implemented.

8. A storage medium, characterized in that: The storage medium stores computer executable instructions, and when the computer executable instructions are loaded and executed by the processor, the steps of the roadheader deviation correction and direction control method as described in any one of claims 1 to 5 are implemented.

Images