A full-cycle management system for tunneling machine cutterhead hobs

Abstract

This application provides a full-cycle management system for tunneling machine cutterhead hobs, comprising: a hob status tracking module for real-time tracking and recording of the position and mileage of each hob using a calibration-then-tracking method; a voice control module for obtaining the current measured wear value of the corresponding hob or controlling the rotation angle of the cutterhead based on voice information; a hob wear prediction module for real-time prediction of the current wear value of the hob based on a preset prediction model; an anomaly warning module for determining whether abnormal wear has occurred in the cutterhead hobs based on the wear value and issuing a timely warning when abnormal wear occurs; and a data processing and storage module for receiving and recording the data output from each module, and real-time matching and storing the hob data with surrounding rock parameters, tunneling parameters, and cutter maintenance status in a database. This application realizes full-cycle management of the cutterhead hobs from parameter acquisition, data processing and storage, wear prediction, anomaly alarms to auxiliary control for cutter inspection and replacement.

Description

Technical Field

[0001] This application relates to the field of tunnel boring machine technology, and in particular, to a full-cycle management system for the cutterhead rollers of a tunnel boring machine. Background Technology

[0002] Full-face tunnel boring machines are large-scale tunnel construction equipment that integrates mechanical, electrical, hydraulic, optical, and pneumatic systems. They can be used for tunneling, support, muck removal, and other construction processes and can be operated continuously. They have advantages such as fast tunneling speed, environmental friendliness, and high comprehensive benefits. Their application in tunnel projects in China's railways, hydropower, transportation, mining, and municipal engineering is growing rapidly.

[0003] The cutterhead is a key component of the tunnel boring machine for rock breaking. As the tunneling cutter that directly contacts the rock strata on the cutterhead, the roller cutter is characterized by easy wear, easy consumption, difficult maintenance and replacement, and high cost. Therefore, whether the management of the cutterhead roller cutter is reasonable will directly affect the project cost and the tunnel construction progress. However, the current management of cutterhead rollers is chaotic and disorganized, specifically in the following ways: (1) When collecting roller data, due to the harsh environment for checking the cutter, workers need to check the cutter on-site first and then make paper records based on their memory. This can easily lead to incomplete or incorrect roller data records. Furthermore, a complete database has not been established, and the roller wear amount has not been matched with the tunneling mileage, tunneling parameters, and surrounding rock conditions. The information data is scattered and difficult to utilize. (2) The inspection cycle relies solely on the driver's experience and is greatly affected by the driver's skill level. If the inspection is done too early, it will waste resources. If the inspection is done too late, it will affect the construction progress. (3) For the prediction of the roller wear state, most existing prediction methods rely on historical data combined with neural networks for learning and training. However, the existing historical data of the rollers is unreliable, making it difficult to control accurately. (4) When checking and replacing the cutter, workers need to perform frequent jogging operations to control the cutterhead to rotate to the appropriate position. This cannot be done in one step, resulting in long inspection and replacement times and low efficiency.

[0004] In conclusion, it is crucial to develop a complete lifecycle management system for tunneling machine cutterhead hobs, enabling full-cycle management of the cutterhead hobs from parameter acquisition, data processing and storage, wear prediction, anomaly alarms to auxiliary control for cutter inspection and replacement. Summary of the Invention

[0005] To address the aforementioned technical problems, this application provides a full-cycle management system for tunneling machine cutterhead hobs.

[0006] The technical solution adopted in this application is as follows:

[0007] A tunneling machine cutterhead hob full-cycle management system includes:

[0008] The hob status tracking module is used to track and record the position and usage mileage of each hob in real time by calibrating first and then tracking, thereby controlling the real-time status of each hob.

[0009] The voice control module is used to obtain the current measured wear value of the hob or control the rotation angle of the cutter head based on voice information.

[0010] The hob wear prediction module is used to predict the current wear value of the hob in real time based on a preset prediction model.

[0011] The abnormal status early warning module is used to determine whether the cutter head hob has abnormal wear based on the wear value, and to issue an early warning in a timely manner when abnormal wear occurs;

[0012] The data processing and storage module is used to receive and record the data output by each module, and match the cutter data with the surrounding rock parameters, tunneling parameters, and cutter maintenance status in real time, and automatically generate forms to be stored in the database.

[0013] Furthermore, the hobbing cutter state tracking module specifically includes:

[0014] The initial position calibration module is used to first set the initial position of the cutter head according to convention, including the 0° position and rotation direction. Then, each hob is numbered, and the initial position of each hob is calibrated, including the initial angle θ. i and distance ψ from the center of the cutter head i , where i is the hob number;

[0015] The tracking and recording module is used to track and record the current status B of each hob in real time while the cutter head is running. i (δ i S i )in:

[0016] Current angle δ of the hob i :

[0017] δ i =MOD(MOD(α-β, 360)+θ i ,360)

[0018] Current mileage S of the hob i :

[0019] S i =2πψ i (α+β) / 360

[0020] Where α is the total forward rotation angle of the cutterhead from the initial tunneling to the current moment, β is the total reverse rotation angle of the cutterhead from the initial tunneling to the current moment, and MOD(x, y) is the remainder function.

[0021] Furthermore, the voice control module includes:

[0022] The voice recognition unit is used to recognize and process the voice control commands given by the worker to obtain the corresponding control parameters, including the hob number, the current radius of the hob, and the target angle.

[0023] The cutter head jog control box is used to control the cutter head to rotate by a corresponding angle when the control parameters recognized by the voice recognition unit include the word "rotate";

[0024] The controller is used to store the recognized radius value as the current radius of the corresponding numbered hob when the control parameters recognized by the voice recognition unit include the word "radius", and to calculate the current measured wear value of the corresponding numbered hob based on the initial radius of the corresponding numbered hob.

[0025] Furthermore, the cutter head jog control box is specifically used for:

[0026] When the control parameters recognized by the voice recognition unit include the word "rotation", the receiving voice recognition unit will identify the hob number i and the corresponding target angle. Automatic calculation of cutter head rotation angle And control the rotation of the cutter head.

[0027] Furthermore, when the cutter head jog control box controls the rotation of the cutter head, it controls the cutter head to rotate along the shortest path, wherein the shortest path is: if 0≤λ≤180, the cutter head is controlled to rotate clockwise by an angle λ; if 180<λ<360, the cutter head is controlled to rotate counterclockwise by an angle 360-λ; if -180≤λ≤0, the cutter head is controlled to rotate counterclockwise by an angle |λ|; if -360<λ<-180, the cutter head is controlled to rotate clockwise by an angle 360+λ.

[0028] Furthermore, the data processing and storage module includes:

[0029] The data receiving unit is used to receive and record the data output by the hob status tracking module, the voice control module, the hob wear prediction module, and the status anomaly early warning module.

[0030] The data processing unit matches and associates the cutter head data with the surrounding rock parameters, tunneling parameters, and cutter maintenance status in real time according to the time sequence, and automatically generates a form. The tunneling parameters include cutterhead rotation speed, propulsion speed, propulsion force, cutterhead torque, and penetration depth. The surrounding rock parameters include the surrounding rock grade. The cutter maintenance status includes maintenance status and maintenance reason.

[0031] Data storage unit, used to store the generated forms.

[0032] Furthermore, the automatically generated form content includes: hob number i, initial angle θ i Current angle δi ψ, distance from the center of the cutter head i Initial radius r i Current radius d i Mileage S i Measured wear value h i Wear prediction value k i , hobbing cutter state.

[0033] Furthermore, the hob wear prediction module includes:

[0034] The prediction model building module is used to build and train a model for predicting hob wear values;

[0035] The online prediction module is used to input the real-time collected feature information parameters into the hob wear value prediction model to perform online prediction and obtain the hob wear prediction value.

[0036] The offline optimization module of the prediction model retrieves newly generated historical data from the historical database of the cutterhead hob full-cycle management system at regular intervals during actual tunneling, using it as input samples to optimize the hob wear value prediction model offline. This makes the prediction model increasingly consistent with the current tunneling conditions, thereby further improving the prediction accuracy.

[0037] Furthermore, the prediction model building module is specifically used for:

[0038] Based on historical databases, model feature information data is selected, including: hob number and the corresponding cutter head rotation speed, feed speed, thrust, feed speed, penetration, surrounding rock grade, distance of hob from cutter head center, hob usage mileage, and measured hob wear value;

[0039] A hob wear prediction model is constructed based on a convolutional neural network. This model consists of four convolutional encoders and three fully connected layers. After processing the input data, the convolutional encoders and fully connected layers directly output the wear value Z = (Z1, Z2, ... Zn). i );

[0040] The feature information data is sequentially segmented and organized into a 2D matrix X×Y containing feature information and tool position information, which serves as the input to the model. The size of X depends on the length of the time series when organizing the data. The output is the tool wear amount Z = (Z1, Z2, ... Zn). i From the network structure, it can be seen that during forward propagation, the l-th layer convolutional encoder maps to L. l The result is:

[0041]

[0042] Where, x ij(l) The output of the l-th layer convolutional encoder is g(x), where g(x) is the activation function and b is the output of the l-th layer convolutional encoder. i (l) The threshold of the l-th layer convolutional encoder. This is a convolutional kernel in the convolutional encoder. This approach allows the network to efficiently describe complex interactions of multiple variables using only sparse interactions. Based on this, the computation results of the previous encoder layer are processed in the same way to the last convolutional encoder layer. The computation results of the last encoder layer are converted into a one-dimensional array and input to the fully connected layer. The fully connected layer consists of multiple neuron nodes. The output of the fully connected layer is:

[0043]

[0044] Where g(x) = max{0,x} is the activation function, b s l w is the hidden layer threshold. ab l The input layer weight matrix;

[0045] Find the error function of the hob wear prediction model:

[0046]

[0047] Among them, e q f is the expected value of the error function, i.e., the measured value of hob wear in the historical sample data. q The actual output of the error function is y, which is the output of the fully connected layer of the convolutional neural network. s ;

[0048] The backpropagation algorithm is used to optimize the parameters of the convolutional neural network and reduce the error. The partial derivatives with respect to the weights of the input layer and hidden layer are taken to obtain the weight update matrix.

[0049]

[0050]

[0051] In the formula, w i,j b represents the weights of the convolutional neural network. j η is the threshold of the convolutional neural network, and η is the learning rate of the neural network;

[0052] When △w i,j ≤ε and △b j When ≤ε, and ε is greater than zero and sufficiently small, the neural network model training ends, and the weight matrix w at this point is... i,j b j The optimal weight parameters are used as the learning parameters of the convolutional neural network to obtain the hob wear value prediction model.

[0053] Furthermore, the abnormal status early warning module includes:

[0054] The limit setting module is used to set the limit value for hob wear on the cutter head. max1 Hob wear limit value 2h max2 Thrust limit value F max Torque limit value T min Excavation speed limit value V min ;

[0055] The hob wear estimation calculation module is used to calculate the hob wear based on the measured value h obtained during the latest hob inspection. i (j) and its corresponding hob service mileage S i (j) Combined with the current hobbing mileage S i (t) Calculate the estimated wear value of the hob h gi (t):

[0056] h gi (t)=h i (j) *(S i (t)-S i (j) ) / S i (j) ;

[0057] The hob status anomaly judgment module is used to judge abnormal hob status conditions, including: prediction value anomaly judgment: if the current hob wear prediction value k i (t)>h max2 The abnormal status warning module will issue an alarm for abnormal predicted value of hob wear; the abnormal judgment of the measured value is as follows: if the predicted value of hob wear at the current time is k i (t)>h max1 And h i (j) +h gi (t)>h max2 The abnormal status warning module will issue an alarm indicating abnormal measured value of cutter wear; abnormal tunneling parameter judgment: if thrust F > F max And torque T <T max , or thrust F>F max And the propulsion speed V <V max If the abnormal status warning module issues an alarm indicating that the cutting tool wear has reached the required level and the tunneling parameters are abnormal, then the abnormal status warning module will issue an alarm.

[0058] Compared with the prior art, this application has the following advantages:

[0059] This application provides a full-cycle management system for the cutterhead hob of a tunneling machine, including a hob status tracking module, a voice control module, a data processing and storage module, a hob wear prediction module, and a status anomaly early warning module. Compared with the prior art, it has the following advantages:

[0060] 1. A full-cycle management system for tunneling machine cutterhead cutters is proposed, which realizes the all-round automatic management process of tunneling machine cutters, improving the level of automation and tunneling efficiency.

[0061] 2. A hob status tracking module was designed to calibrate each hob, thereby enabling precise control of the real-time status of each hob and improving the management accuracy of the cutter head hobs.

[0062] 3. A voice control module was designed, allowing workers to input hobbing information in real time via voice, avoiding omissions and errors caused by checking the cutter head before recording, thus improving the reliability of hobbing data. At the same time, voice control of the cutter head rotation avoids the need for workers to repeatedly rotate the cutter head when checking or changing cutters, saving time, improving construction efficiency, and protecting personnel safety.

[0063] 4. A data processing and storage module was designed, which can not only process the current cutter information, but also match the cutter data with tunneling parameters, surrounding rock conditions, and cutter maintenance status in real time and store it in the database, thereby improving the reference value and usability of historical cutter data.

[0064] 5. A hob wear prediction and abnormal condition early warning module was designed, which can predict the wear status of the cutter head hob in real time and provide timely warnings when the hob is abnormal. This allows for accurate control of the time for checking and replacing the cutter, extending the service life of key components such as the hob, cutter head, and bearings, improving the safety performance of the TBM, and reducing construction costs.

[0065] In addition to the purposes, features, and advantages described above, this application has other purposes, features, and advantages. The application will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0066] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0067] Figure 1 This is a schematic diagram of the tunneling machine cutterhead hob full-cycle management system module according to a preferred embodiment of this application.

[0068] Figure 2 This is a schematic diagram of the installation position of the cutterhead of a tunneling machine according to a preferred embodiment of this application.

[0069] Figure 3This is a schematic diagram of the control flow of the voice control module according to a preferred embodiment of this application.

[0070] Figure 4 This is a schematic diagram of data transmission of each module in the cutter head hob full life cycle management system according to another preferred embodiment of this application.

[0071] Figure 5 This is a schematic diagram of the convolutional neural network model structure of the hob wear prediction module according to a preferred embodiment of this application. Detailed Implementation

[0072] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0073] Reference Figure 1 A preferred embodiment of the present invention provides a full-cycle management system for the cutterhead hob of a tunneling machine, comprising:

[0074] The hob status tracking module is used to track and record the position and usage mileage of each hob in real time by calibrating first and then tracking, thereby controlling the real-time status of each hob.

[0075] The voice control module is used to obtain the current measured wear value of the hob or control the rotation angle of the cutter head based on voice information.

[0076] The hob wear prediction module is used to predict the current wear value of the hob in real time based on a preset prediction model.

[0077] The abnormal status early warning module is used to determine whether the cutter head hob has abnormal wear based on the wear value, and to issue an early warning in a timely manner when abnormal wear occurs;

[0078] The data processing and storage module is used to receive and record the data output by each module, and match the cutter data with the surrounding rock parameters, tunneling parameters, and cutter maintenance status in real time, and automatically generate forms to be stored in the database.

[0079] This embodiment provides a full-cycle management system for the cutterhead hob of a tunneling machine, including a hob status tracking module, a voice control module, a data processing and storage module, a hob wear prediction module, and a status anomaly early warning module, which has the following beneficial effects:

[0080] 1. A full-cycle management system for tunneling machine cutterhead cutters is proposed, which realizes the all-round and full-cycle automatic management of tunneling machine cutters, improving the level of automation and tunneling efficiency.

[0081] 2. A hob status tracking module was designed to calibrate each hob, thereby enabling precise control of the real-time status of each hob and improving the management accuracy of the cutter head hobs.

[0082] 3. A voice control module was designed, allowing workers to input hobbing information in real time via voice, avoiding omissions and errors caused by checking the cutter head before recording, thus improving the reliability of hobbing data. At the same time, voice control of the cutter head rotation avoids the need for workers to repeatedly rotate the cutter head when checking or changing cutters, saving time, improving construction efficiency, and protecting personnel safety.

[0083] 4. A data processing and storage module was designed, which can not only process the current cutter information, but also match the cutter data with tunneling parameters, surrounding rock conditions, and cutter maintenance status in real time and store it in the database, thereby improving the reference value and usability of historical cutter data.

[0084] 5. A hob wear prediction and abnormal condition early warning module was designed, which can predict the wear status of the cutter head hob in real time and provide timely warnings when the hob is abnormal. This allows for accurate control of the time for checking and replacing the cutter, extending the service life of key components such as the hob, cutter head, and bearings, improving the safety performance of the TBM, and reducing construction costs.

[0085] Specifically, in a preferred embodiment of this application, the hobbing cutter state tracking module specifically includes:

[0086] The initial position calibration module is used to first set the initial position A of the cutter head according to convention. i (θ i , ψ i This includes the 0° position and rotation direction. Then, each hob is numbered, and the initial position of each hob is marked, including the initial angle θ. i and distance ψ from the center of the cutter head i 'i' represents the hob number, and the cutter head diagram is shown below. Figure 2 As shown;

[0087] The tracking and recording module is used to track and record the current status B of each hob in real time while the cutter head is running. i (δ i S i )in:

[0088] Current angle δ of the hob i :

[0089] δ i =MOD(MOD(α-β, 360)+θ i ,360)

[0090] Current mileage S of the hob i :

[0091] S i =2πψ i (α+β) / 360

[0092] Where α is the total forward rotation angle of the cutterhead from the initial tunneling to the current moment, β is the total reverse rotation angle of the cutterhead from the initial tunneling to the current moment, and MOD(x, y) is the remainder function.

[0093] The hob status tracking module in this embodiment adopts a calibration-then-tracking approach to accurately control the real-time status of each hob.

[0094] Specifically, in a preferred embodiment of this application, the voice control module includes:

[0095] The voice recognition unit is used to recognize and process the voice control commands given by the worker to obtain the corresponding control parameters, including the hob number, the current radius of the hob, and the target angle.

[0096] The cutter head jog control box is used to control the cutter head to rotate by a corresponding angle when the control parameters recognized by the voice recognition unit include the word "rotate";

[0097] The controller is used to store the recognized radius value as the current radius of the corresponding numbered hob when the control parameters recognized by the voice recognition unit include the word "radius", and to calculate the current measured wear value of the corresponding numbered hob based on the initial radius of the corresponding numbered hob.

[0098] Specifically, in a preferred embodiment of this application, the cutter head jogging operation box is specifically used for:

[0099] When the control parameters recognized by the voice recognition unit include the word "rotation", the receiving voice recognition unit will identify the hob number i and the corresponding target angle. Automatic calculation of cutter head rotation angle And control the rotation of the cutter head.

[0100] Specifically, in a preferred embodiment of this application, when the cutter head jog control box controls the cutter head to rotate, it controls the cutter head to rotate along the shortest path, wherein the shortest path is: if 0≤λ≤180, the cutter head is controlled to rotate clockwise by an angle λ; if 180<λ<360, the cutter head is controlled to rotate counterclockwise by an angle 360-λ; if -180≤λ≤0, the cutter head is controlled to rotate counterclockwise by an angle |λ|; if -360<λ<-180, the cutter head is controlled to rotate clockwise by an angle 360+λ.

[0101] like Figure 3 As shown, the specific control flow of the voice control module in the above embodiment is as follows:

[0102] 1) Workers give control instructions, such as "i number + radius + value" or "i number + rotate to + angle", etc.

[0103] 2) The speech recognition unit recognizes and processes the commands, generating corresponding data, such as "N=i, D". i =d i "or "N=i, "

[0104] 3) After analyzing the command, the speech recognition unit proceeds to the corresponding process. For example, if the speech recognition unit recognizes the word "radius", it proceeds to step 4); if it recognizes the word "rotate", it proceeds to step 5.

[0105] 4) The controller receives data "N=i, D i =d i "To the data processing and storage module, the data processing and storage module automatically transfers d..." i Stored in the data table i The "Current Radius" column of hob number i (the first hob radius value entered will also be entered into the "Initial Radius" column of hob number i, i.e., r) i =d i It automatically generates the current measured wear value h of hob number i. i =r i -d i .

[0106] 5) The tool turret jog control box receives data "N=i, Then, the rotation angle of the cutter head is automatically calculated. The cutter head is controlled to rotate along the shortest path: if 0 ≤ λ ≤ 180, the cutter head rotates clockwise by an angle λ; if 180 < λ < 360, the cutter head rotates counterclockwise by an angle 360 ​​- λ; if -180 ≤ λ ≤ 0, the cutter head rotates counterclockwise by an angle |λ|; if -360 < λ < -180, the cutter head rotates clockwise by an angle 360 ​​+ λ. The control flow is as follows: Figure 3 As shown.

[0107] The above embodiments, by setting up a voice control module, ensure that workers can record hobbing information in a timely manner even in narrow and harsh environments, avoiding information loss and errors. At the same time, voice-assisted tool checking and tool changing, and control of the cutter head to rotate to the designated position along the shortest path, make operation simple, convenient and fast, greatly reducing the steps and time required for workers to operate.

[0108] Specifically, in a preferred embodiment of this application, the data processing and storage module includes:

[0109] The data receiving unit is used to receive and record the data output by the hob status tracking module, the voice control module, the hob wear prediction module, and the status anomaly early warning module.

[0110] The data processing unit matches and associates the cutter head data with the surrounding rock parameters, tunneling parameters, and cutter maintenance status in real time according to the time sequence, and automatically generates forms. The tunneling parameters include cutterhead rotation speed, propulsion speed, propulsion force, cutterhead torque, penetration depth, etc. The surrounding rock parameters include surrounding rock grade, etc. The cutter maintenance status includes maintenance status, maintenance reason, etc.

[0111] Data storage unit, used to store the generated forms.

[0112] In this embodiment, the data transmission relationship between the data processing and storage module and other modules is as follows: Figure 4 As shown, in this embodiment, the real-time information of the cutting tool is associated and matched with tunneling parameters, surrounding rock parameters, and tool maintenance status through a data processing and storage module before being stored in the database to ensure data integrity and effectiveness.

[0113] Specifically, in a preferred embodiment of this application, the automatically generated form content includes: hob number i, initial angle θ. i Current angle δ i ψ, distance from the center of the cutter head i Initial radius r i Current radius d i Mileage S i Measured wear value h i Wear prediction value k i , hobbing cutter state.

[0114] Specifically, in a preferred embodiment of this application, the hob wear prediction module includes:

[0115] The prediction model building module is used to build and train a model for predicting hob wear values;

[0116] The online prediction module is used to input the real-time collected feature information parameters into the hob wear value prediction model to perform online prediction and obtain the hob wear prediction value, so that an early warning can be given when the hob wear reaches the target level;

[0117] The offline optimization module of the prediction model retrieves newly generated historical data from the historical database of the cutterhead hob full-cycle management system at regular intervals during actual tunneling, using it as input samples to optimize the hob wear value prediction model offline. This makes the prediction model increasingly consistent with the current tunneling conditions, thereby further improving the prediction accuracy.

[0118] Specifically, the prediction model building module is used for:

[0119] Based on historical databases, model feature information data is selected, including: hob number and the corresponding cutter head rotation speed, feed speed, thrust, feed speed, penetration, surrounding rock grade, distance of hob from cutter head center, hob usage mileage, and measured hob wear value;

[0120] A hobbing cutter wear value prediction model is constructed based on a convolutional neural network, with the structure as follows: Figure 5 As shown, the model consists of 4 sets of convolutional encoders and 3 fully connected layers. After the data is processed, the convolutional encoders and fully connected layers within the model calculate and directly output the tool wear value Z = (Z1, Z2, ... Zn). i );

[0121] The feature information data is sequentially segmented into a 2D matrix X×Y containing both feature information and tool position information, which serves as the input to the model. The size of X depends on the length of the time series when the data is processed. Assuming the model input is a 10 (feature information category) × 256 (time series data for a certain time period) 2D matrix, the output is the tool wear Z = (Z1, Z2, ... Z...). i From the network structure, it can be seen that during forward propagation, the l-th layer convolutional encoder maps to L. l The result is:

[0122]

[0123] Where, x ij (l) The output of the l-th layer convolutional encoder is g(x), where g(x) is the activation function and b is the output of the l-th layer convolutional encoder. i (l) The threshold of the l-th layer convolutional encoder. This is a convolutional kernel in the convolutional encoder. This approach allows the network to efficiently describe complex interactions of multiple variables using only sparse interactions. Based on this, the computation results of the previous encoder layer are processed in the same way to the last convolutional encoder layer. The computation results of the last encoder layer are converted into a one-dimensional array and input to the fully connected layer. The fully connected layer consists of multiple neuron nodes. The output of the fully connected layer is:

[0124]

[0125] Where g(x) = max{0,x} is the activation function, b s l w is the hidden layer threshold. ab l The input layer weight matrix;

[0126] Find the error function of the hob wear prediction model:

[0127]

[0128] Among them, e q f is the expected value of the error function, i.e., the measured value of hob wear in the historical sample data. q The actual output of the error function is y, which is the output of the fully connected layer of the convolutional neural network. s ;

[0129] The backpropagation algorithm is used to optimize the parameters of the convolutional neural network and reduce the error. The partial derivatives with respect to the weights of the input layer and hidden layer are taken to obtain the weight update matrix.

[0130]

[0131]

[0132] In the formula, w i,j b represents the weights of the convolutional neural network. j η is the threshold of the convolutional neural network, and η is the learning rate of the neural network;

[0133] When △w i,j ≤ε and △b j When ≤ε, and ε is greater than zero and sufficiently small, the neural network model training ends, and the weight matrix w at this point is... i,j b j The optimal weight parameters are used as the learning parameters of the convolutional neural network to obtain the hob wear value prediction model.

[0134] This embodiment constructs a cutterhead cutter wear value prediction model through a convolutional neural network. The model has high prediction accuracy. At the same time, it comprehensively judges the cutterhead cutter status from three dimensions: abnormal predicted values, abnormal measured values, and abnormal tunneling parameters, and issues timely alarms, thereby improving the reliability of cutter status abnormality warnings.

[0135] Specifically, in a preferred embodiment of this application, the abnormal status early warning module includes:

[0136] The limit setting module is used to set the limit value for hob wear on the cutter head. max1 Hob wear limit value 2h max2 Thrust limit value F max Torque limit value T min Excavation speed limit value V min ;

[0137] The hob wear estimation calculation module is used to calculate the hob wear based on the measured value h obtained during the latest hob inspection. i (j) and its corresponding hob service mileage S i (j) Combined with the current hobbing mileage S i(t) Calculate the estimated wear value of the hob h gi (t):

[0138] h gi (t)=h i (j) *(S i (t)-S i (j) ) / S i (j) ;

[0139] The hob status anomaly judgment module is used to judge abnormal hob status conditions, including: prediction value anomaly judgment: if the current hob wear prediction value k i (t)>h max2 The abnormal status warning module will issue an alarm for abnormal predicted value of hob wear; the abnormal judgment of the measured value is as follows: if the predicted value of hob wear at the current time is k i (t)>h max1 And h i (j) +h gi (t)>h max2 The abnormal status warning module will issue an alarm indicating abnormal measured value of cutter wear; abnormal tunneling parameter judgment: if thrust F > F max And torque T <T max , or thrust F>F max And the propulsion speed V <V max If the abnormal status warning module issues an alarm indicating that the cutting tool wear has reached the required level and the tunneling parameters are abnormal, then the abnormal status warning module will issue an alarm.

[0140] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0141] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A full-cycle management system for the cutterhead hob of a tunneling machine, characterized in that, include: The hob status tracking module is used to track and record the position and usage mileage of each hob in real time by calibrating first and then tracking, thereby controlling the real-time status of each hob. The voice control module is used to obtain the current measured wear value of the hob or control the rotation angle of the cutter head based on voice information. The hob wear prediction module is used to predict the current wear value of the hob in real time based on a preset prediction model, and obtain the predicted hob wear value at the current moment. k i (t) ; The status anomaly early warning module is used to predict values. k i (t) The system determines whether the hobbing cutter on the cutter head exhibits abnormal wear and issues a timely warning when abnormal wear occurs. The abnormal wear warning module includes: The limit setting module is used to set the limit value for the wear of the hob on the cutter head. h max1 Hob wear limit value 2 h max2 Thrust limit value F max Torque limit value T min Excavation speed limit V min ; The hob wear estimation calculation module is used to calculate the hob wear based on the measured value obtained during the latest hob inspection. h i (j) and the corresponding hob service mileage S i (j) Combined with the current mileage of the hobbing cutter S i (t) Calculate the estimated value of hob wear h gi (t) : h gi (t) = h i (j) *( S i (t)-S i (j) ) / S i (j) ； The hob status anomaly detection module is used to determine abnormal hob status conditions, including: prediction value anomaly detection: if the current hob wear prediction value... k i (t) > h max2 The abnormal status warning module will issue an alarm indicating an abnormality in the predicted value of the hob wear; the abnormality judgment of the measured value is as follows: if the predicted value of the hob wear at the current moment is abnormal... k i (t) > h max1 and h i (j) + h gi (t) > h max2 The abnormal status warning module will issue an alarm indicating abnormal measured value of cutter wear; abnormal tunneling parameter judgment: if the thrust F > F max And torque T < T max or thrust F > F max And propulsion speed V < V max If the abnormal status warning module issues an alarm for abnormal tunneling parameters due to cutter wear; The data processing and storage module is used to receive and record the data output by each module, and match the cutter data with the surrounding rock parameters, tunneling parameters, and cutter maintenance status in real time, and automatically generate forms to be stored in the database.

2. The tunneling machine cutterhead hob full-cycle management system according to claim 1, characterized in that, The hobbing cutter status tracking module specifically includes: The initial position calibration module is used to first set the initial position of the cutter head according to convention, including the 0° position and rotation direction. Then, each hob is numbered, and the initial position of each hob is calibrated, including the initial angle. θ i and distance from the center of the cutter head ψ i , i Number the hobbing cutter; The tracking and recording module is used to track and record the current status of each hob in real time while the cutter head is running. B i ( δ i , S i )in: Current angle of the hobbing cutter δ i : δ i =MOD （ MOD （ α - β，360 ） +θ i ，360 ）； Current mileage of hobbing cutter S i : S i =2πψ i （ α+β ） / 360 ； in, α This represents the total forward rotation angle of the cutterhead from the initial excavation to the current moment. β This represents the total angle of reversal of the cutterhead from the initial digging to the current moment. MOD ( x, y ) is the remainder function.

3. The tunneling machine cutterhead hob full-cycle management system according to claim 1, characterized in that, The voice control module includes: The voice recognition unit is used to recognize and process the voice control commands given by the worker to obtain the corresponding control parameters, including the hob number, the current radius of the hob, and the target angle. The cutter head jog control box is used to control the cutter head to rotate by a corresponding angle when the control parameters recognized by the voice recognition unit include the word "rotate"; The controller is used to store the recognized radius value as the current radius of the corresponding numbered hob when the control parameters recognized by the voice recognition unit include the word "radius", and to calculate the current measured wear value of the corresponding numbered hob based on the initial radius of the corresponding numbered hob.

4. The tunneling machine cutterhead hob full-cycle management system according to claim 3, characterized in that, The cutter head jog control box is specifically used for: When the control parameters recognized by the voice recognition unit include the word "rotation", the receiving voice recognition unit will identify the hob number. i and the corresponding target angle φ i Automatically calculate the rotation angle of the cutter head λ=φ i - δ i And control the rotation of the cutter head.

5. The tunneling machine cutterhead hob full-cycle management system according to claim 4, characterized in that, When the cutter head jog control box controls the rotation of the cutter head, it controls the cutter head to rotate along the shortest path, wherein the shortest path is: if 0 ≤ λ ≤180, control the clockwise rotation angle of the cutter head. λ If 180 < λ <360, control the cutter head to rotate counterclockwise by 360 degrees. λ If -180≤ λ ≤0, controls the counterclockwise rotation angle of the cutter head. λ |;If -360< λ <-180, controls the cutter head to rotate clockwise by 360+ degrees. λ .

6. The tunneling machine cutterhead hob full-cycle management system according to claim 1, characterized in that, The data processing and storage module includes: The data receiving unit is used to receive and record the data output by the hob status tracking module, the voice control module, the hob wear prediction module, and the status anomaly early warning module. The data processing unit matches and associates the cutter head data with the surrounding rock parameters, tunneling parameters, and cutter maintenance status in real time according to the time sequence, and automatically generates a form. The tunneling parameters include cutterhead rotation speed, propulsion speed, propulsion force, cutterhead torque, and penetration depth. The surrounding rock parameters include the surrounding rock grade. The cutter maintenance status includes maintenance status and maintenance reason. Data storage unit, used to store the generated forms.

7. The tunneling machine cutterhead hob full-cycle management system according to claim 6, characterized in that, The automatically generated form content includes: the rolling cutter number. i Initial angle θ i Current perspective δ i Distance from the center of the cutter head ψ i Initial radius r i Current radius d i Mileage S i Measured wear values h i Wear prediction value k i , hobbing cutter state.

8. The tunneling machine cutterhead hob full-cycle management system according to claim 1, characterized in that, The hobbing cutter wear prediction module includes: The prediction model building module is used to build and train a model for predicting hob wear values; The online prediction module is used to input the real-time collected feature information parameters into the hob wear value prediction model for online prediction to obtain the hob wear prediction value; The offline optimization module of the prediction model retrieves newly generated historical data from the historical database of the cutterhead hob full-cycle management system at regular intervals during actual tunneling, using it as input samples to optimize the hob wear value prediction model offline. This makes the prediction model increasingly consistent with the current tunneling conditions, thereby further improving the prediction accuracy.

9. The tunneling machine cutterhead hob full-cycle management system according to claim 8, characterized in that, The prediction model building module is specifically used for: Based on historical databases, model feature information data is selected, including: hob number and the corresponding cutter head rotation speed, feed speed, thrust, feed speed, penetration, surrounding rock grade, distance of hob from cutter head center, hob usage mileage, and measured hob wear value; A hob wear prediction model is constructed based on a convolutional neural network. This model consists of four convolutional encoders and three fully connected layers. After processing the input data, the convolutional encoders and fully connected layers within the model directly output the wear values ​​of each tool. Z= ( Z 1 ，Z 2 , … Z i ); The various feature information data are organized sequentially in chronological order into segments that include both feature information and tool position information. X × Y The 2D matrix is ​​used as the input to this model. X The size depends on the length of the time series when the data is processed, and the output is the tool wear amount. Z = (Z 1 ，Z 2 , … Z i ） From the network structure, we can deduce that during forward propagation, the th... l Layer convolutional encoder mapping L l The result is: ； in, x ij (l) For the first l The output of the layer convolutional encoder, g(x) For activation function, b i (l) For the first l Layer convolutional encoder threshold, This is a convolutional kernel for the convolutional encoder. This approach allows the network to efficiently describe the complex interactions of multiple variables using only sparse interactions. Based on this, the computation results of the previous encoder layer are processed in the same way to the last convolutional encoder layer. The computation results of the last encoder layer are converted into a one-dimensional array and input to the fully connected layer. The fully connected layer consists of multiple neuron nodes. The output of the fully connected layer is: ； Where g(x) = max{0,x} is the activation function. b s l The hidden layer threshold, l The input layer weight matrix; Find the error function of the hob wear prediction model: ； in, e q The expected value of the error function is the measured value of hob wear from historical sample data. f q This is the actual output of the error function, i.e., the output of the fully connected layer of the convolutional neural network. y s ; The backpropagation algorithm is used to optimize the parameters of the convolutional neural network and reduce the error. The partial derivatives with respect to the weights of the input layer and hidden layer are taken to obtain the weight update matrix. ； In the formula, w i,j For the weights of a convolutional neural network, b j η is the threshold of the convolutional neural network, and η is the learning rate of the neural network; When △ w i,j ≤ ε And △ b j ≤ ε hour, ε When the weights are greater than zero and sufficiently small, the neural network model training ends, and the weight matrix at this point is... w i,j , b j The optimal weight parameters are used as the learning parameters of the convolutional neural network to obtain the hob wear value prediction model.

Images