A method, device and equipment for calculating torque of a tunneling machine cutter head and a storage medium

Abstract

The application discloses a tunneling machine cutter head torque calculation method, device and equipment and a storage medium. The method comprises the following steps: S1, importing historical geological parameters and cutter head structure parameters into tunneling machine historical operation data to form an original data sample; S2, preprocessing geological parameters, cutter head structure parameters and construction parameters in the original data sample; S3, analyzing the preprocessed data sample set to determine main characteristic parameters in each parameter; S4, establishing a cutter head torque multivariate nonlinear regression model structure based on a multilayer perception machine MLP and a long short-term memory LSTM according to the main characteristic parameters, and determining an optimal model through model data construction and model parameter identification; and S5, acquiring actual tunneling line geological parameters, cutter head structure parameters and construction parameters, constructing calculation data according to the model structure data requirements, and inputting the calculation data into the optimal cutter head torque multivariate nonlinear regression model to obtain the tunneling machine cutter head torque. The application is more efficient, more accurate and has a small calculation amount.

Description

Technical Field

[0001] The present application relates to the field of automatic control, and in particular, to a method, device, equipment and storage medium for calculating the cutter head torque of a tunnel boring machine. Background Art

[0002] With the development of tunnel engineering and underground space development, tunnel boring machines (TBMs), as highly automated and efficient tunnel construction equipment, are widely used in urban subway and river-crossing tunnel construction. The cutterhead of a TBM is a crucial component, and the rationality of its torque calculation is directly related to the safe and stable operation of the TBM, the overall cost of the machine, and power loss. If the cutterhead torque is calculated too low, the main drive motor and transformer capacity will be undersized, resulting in an excessively high actual load rate after commissioning, affecting the safe and stable operation of the shield machine. If the cutterhead torque is calculated too high, the main drive motor and transformer capacity will be oversized, resulting in a low actual load rate after commissioning, which will result in significant energy waste and increase the overall cost of the machine.

[0003] Currently, cutterhead torque calculations suffer from high computational complexity and low accuracy. For example, empirical estimation algorithms typically provide a wide range of estimates, and the empirical coefficients lack a rigorous basis. Furthermore, human experience varies, leading to large deviations in cutterhead torque calculations. Consequently, this leads to large deviations in the motor and cutterhead main drive transformer, ultimately impacting the safe and stable operation of the shield machine and the overall equipment cost. Therefore, accurately calculating cutterhead torque is crucial to the safe and stable operation of the tunnel boring machine and the overall equipment cost. Summary of the Invention

[0004] On the one hand, the present application provides a method for calculating the cutter head torque of a tunnel boring machine to solve the technical problems of large computational complexity and low accuracy in existing cutter head torque calculations.

[0005] This application is implemented through the following scheme:

[0006] A method for calculating the torque of a tunnel boring machine cutter head comprises the following steps:

[0007] S1. Import historical geological parameters and cutterhead structural parameters into the historical operation data of the roadheader to form a complete original data sample;

[0008] S2. Preprocessing the geological parameters, cutterhead structural parameters, and construction parameters in the original data samples, screening and cleaning the data, and normalizing the data sample sets after data screening and cleaning to convert them into dimensionless values;

[0009] S3, analyzing the preprocessed data sample set to determine the main characteristic parameters of each parameter;

[0010] S4. Establish the structure of the cutterhead torque multivariate nonlinear regression model based on the multilayer perceptron (MLP) and long short-term memory (LSTM) based on the main characteristic parameters, and determine the optimal model by constructing model data and identifying model parameters. When constructing the model data, the state vector is introduced as the input of the MLP and LSTM. When identifying the model parameters, the coefficients of the multivariate nonlinear regression model are fitted by weighted averaging the outputs of the MLP and LSTM.

[0011] S5. Obtain the actual geological parameters of the tunneling route, the cutterhead structural parameters, and the construction parameters, construct the calculation data according to the requirements of the model structural data, and input them into the optimal cutterhead torque multivariate nonlinear regression model to obtain the tunneling machine cutterhead torque.

[0012] Furthermore, the step S2 specifically includes the steps of:

[0013] S21. Collect and summarize geological parameters and cutterhead structural parameters of each tunneling interval and insert them into the corresponding tunneling machine operation history database;

[0014] S22. For the complete database after insertion, data is screened and cleaned according to certain rules and conditions;

[0015] S23. In order to make all parameter indicators at the same quantitative level, the data sample set after data screening and cleaning is normalized and converted into dimensionless values.

[0016] Furthermore, the geological parameters include tunnel burial depth, silt thickness, silty clay thickness, gravel thickness, silty soil thickness, residual clay thickness, fully weathered tuff thickness, strongly weathered tuff thickness, strongly weathered granite thickness, and moderately weathered granite thickness;

[0017] The construction parameters include cutter head torque, total propulsion force, propulsion pressure of each group, cutter head speed, propulsion speed, penetration rate, screw machine speed, foam pressure, foam mixture flow rate, foam air flow rate, foam mixture concentration, and soil pressure;

[0018] The cutterhead structural parameters refer to the structural design parameters related to the cutterhead torque, including the cutterhead diameter, cutterhead thickness, stirring rod length, stirring rod width, stirring rod height, stirring rod soil internal friction angle, number of stirring rods, cutterhead leg length, cutterhead leg width, cutterhead leg height, cutterhead leg force arm, number of cutters, number of edge scrapers, average cutter force arm, cutter cutting force, average edge scraper force arm, number of hobs, rolling force, cutterhead opening rate, and front hob cutter spacing.

[0019] Furthermore, in step S22, the rule conditions for data screening and cleaning refer to removing data in which any parameter of the cutter disc speed, total propulsion force, or cutter disc torque is zero within a certain range of the cutter disc rated speed.

[0020] Furthermore, the step S3 specifically includes the steps of:

[0021] S31. Apply the principal component analysis method to the preprocessed data sample set to determine the relationship between the various variables in the parameters and the contribution rate, sort and select them according to the contribution rate, and determine the main characteristic parameters in the parameters.

[0022] Furthermore, the step S4 specifically includes the steps of:

[0023] S41. Construct a model structure and establish a multivariate nonlinear regression model of the cutter head torque. Then, use the multi-layer perceptron (MLP) and long short-term memory (LSTM) weighted average fitting to fit the coefficients of the multivariate nonlinear regression model. This allows the model to have a strong ability to describe dynamic nonlinear characteristics. The model structure is as follows:

[0024]

[0025] Where: T is the cutter head torque, t represents the current time; p represents the number of parameters in the multivariate nonlinear regression model, which refers to the main characteristic parameters determined by the principal component analysis method; A 0,t-1 is the state-dependent bias of the model, A s,t-1 is the state-dependent coefficient of the sth parameter in the model; is the state dependency vector fitted by the multi-layer perceptron MLP, is the state dependency vector fitted by LSTM; Inputx is the common state vector of the multi-layer perceptron MLP and long short-term memory LSTM, where n w is the order of the state vector, R is the penetration;

[0026] S42. Model data construction, including:

[0027] The input layer data structure of the MLP and LSTM terminals is defined as follows:

[0028] Inputx=[T(tn w ),R(tn w ),...,T(t-1),R(t-1)] T

[0029] The input layer data structure of the multivariate nonlinear regression model is defined as follows:

[0030] InputDx=[1,x1,x2,...,x p-1 ,x p ] T

[0031] Initialize model structure parameters m, n i 、n wAnd the output dimension n of each LSTM layer;

[0032] According to the input layer data structure of the MLP and LSTM end and the multiple nonlinear regression model, construct the input layer data of the MLP and LSTM end and the input data of the multiple nonlinear regression model;

[0033] S43. Model parameter identification, including:

[0034] According to the calculation process of the coefficients of the multi-layer perceptron MLP and long short-term memory LSTM weighted average fitting multivariate nonlinear regression model, the data of the Inputx structure is used as the input of the MLP and LSTM terminals to calculate A s,t-1 , which is the regression coefficient of the multivariate nonlinear regression model. At this time, the cutter head torque prediction output is obtained by the following formula:

[0035]

[0036] in: A is the predicted output value of the cutter head torque; s,t-1 is the sth state dependency coefficient in the model;

[0037] Construct the loss function e and continuously update the model parameters through the back propagation algorithm until the loss function is minimized. The formula of the loss function e is as follows:

[0038]

[0039] Where: e represents the mean square error between the predicted value and the actual value, N represents the number of training samples; T is the cutter head torque, and I is the training sample number;

[0040] The gradient method is used to optimize the parameters of the model, the Adam optimization method is used to optimize the objective function in the formula, and the model parameters are updated by gradually decreasing the learning rate;

[0041] Changing the model structure parameters m and n by grid search method i 、n w And the output dimension n of each LSTM layer, repeat the above steps, traverse all model structures, compare the loss values ​​e under different model structures, and select the model structure and model parameters with the smallest loss function value as the final model.

[0042] Furthermore, in step S41, the multi-layer perceptron MLP is fitted to obtain The process is as follows:

[0043]

[0044] Where: m is the number of hidden layers in MLP; n i is the number of nodes in the i-th hidden layer; φs is the bias of the s-th output of the MLP model; is the output of the kth node in the mth hidden layer; is the weight from the kth node to the sth output coefficient of the mth hidden layer; k represents the node number; is the output of the jth node in the i-th hidden layer; is the weight from the kth node in the i-1th hidden layer to the jth node in the i-th hidden layer; is the bias of the jth node in the i-th hidden layer; is the activation function; L represents the number of input layer nodes, which here refers to the number of elements in the state vector; is the output of the jth node in the first hidden layer; is the weight from the kth node in the input layer to the jth node in the first hidden layer; is the bias of the output of the jth node in the first hidden layer; Inputx k is the kth node of the state vector;

[0045] Long short memory LSTM fitting The process is as follows:

[0046]

[0047] in, is the cell state of the first time step in the z-th layer LSTM; is the output of the first time step in the z-th layer LSTM; is the cell state of the lth time step in the first layer of LSTM; is the cell state at the l-1th time step in the first layer of LSTM; is the output of the lth time step in the first layer of LSTM; is the output of the l-1th time step in the first layer of LSTM; is the weight of the forget gate of the first layer LSTM; is the weight of the input gate of the first layer LSTM; The weight converted to the input gate of the first layer LSTM; is the weight of the output gate of the first layer LSTM; is the bias of the forget gate of the first layer LSTM; is the bias of the input gate of the first layer LSTM; Bias for the input gate transformation of the first layer LSTM; is the bias of the output gate of the first layer LSTM; is the weight of the forget gate of the Z-th layer LSTM; is the weight of the input gate of the Z-th layer LSTM; The weight converted from the input gate of the Z-th layer LSTM; is the weight of the output gate of the Z-th layer LSTM; is the bias of the forget gate of the Z-th layer LSTM; is the bias of the input gate of the Z-th layer LSTM; Bias for the input gate conversion of the Z-th layer LSTM; is the bias of the output gate of the Z-th layer LSTM; is the cell state at the lth time step in the Zth layer LSTM; is the output of the lth time step in the Zth layer LSTM; is the cell state at the l-1th time step in the Zth layer LSTM; is the output of the l-1th time step in the Zth layer LSTM; θ LSTM is the state dependency vector output by the LSTM end of the model, is the element of the state dependency vector; N represents the number of LSTM layers in series; is the feature vector of the Nth layer LSTM network at the lth time step, W fc is the weight of the fully connected layer; b fc is the bias of the fully connected layer; λ is the activation function of the fully connected layer; Represents the term-wise multiplication of two matrices of the same size.

[0048] On the other hand, the present application also provides a device for calculating the torque of a cutterhead of a roadheader, comprising:

[0049] The parameter import module is used to import historical geological parameters and cutterhead structural parameters into the historical operation data of the roadheader to form a complete original data sample;

[0050] The data preprocessing module is used to preprocess the geological parameters, cutterhead structural parameters and construction parameters in the original data samples, screen and clean the data, and normalize the data sample sets after data screening and cleaning to convert them into dimensionless values;

[0051] The main parameter analysis module is used to analyze the pre-processed data sample set and determine the main characteristic parameters of each parameter;

[0052] The model construction and identification module is used to establish the structure of the cutterhead torque multivariate nonlinear regression model based on the multilayer perceptron (MLP) and long short-term memory (LSTM) according to the main characteristic parameters, and to determine the optimal model by constructing model data and identifying model parameters. When constructing the model data, the state vector is introduced as the input of the MLP and LSTM. When identifying the model parameters, the coefficients of the multivariate nonlinear regression model are fitted by weighted averaging the outputs of the MLP and LSTM.

[0053] The cutterhead torque calculation module is used to obtain the actual geological parameters of the tunneling route, cutterhead structural parameters, and construction parameters. The calculation data is constructed according to the requirements of the model structure data and input into the optimal cutterhead torque multivariate nonlinear regression model to obtain the tunneling machine cutterhead torque.

[0054] On the other hand, the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for calculating the cutter head torque of the tunnel boring machine when executing the computer program.

[0055] On the other hand, the present application further provides a storage medium, which includes a stored program, and when the program is run, controls the device where the storage medium is located to execute the steps of the tunnel boring machine cutter head torque calculation method.

[0056] Compared with the existing technology, this application has the following beneficial effects:

[0057] (1) Based on data identification, this application comprehensively considers the influence of geological parameters, cutterhead structural parameters, and construction parameters on the calculation of cutterhead torque. The proposed cutterhead torque multivariate nonlinear regression model based on multilayer perceptron (MLP) and long short-term memory (LSTM) reduces the input dimension of the multilayer perceptron (MLP) and long short-term memory (LSTM) by introducing a state vector, thus reducing the amount of calculation and making the calculation more efficient. At the same time, by fitting the coefficients of the multivariate nonlinear regression model, the dynamic nonlinear description capability of the cutterhead torque calculation is enhanced, the parameters are comprehensive, and the calculated cutterhead torque is more accurate.

[0058] (2) The multivariate nonlinear regression model of cutter disc torque based on multilayer perceptron MLP and long short-term memory LSTM proposed in this application fits the coefficients of the multivariate nonlinear regression model by weighted average of the outputs of MLP and LSTM, making the parameters of the model more accurate and the calculated cutter disc torque more accurate.

[0059] (3) The motor, cutterhead main drive transformer and corresponding electrical control equipment selected in this application are based on the precise cutterhead torque and rated design speed, which not only ensures the safe and stable operation of the tunnel boring machine, but also makes the cost of the entire machine more reasonable.

[0060] In addition to the above-described purposes, features and advantages, the present application also has other purposes, features and advantages. The present application will be further described in detail below with reference to the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The drawings that constitute a part of this application are used to provide further understanding of this application. The illustrative embodiments of this application and their descriptions are used to explain this application and do not constitute improper limitations on this application.

[0062] Figure 1 This is a flow chart of a method for calculating the cutter head torque of a tunnel boring machine according to a preferred embodiment of the present application.

[0063] Figure 2 This is a flowchart of the sub-steps of step S2 in the preferred embodiment of the present application.

[0064] Figure 3 This is a flowchart of the sub-steps of step S3 in the preferred embodiment of the present application.

[0065] Figure 4 This is a schematic diagram of the structure of the cutterhead torque multivariate nonlinear regression model based on the multi-layer perceptron MLP and long short-term memory LSTM in the preferred embodiment of the present application.

[0066] Figure 5 This is a schematic diagram of the module of the tunnel boring machine cutter head torque calculation device in the preferred embodiment of the present application.

[0067] Figure 6 This is a schematic block diagram of an electronic device entity according to a preferred embodiment of the present application.

[0068] Figure 7 It is a diagram of the internal structure of a computer device according to a preferred embodiment of the present application. DETAILED DESCRIPTION

[0069] The embodiments of the present application are described in detail below with reference to the accompanying drawings, but the present application can be implemented in a variety of different ways defined and covered below.

[0070] like Figure 1 As shown, the preferred embodiment of the present application provides a method for calculating the torque of a tunnel boring machine cutter head, comprising the steps of:

[0071] S1. Import historical geological parameters and cutterhead structural parameters into the historical operation data of the roadheader to form a complete original data sample;

[0072] S2. Preprocessing the geological parameters, cutterhead structural parameters, and construction parameters in the original data samples, screening and cleaning the data, and normalizing the data sample sets after data screening and cleaning to convert them into dimensionless values;

[0073] S3, analyzing the preprocessed data sample set to determine the main characteristic parameters of each parameter;

[0074] S4. Establish the structure of the cutterhead torque multivariate nonlinear regression model based on the multilayer perceptron (MLP) and long short-term memory (LSTM) based on the main characteristic parameters, and determine the optimal model by constructing model data and identifying model parameters. When constructing the model data, the state vector is introduced as the input of the MLP and LSTM. When identifying the model parameters, the coefficients of the multivariate nonlinear regression model are fitted by weighted averaging the outputs of the MLP and LSTM.

[0075] S5. Obtain the actual geological parameters of the tunneling route, the cutterhead structural parameters, and the construction parameters, construct the calculation data according to the requirements of the model structural data, and input them into the optimal cutterhead torque multivariate nonlinear regression model to obtain the tunneling machine cutterhead torque.

[0076] Compared with the prior art, this embodiment has the following beneficial effects:

[0077] (1) Based on data identification, this embodiment comprehensively considers the influence of geological parameters, cutterhead structural parameters, and construction parameters on the calculation of cutterhead torque. The proposed cutterhead torque multivariate nonlinear regression model based on the multilayer perceptron (MLP) and long short-term memory (LSTM) reduces the input dimension of the multilayer perceptron (MLP) and long short-term memory (LSTM) by introducing the state vector, thus reducing the amount of calculation and making the calculation more efficient. At the same time, by fitting the coefficients of the multivariate nonlinear regression model, the dynamic nonlinear description capability of the cutterhead torque calculation is enhanced, the parameters are comprehensive, and the calculated cutterhead torque is more accurate.

[0078] (2) The cutterhead torque multivariate nonlinear regression model based on the multilayer perceptron MLP and long short-term memory LSTM proposed in this embodiment fits the coefficients of the multivariate nonlinear regression model by weighted average fitting of the outputs of the MLP and LSTM, making the model parameters more accurate and the calculated cutterhead torque more accurate.

[0079] (3) In this embodiment, the motor, cutterhead main drive transformer and corresponding electrical control equipment are selected based on the precise cutterhead torque and rated design speed, which not only ensures the safe and stable operation of the tunnel boring machine, but also makes the cost of the entire machine more reasonable.

[0080] Preferably, if Figure 2 As shown, the step S2 specifically includes the following steps:

[0081] S21. Collect and summarize geological parameters and cutterhead structural parameters of each tunneling interval and insert them into the corresponding tunneling machine operation history database;

[0082] S22. For the complete database after insertion, data is screened and cleaned according to certain rules and conditions;

[0083] S23. In order to make all parameter indicators at the same quantitative level, the data sample set after data screening and cleaning is normalized and converted into dimensionless values.

[0084] In this example, geological parameters and cutterhead structural parameters for each tunneling section are collected and aggregated and inserted into the corresponding tunneling machine operation history database. This complete database is then filtered and cleaned according to specific criteria. To ensure that all parameters are at the same level, the filtered and cleaned data samples are normalized and converted to dimensionless values.

[0085] Preferably, the geological parameters include tunnel burial depth, silt thickness, silty clay thickness, gravel thickness, silty soil thickness, residual clay thickness, fully weathered tuff thickness, strongly weathered tuff thickness, strongly weathered granite thickness, and moderately weathered granite thickness;

[0086] The construction parameters include a target parameter such as cutter head torque, total propulsion force, group A propulsion pressure, group B propulsion pressure, group C propulsion pressure, group D propulsion pressure, cutter head speed, propulsion speed, penetration rate, screw machine speed, foam pressure, foam mixture flow rate, foam air flow rate, foam mixture concentration, 1# soil pressure, 2# soil pressure, 3# soil pressure, 4# soil pressure, 5# soil pressure, and 6# soil pressure;

[0087] The cutterhead structural parameters refer to the structural design parameters related to the cutterhead torque, including the cutterhead diameter, cutterhead thickness, stirring rod length, stirring rod width, stirring rod height, stirring rod soil internal friction angle, number of stirring rods, cutterhead leg length, cutterhead leg width, cutterhead leg height, cutterhead leg force arm, number of cutters, number of edge scrapers, average cutter force arm, cutter cutting force, average edge scraper force arm, number of hobs, rolling force, cutterhead opening rate, and front hob cutter spacing.

[0088] The present embodiment has rich data on geological parameters, construction parameters and cutterhead structural parameters, which ensures that the calculation results are closer to the actual working conditions and facilitates more accurate calculation of the cutterhead torque.

[0089] Preferably, in step S22, the rule conditions for data screening and cleaning refer to removing data in which any parameter of the cutter disc speed, total propulsion force, and cutter disc torque is zero within a certain range of the cutter disc rated speed.

[0090] Preferably, if Figure 3 As shown, the step S3 specifically includes the following steps:

[0091] S31. Apply the principal component analysis method to the preprocessed data sample set to determine the relationship between the various variables in the parameters and the contribution rate, sort and select them according to the contribution rate, and determine the main characteristic parameters in the parameters.

[0092] In this embodiment, the specific steps of the principal component analysis method are as follows:

[0093] (1) Apply z-score to standardize the data sample set and combine the samples into a matrix;

[0094] (2) Calculate the covariance matrix of the combined matrix;

[0095] (3) Calculate the eigenvalues ​​and eigenvectors of the covariance matrix in step (2);

[0096] (4) The solved eigenvectors are recombined in the order of eigenvalues ​​to form a mapping matrix, and the first n rows or columns of the mapping matrix are taken as the final mapping matrix according to the number of features retained by principal component analysis.

[0097] (5) Use the mapping matrix in step (4) to map the standardized data samples to achieve the purpose of data dimensionality reduction.

[0098] The embodiment is not only conducive to reducing the amount of calculation, but also can ensure that the calculation results are closer to the actual working conditions, which is conducive to more accurate calculation of the cutter head torque.

[0099] Preferably, the step S4 specifically includes the steps of:

[0100] S41. Construct the model structure. In the actual operation of the tunnel boring machine, the cutterhead torque is closely related to the geological parameters, construction parameters and cutterhead structural parameters. In addition, the cutterhead torque is dynamically nonlinear. Therefore, a multivariate nonlinear regression model of the cutterhead torque is first established. Then, the coefficients of the multivariate nonlinear regression model are fitted by the multi-layer perceptron MLP and the long short-term memory LSTM weighted average, so that the model has a strong ability to describe dynamic nonlinear characteristics. The specific model structure is as follows: Figure 4 As shown in the figure, the state vector is first used as the input of the MLP and LSTM networks. The LSTM network calculates the feature vector, and then the fully connected layer converts the feature vector into the state dependency vector fitted by the LSTM. At the same time, the MLP network calculates the state dependency vector fitted by the MLP. The state dependency vector fitted by the LSTM and the state dependency vector fitted by the MLP are averaged as the coefficients of the multiple nonlinear regression model. At this time, they are input into the multiple nonlinear regression model together with the input of the multiple nonlinear regression model to finally obtain the cutter head torque. Its model structure is:

[0101]

[0102] Where: T is the cutter head torque, t represents the current time; p represents the number of parameters in the multivariate nonlinear regression model, which refers to the main characteristic parameters determined by the principal component analysis method; A 0,t-1 is the state-dependent bias of the model, A s,t-1 is the state-dependent coefficient of the sth parameter in the model; is the state dependency vector fitted by the multi-layer perceptron MLP, is the state dependency vector fitted by LSTM; Inputx is the common state vector of the multi-layer perceptron MLP and long short-term memory LSTM, where n w is the order of the state vector, R is the penetration;

[0103] Specifically, the multi-layer perceptron MLP fitting is obtained The process is as follows:

[0104]

[0105] Where: m is the number of hidden layers in MLP; n i is the number of nodes in the i-th hidden layer; φ s is the bias of the s-th output of the MLP model; is the output of the kth node in the mth hidden layer; is the weight from the kth node to the sth output coefficient of the mth hidden layer; k represents the node number; is the output of the jth node in the i-th hidden layer; is the weight from the kth node in the i-1th hidden layer to the jth node in the i-th hidden layer; is the bias of the jth node in the i-th hidden layer; is the activation function; L represents the number of input layer nodes, which here refers to the number of elements in the state vector; is the output of the jth node in the first hidden layer; is the weight from the kth node in the input layer to the jth node in the first hidden layer; is the bias of the output of the jth node in the first hidden layer; Inputx k is the kth node of the state vector;

[0106] Long short memory LSTM fitting The process is as follows:

[0107]

[0108] in, is the cell state of the first time step in the z-th layer LSTM; is the output of the first time step in the z-th layer LSTM; is the cell state of the lth time step in the first layer of LSTM; is the cell state at the l-1th time step in the first layer of LSTM; is the output of the lth time step in the first layer of LSTM; is the output of the l-1th time step in the first layer of LSTM; is the weight of the forget gate of the first layer LSTM; is the weight of the input gate of the first layer LSTM; The weight converted to the input gate of the first layer LSTM; is the weight of the output gate of the first layer LSTM; is the bias of the forget gate of the first layer LSTM; is the bias of the input gate of the first layer LSTM; Bias for the input gate transformation of the first layer LSTM; is the bias of the output gate of the first layer LSTM; is the weight of the forget gate of the Z-th layer LSTM; is the weight of the input gate of the Z-th layer LSTM; The weight converted from the input gate of the Z-th layer LSTM; is the weight of the output gate of the Z-th layer LSTM; is the bias of the forget gate of the Z-th layer LSTM; is the bias of the input gate of the Z-th layer LSTM; Bias for the input gate conversion of the Z-th layer LSTM; is the bias of the output gate of the Z-th layer LSTM; is the cell state at the lth time step in the Zth layer LSTM; is the output of the lth time step in the Zth layer LSTM; is the cell state at the l-1th time step in the Zth layer LSTM; is the output of the l-1th time step in the Zth layer LSTM; θ LSTM is the state dependency vector output by the LSTM end of the model, is the element of the state dependency vector; N represents the number of LSTM layers in series; is the feature vector of the Nth layer LSTM network at the lth time step, W fc is the weight of the fully connected layer; b fc is the bias of the fully connected layer; λ is the activation function of the fully connected layer; Represents the term-wise multiplication of two matrices of the same size.

[0109] S42. Model data construction, including:

[0110] The input layer data structure of the MLP and LSTM terminals is defined as follows:

[0111] Inputx=[T(tn w ),R(tn w ),...,T(t-1),R(t-1)] T (4)

[0112] The input layer data structure of the multivariate nonlinear regression model is defined as follows:

[0113] InputDx=[1,x1,x2,...,x p-1 ,x p ] T (5)

[0114] Initialize model structure parameters m, n i 、n w And the output dimension n of each LSTM layer;

[0115] According to the MLP and LSTM end and the input layer data structure of the multivariate nonlinear regression model in formula (4) and formula (5), the input layer data of the MLP and LSTM end and the input data of the multivariate nonlinear regression model are constructed;

[0116] S43. Model parameter identification, including:

[0117] S431, according to the calculation process of the coefficients of the multi-layer perceptron MLP and long short-term memory LSTM weighted average fitting multivariate nonlinear regression model from formula (1) to formula (3), the data of the Inputx structure is used as the input of the MLP and LSTM terminals, and A is calculated. s,t-1 , which is the regression coefficient of the multivariate nonlinear regression model. At this time, the cutter head torque prediction output is obtained by the following formula:

[0118]

[0119] in: A is the predicted output value of the cutter head torque; s,t-1 is the sth state dependency coefficient in the model.

[0120] S432. Construct a loss function e and continuously update the model parameters through the back propagation algorithm until the loss function is minimized. The formula of the loss function e is as follows:

[0121]

[0122] Where: e represents the mean square error between the predicted value and the actual value, N represents the number of training samples; T is the cutter head torque, and I is the training sample number;

[0123] The gradient method is used to optimize the parameters of the model, the Adam optimization method is used to optimize the objective function in the formula, and the model parameters are updated by gradually decreasing the learning rate;

[0124] Changing the model structure parameters m and n by grid search method i 、n w And the output dimension n of each LSTM layer, repeat the above steps S431 and S432, after traversing all model structures, compare the loss values ​​e under different model structures, and select the model structure and model parameters with the smallest loss function value as the final model.

[0125] In step S5, when the actual line geological parameters, construction parameters, and cutterhead structural parameters are input into the model to calculate the actual cutterhead torque, the line geological data, cutterhead structural parameters, and construction parameters provided by the customer are constructed into calculation data according to the requirements of the model structure data and input into the cutterhead torque multivariate nonlinear regression model based on the multi-layer perceptron MLP and long short-term memory LSTM, thereby obtaining the cutterhead torque.

[0126] like Figure 5 As shown, the present application also provides a device for calculating the torque of a cutter head of a tunnel boring machine, comprising:

[0127] The parameter import module is used to import historical geological parameters and cutterhead structural parameters into the historical operation data of the roadheader to form a complete original data sample;

[0128] The data preprocessing module is used to preprocess the geological parameters, cutterhead structural parameters and construction parameters in the original data samples, screen and clean the data, and normalize the data sample sets after data screening and cleaning to convert them into dimensionless values;

[0129] The main parameter analysis module is used to analyze the pre-processed data sample set and determine the main characteristic parameters of each parameter;

[0130] The model construction and identification module is used to establish the structure of the cutterhead torque multivariate nonlinear regression model based on the multilayer perceptron (MLP) and long short-term memory (LSTM) according to the main characteristic parameters, and to determine the optimal model by constructing model data and identifying model parameters. When constructing the model data, the state vector is introduced as the input of the MLP and LSTM. When identifying the model parameters, the coefficients of the multivariate nonlinear regression model are fitted by weighted averaging the outputs of the MLP and LSTM.

[0131] The cutterhead torque calculation module is used to obtain the actual geological parameters of the tunneling route, cutterhead structural parameters, and construction parameters. The calculation data is constructed according to the requirements of the model structure data and input into the optimal cutterhead torque multivariate nonlinear regression model to obtain the tunneling machine cutterhead torque.

[0132] like Figure 6 As shown, a preferred embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for calculating the cutter head torque of the tunnel boring machine in the above embodiment when executing the computer program.

[0133] like Figure 7 As shown, the preferred embodiment of the present application further provides a computer device, which can be a terminal or a liveness detection server, and its internal structure diagram can be as shown in FIG. Figure 7 As shown. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with other external computer devices via a network connection. When executed by the processor, the computer program implements the steps of the above-mentioned method for calculating the cutterhead torque of a tunnel boring machine.

[0134] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0135] A preferred embodiment of the present application further provides a storage medium, which includes a stored program. When the program is run, the device where the storage medium is located is controlled to execute the steps of the method for calculating the cutter head torque of the tunnel boring machine in the above embodiment.

[0136] In summary, the above embodiments of the present application have the following technical features compared to the prior art:

[0137] (1) The method for calculating the cutterhead torque of a tunnel boring machine proposed in this application comprehensively considers geological data, cutterhead structural parameters and construction parameters, and establishes a cutterhead torque multivariate nonlinear regression model based on a multilayer perceptron (MLP) and a long short-term memory (LSTM). The model fully utilizes the deep learning capability of MLP and the advantage of LSTM in processing long interval and long delay time series, so that the model has a strong dynamic nonlinear description capability. For different working points of the cutterhead, the model can be converted into a multivariate linear regression model with different parameters, which is more in line with the actual working conditions of the cutterhead.

[0138] (2) The cutterhead torque multivariate nonlinear regression model based on multilayer perceptron MLP and long short-term memory LSTM proposed in this application reduces the input dimension of the multilayer perceptron MLP and long short-term memory LSTM by taking the cutterhead torque and penetration as state vectors, thereby reducing the amount of calculation and making the calculation more efficient.

[0139] (3) The multivariate nonlinear regression model of cutter disc torque based on multilayer perceptron MLP and long short-term memory LSTM proposed in this application fits the coefficients of the multivariate nonlinear regression model by weighted average of the outputs of MLP and LSTM, making the parameters of the model more accurate and the calculated cutter disc torque more accurate.

[0140] (4) The present invention proposes a cutterhead torque calculation device for a tunnel boring machine, which includes a parameter import module, a data preprocessing module, a main parameter analysis module, a model construction and identification module, and a cutterhead torque calculation module. The device improves the convenience of cutterhead torque calculation.

[0141] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0142] If the functions described in the method of this embodiment are implemented in the form of a software functional unit and sold or used as an independent product, they can be stored in a storage medium readable by one or more computing devices. Based on this understanding, the part of the embodiment of the present application that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for enabling a computing device (which can be a personal computer, server, mobile computing device or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.

[0143] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiment of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal translation scripting language JavaScript, etc.

[0144] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0145] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0146] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0147] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic accurate concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0148] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A method for calculating the cutter head torque of a tunnel boring machine, characterized in that: Including steps: S1. Import historical geological parameters and cutterhead structural parameters into the historical operation data of the roadheader to form a complete original data sample; S2. Preprocessing the geological parameters, cutterhead structural parameters, and construction parameters in the original data samples, screening and cleaning the data according to the rules and conditions, and normalizing the data sample set after data screening and cleaning to convert them into dimensionless values; S3, analyzing the preprocessed data sample set to determine the main characteristic parameters of each parameter; S4. Establish the structure of the cutter head torque multivariate nonlinear regression model based on the multilayer perceptron MLP and long short-term memory LSTM according to the main characteristic parameters, and determine the optimal model by constructing model data and identifying model parameters. When constructing the model data, introduce the state vector as the input of the MLP and LSTM. When identifying the model parameters, the coefficients of the multivariate nonlinear regression model are fitted by weighted average of the outputs of the MLP and LSTM. Specifically, the steps include: S41. Construct the model structure, establish the multivariate nonlinear regression model of the cutter head torque, and then fit the coefficients of the multivariate nonlinear regression model by weighted average of the multilayer perceptron MLP and long short-term memory LSTM, so that the model has a strong dynamic nonlinear characteristic description capability. The model structure is: Where: T is the cutter head torque, t represents the current time; p represents the number of parameters in the multivariate nonlinear regression model, which refers to the main characteristic parameters determined by the principal component analysis method; A 0,t-1 is the state-dependent bias of the model, A s,t-1 is the state-dependent coefficient of the sth parameter in the model; is the state dependency vector fitted by the multi-layer perceptron MLP, is the state dependency vector fitted by LSTM; Inputx is the common state vector of the multi-layer perceptron MLP and long short-term memory LSTM, where n w is the order of the state vector, R is the penetration; S5. Obtain the actual geological parameters of the tunneling route, the cutterhead structural parameters, and the construction parameters, construct the calculation data according to the requirements of the model structural data, and input them into the optimal cutterhead torque multivariate nonlinear regression model to obtain the tunneling machine cutterhead torque.

2. The method for calculating the cutter head torque of a roadheader according to claim 1, characterized in that: Said step S1 specifically comprises the steps of collecting and summarizing geological parameters and cutterhead structural parameters of each tunneling interval, and inserting them into the corresponding tunneling machine operation history database.

3. The method for calculating cutter head torque of a roadheader according to claim 1, characterized in that: The geological parameters include tunnel burial depth, silt thickness, silty clay thickness, gravel thickness, silty soil thickness, residual clay thickness, fully weathered tuff thickness, strongly weathered tuff thickness, strongly weathered granite thickness, and moderately weathered granite thickness; The construction parameters include cutter head torque, total propulsion force, propulsion pressure of each group, cutter head speed, propulsion speed, penetration rate, screw machine speed, foam pressure, foam mixture flow rate, foam air flow rate, foam mixture concentration, and soil pressure; The cutterhead structural parameters refer to the structural design parameters related to the cutterhead torque, including the cutterhead diameter, cutterhead thickness, stirring rod length, stirring rod width, stirring rod height, stirring rod soil internal friction angle, number of stirring rods, cutterhead leg length, cutterhead leg width, cutterhead leg height, cutterhead leg force arm, number of cutters, number of edge scrapers, average cutter force arm, cutter cutting force, average edge scraper force arm, number of hobs, rolling force, cutterhead opening rate, and front hob cutter spacing.

4. The method for calculating the cutter head torque of a roadheader according to claim 3, wherein: In step S2, the rule condition for data screening and cleaning is to eliminate data in which any parameter of the cutter head speed, total propulsion force, or cutter head torque is zero within a certain range of the cutter head rated speed.

5. The method for calculating the cutter head torque of a roadheader according to claim 4, characterized in that: The step S3 specifically includes the following steps: S31. Apply the principal component analysis method to the preprocessed data sample set to determine the relationship between the various variables in the parameters and the contribution rate, sort and select them according to the contribution rate, and determine the main characteristic parameters in the parameters.

6. The method for calculating the cutter head torque of a roadheader according to claim 5, characterized in that: The step S4 specifically further includes the steps of: S42. Model data construction, including: The input layer data structure of the MLP and LSTM terminals is defined as follows: Inputx=[T(t-n w ),R(t-n w ),...,T(t-1),R(t-1)] T ; The input layer data structure of the multivariate nonlinear regression model is defined as follows: InputDx=[1,x1,x2,...,x p-1 ,x p ] T ; Initialize model structure parameters m, n i 、n w And the output dimension n of each LSTM layer; According to the input layer data structure of the MLP and LSTM end and the multiple nonlinear regression model, construct the input layer data of the MLP and LSTM end and the input data of the multiple nonlinear regression model; S43. Model parameter identification, including: According to the calculation process of the coefficients of the multi-layer perceptron MLP and long short-term memory LSTM weighted average fitting multivariate nonlinear regression model, the data of the Inputx structure is used as the input of the MLP and LSTM terminals to calculate A s,t-1 , which is the regression coefficient of the multivariate nonlinear regression model. At this time, the cutter head torque prediction output is obtained by the following formula: in: A is the predicted output value of the cutter head torque; s,t-1 is the sth state dependency coefficient in the model; Construct the loss function e and continuously update the model parameters through the back propagation algorithm until the loss function is minimized. The formula of the loss function e is as follows: Where: e represents the mean square error between the predicted value and the actual value, N represents the number of training samples; T is the cutter head torque, and I is the training sample number; The gradient method is used to optimize the parameters of the model, and the model parameters are updated by gradually decreasing the learning rate; Changing the model structure parameters m and n by grid search method i 、n w And the output dimension n of each LSTM layer, repeat the above steps, traverse all model structures, compare the loss values ​​e under different model structures, and select the model structure and model parameters with the smallest loss function value as the final model.

7. The method for calculating the cutter head torque of a roadheader according to claim 6, characterized in that: In step S41, the multi-layer perceptron MLP is fitted to obtain The process is as follows: Where: m is the number of hidden layers in MLP; n i is the number of nodes in the i-th hidden layer; φ s is the bias of the s-th output of the MLP model; is the output of the kth node in the mth hidden layer; is the weight from the kth node to the sth output coefficient of the mth hidden layer; k represents the node number; is the output of the jth node in the i-th hidden layer; is the weight from the kth node in the i-1th hidden layer to the jth node in the i-th hidden layer; is the bias of the jth node in the i-th hidden layer; is the activation function; L represents the number of input layer nodes, which here refers to the number of elements in the state vector; is the output of the jth node in the first hidden layer; is the weight from the kth node in the input layer to the jth node in the first hidden layer; is the bias of the output of the jth node in the first hidden layer; Inputx k is the kth node of the state vector; Long short memory LSTM fitting The process is as follows: in, is the cell state of the first time step in the z-th layer LSTM; is the output of the first time step in the z-th layer LSTM; is the cell state of the lth time step in the first layer of LSTM; is the cell state at the l-1th time step in the first layer of LSTM; is the output of the lth time step in the first layer of LSTM; is the output of the l-1th time step in the first layer of LSTM; is the weight of the forget gate of the first layer LSTM; is the weight of the input gate of the first layer LSTM; The weight converted to the input gate of the first layer LSTM; is the weight of the output gate of the first layer LSTM; is the bias of the forget gate of the first layer LSTM; is the bias of the input gate of the first layer LSTM; Bias for the input gate transformation of the first layer LSTM; is the bias of the output gate of the first layer LSTM; is the weight of the forget gate of the Z-th layer LSTM; is the weight of the input gate of the Z-th layer LSTM; The weight converted from the input gate of the Z-th layer LSTM; is the weight of the output gate of the Z-th layer LSTM; is the bias of the forget gate of the Z-th layer LSTM; is the bias of the input gate of the Z-th layer LSTM; Bias for the input gate conversion of the Z-th layer LSTM; is the bias of the output gate of the Z-th layer LSTM; is the cell state at the lth time step in the Zth layer LSTM; is the output of the lth time step in the Zth layer LSTM; is the cell state at the l-1th time step in the Zth layer LSTM; is the output of the l-1th time step in the Zth layer LSTM; θ LSTM is the state dependency vector output by the LSTM end of the model, is the element of the state dependency vector; N represents the number of LSTM layers in series; is the feature vector of the Nth layer LSTM network at the lth time step, W fc is the weight of the fully connected layer; b fc is the bias of the fully connected layer; λ is the activation function of the fully connected layer; Represents the term-wise multiplication of two matrices of the same size.

8. A device for calculating the cutter head torque of a tunnel boring machine, characterized in that: include: The parameter import module is used to import historical geological parameters and cutterhead structural parameters into the historical operation data of the roadheader to form a complete original data sample; The data preprocessing module is used to preprocess the geological parameters, cutterhead structural parameters, and construction parameters in the original data samples, screen and clean the data according to the rules and conditions, and normalize the data sample sets after data screening and cleaning to convert them into dimensionless values; The main parameter analysis module is used to analyze the pre-processed data sample set and determine the main characteristic parameters of each parameter; The model construction and identification module is used to establish the structure of the cutterhead torque multivariate nonlinear regression model based on the multilayer perceptron MLP and long short-term memory LSTM according to the main characteristic parameters, and determine the optimal model by constructing model data and identifying model parameters. When constructing the model data, the state vector is introduced as the input of the MLP and LSTM. When identifying the model parameters, the coefficients of the multivariate nonlinear regression model are fitted by weighted average of the outputs of the MLP and LSTM. Specifically, it is used to: construct the model structure, establish the multivariate nonlinear regression model of the cutterhead torque, and then fit the coefficients of the multivariate nonlinear regression model by weighted average of the multilayer perceptron MLP and long short-term memory LSTM, so that the model has a strong ability to describe dynamic nonlinear characteristics. The model structure is: Where: T is the cutter head torque, t represents the current time; p represents the number of parameters in the multivariate nonlinear regression model, which refers to the main characteristic parameters determined by the principal component analysis method; A 0,t-1 is the state-dependent bias of the model, A s,t-1 is the state-dependent coefficient of the sth parameter in the model; is the state dependency vector fitted by the multi-layer perceptron MLP, is the state dependency vector fitted by LSTM; Inputx is the common state vector of the multi-layer perceptron MLP and long short-term memory LSTM, where n w is the order of the state vector, R is the penetration; The cutterhead torque calculation module is used to obtain the actual geological parameters of the tunneling route, cutterhead structural parameters, and construction parameters. The calculation data is constructed according to the requirements of the model structure data and input into the optimal cutterhead torque multivariate nonlinear regression model to obtain the tunneling machine cutterhead torque.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method for calculating the cutter head torque of a tunnel boring machine according to any one of claims 1 to 7 are implemented.

10. A storage medium comprising a stored program, characterized in that: When the program is running, the device where the storage medium is located is controlled to execute the steps of the method for calculating the cutter head torque of a tunnel boring machine as claimed in any one of claims 1 to 7.

Images