Risk Dynamic Assessment Method Based on TBM Operating Parameters

Abstract

The invention belongs to the field of tunnel boring machine construction, and provides a risk dynamic assessment method based on TBM operation parameters, comprising: obtaining TBM operation parameter data, using the TOPSIS method to comprehensively evaluate the data of each cycle, and generating corresponding evaluation scores; constructing and training a first neural network model, with the tunneling speed, cutter head torque and total propulsion force of the TBM as inputs, and the evaluation scores as outputs; constructing and training a second neural network model, with future values ​​of three features as inputs of the first neural network model; inputting the future values ​​predicted by the second neural network model into the first neural network model, so as to dynamically output the surrounding rock adaptability evaluation score of the TBM in real time; the invention can predict the tunneling speed, cutter head torque and total propulsion force of the TBM in real time, and output a dynamic surrounding rock adaptability evaluation score accordingly, so as to realize adaptive adjustment of a construction plan, and ensure the safety and efficiency of the construction.

Description

Technical Field

[0001] The invention belongs to the field of tunnel boring machine construction, and in particular relates to a risk dynamic assessment method based on TBM operation parameters. Background Art

[0002] Tunnel boring machine (TBM) is a key equipment in modern tunnel construction and is widely used in tunnel projects with long distance, deep burial and complex geological conditions. TBM construction has the advantages of high efficiency and safety. However, due to the complex and changeable geological conditions, TBM often faces the challenge of surrounding rock changes during construction, especially in weak surrounding rock, fault fracture zone and high ground stress area, which may lead to reduced excavation efficiency, equipment jamming and even construction accidents. Therefore, surrounding rock adaptability evaluation is crucial in TBM construction.

[0003] In the existing technology, the evaluation of surrounding rock adaptability in TBM construction usually relies on experience judgment or static analysis based on a single parameter, which fails to make full use of the real-time data that changes constantly during the construction process and makes it difficult to accurately predict the construction status under complex geological conditions. This method that relies on experience or a single parameter usually lacks sufficient foresight when geological conditions change dramatically, which easily leads to the construction plan lagging behind the actual situation, reducing construction efficiency and increasing safety risks.

[0004] With the rapid development of data-driven technology and artificial intelligence, prediction and evaluation models based on neural networks provide new solutions for the adaptability of TBM surrounding rocks. By building a multidimensional data model and using historical operation data and real-time monitoring data to conduct dynamic prediction and comprehensive evaluation, the adaptability of TBM under complex geological conditions can be more accurately reflected. At the same time, neural network technologies such as bidirectional long short-term memory networks (Bi-LSTM) perform well in processing time series data and can effectively capture the temporal dependence of operating parameters, thereby improving prediction accuracy and model robustness.

[0005] However, the existing technology has not yet fully utilized this neural network technology to predict TBM operating parameters in real time and combine it with surrounding rock adaptability evaluation. Therefore, there is an urgent need for a comprehensive method that integrates neural network prediction and dynamic evaluation, which can predict the TBM's excavation speed, cutterhead torque and total propulsion force in real time, and output a dynamic surrounding rock adaptability evaluation score based on this, so as to achieve adaptive adjustment of the construction plan and ensure the safety and efficiency of the construction. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a dynamic risk assessment method based on TBM operation parameters to solve the problems in the prior art. The technical solution adopted by the present invention is:

[0007] The dynamic risk assessment method based on TBM operation parameters includes the following steps:

[0008] Step 1: Obtain and continuously collect the operating parameter data of each cycle of the TBM during normal excavation, including the excavation speed, cutter head torque and total propulsion force, and establish a complete database;

[0009] Step 2: Process the operating parameter data by using the coefficient of variation method, calculate the coefficient of variation of each operating parameter, and use the TOPSIS method to comprehensively evaluate the data of each cycle to generate the corresponding evaluation score;

[0010] Step 3: Build and train the first neural network model, taking the TBM’s excavation speed, cutterhead torque, and total thrust as inputs and the evaluation score as outputs to optimize the weight parameters of the neural network.

[0011] Step 4, construct and train a second neural network model, using historical data to predict the future values ​​of three characteristics of the TBM: tunneling speed, cutterhead torque, and total thrust, as input to the first neural network model;

[0012] Step 5, inputting the tunneling speed, cutter head torque, and total propulsion force predicted by the second neural network model into the first neural network model to dynamically output the surrounding rock adaptability evaluation score of the TBM in real time;

[0013] Step 6: During the construction process, the actual and predicted operating parameters of the TBM are input in real time to dynamically adjust the excavation parameters of the TBM and realize dynamic evaluation of the surrounding rock adaptability.

[0014] Furthermore, the operating parameter data of each cycle includes excavation speed, cutter head torque and total propulsion force.

[0015] Furthermore, in step 2, the calculation formula of the coefficient of variation is:

[0016] ;

[0017] is the coefficient of variation of the ith operating parameter, is the standard deviation of the ith operating parameter, is the mean of the ith operating parameter.

[0018] Furthermore, the TOPSIS method is used to comprehensively evaluate the data of each cycle and generate corresponding evaluation scores, including the following steps:

[0019] Step 2.1: Standardize the TBM operating parameters. The standardization formula is:

[0020] ;

[0021] is the normalized value of the ith parameter at the jth cycle,

[0022] is the original value of the ith parameter at the jth loop;

[0023] Step 2.2: Calculate the positive and negative ideal solutions, the formula is as follows:

[0024] ;

[0025] ;

[0026] in, is the maximum value after normalization, is the minimum value after standardization;

[0027] Step 2.3: Calculate the Euclidean distance of each cycle's data to the positive ideal solution and the negative ideal solution respectively:

[0028] ;

[0029] ;

[0030] Among them, the distance between the i-th evaluation object and the maximum value is defined as , define the distance between the i-th evaluation object and the minimum value as ;

[0031] Step 2.4: Calculate the comprehensive evaluation score for each cycle using the following formula:

[0032] ;

[0033] Fraction It is used to indicate the adaptability score of the surrounding rock in each cycle. The closer the value is to 1, the more ideal it is, and the closer the value is to 0, the worse the adaptability of the surrounding rock is.

[0034] Furthermore, step 3 includes the following steps:

[0035] Step 3.1: The excavation speed, cutter head torque and total thrust of the TBM are used as inputs of the first neural network model. The network structure of the first neural network model includes two hidden layers, each layer includes 128 neurons;

[0036] Step 3.2: The first neural network model uses the Adam optimization algorithm, the initial learning rate is set to 0.001, and the loss function selects the mean square error to measure the difference between the predicted value and the true value;

[0037] Step 3.3: During the training of the first neural network model, the historical data set is divided into a training set and a validation set. The network weights and biases are optimized through repeated iterations to gradually improve the prediction performance.

[0038] Step 3.4: After training, the first neural network model is applied in a real-time environment to output the surrounding rock adaptability evaluation score based on the current operating parameters of the TBM.

[0039] Furthermore, step 5 includes the following steps:

[0040] Step 5.1: The input of the second neural network model is the historical operating parameters of the TBM in the past several time steps, including the excavation speed, cutterhead torque and total thrust;

[0041] Step 5.2: The second neural network model adopts the structure design of LSTM network, including one LSTM hidden layer, including 64 neurons, and one fully connected layer after the LSTM hidden layer, outputting future operating parameters, and using linear activation function to make the output a continuous value;

[0042] Step 5.3: The second neural network model is trained using the Adam optimizer. The initial learning rate is set to 0.0005. The loss function is the mean square error, which is used to measure the difference between the model's predicted value and the actual value. During the training process, the data is divided into a training set and a validation set. The model weights and bias parameters are optimized through repeated iterations.

[0043] Step 5.4: After the training is completed, the second neural network model predicts the future parameters based on the input historical operation data, providing a basis for dynamic adjustment of TBM operation; Step 5.4: The predicted future operation parameters are input into the first neural network to achieve real-time dynamic update of TBM adaptability evaluation.

[0044] The present invention has the following beneficial effects: the present invention is particularly suitable for TBM construction under complex geological conditions such as weak surrounding rock, fault fracture zone, etc. Through the real-time dynamic evaluation system, the risk of surrounding rock inadaptability can be identified in advance, equipment jamming and other construction accidents can be avoided, and construction efficiency and safety can be significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 An overall flow chart of the implementation method provided by the embodiment of the present invention;

[0046] Figure 2 It is the neural network topology diagram of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the embodiments of the present invention Figure 1-Figure 2, the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art.

[0048] By introducing neural networks and real-time data monitoring means, the present invention can continuously monitor and comprehensively evaluate the excavation process of the TBM, and dynamically adjust the construction plan according to the prediction results, including:

[0049] Data collection and processing: Continuously collect the operating parameters of each cycle of the TBM during normal excavation, including excavation speed, cutterhead torque and total thrust, and establish an operating parameter database. The coefficient of variation method is used to standardize these operating parameters, and TOPSIS (topic ideal solution ranking method) is used to conduct a multi-index comprehensive evaluation of each cycle to generate an evaluation score for each cycle.

[0050] Neural network model construction: The present invention designs two neural network models. The input of the first neural network model is the operating parameters of the TBM (excavation speed, cutterhead torque, and total propulsion force), and the output is the comprehensive evaluation score for each cycle. The second neural network model predicts the TBM operating parameters (excavation speed, cutterhead torque, and total propulsion force) at the next moment by analyzing historical data. The parameter results predicted by the second neural network will be input into the first neural network to achieve a dynamic evaluation of the adaptability of the surrounding rock during the TBM construction process.

[0051] Model training and optimization: supervised learning of neural networks is performed through a large amount of historical data. Bi-LSTM (bidirectional long short-term memory) network is used to analyze time series data, which can capture the temporal dependencies of TBM operation parameters, thereby improving the prediction accuracy of future parameters. At the same time, through the optimization of loss function (such as minimizing mean square error MSE) and early stopping mechanism, the model is prevented from overfitting and its generalization ability is improved.

[0052] Dynamic adaptability evaluation and construction optimization: During the actual construction process, the operating parameters of the TBM are collected in real time and input into the neural network model to output a real-time surrounding rock adaptability evaluation score. According to the score, construction parameters such as excavation speed, cutter head torque and total propulsion force are dynamically adjusted to optimize the construction plan and ensure the safety and efficiency of TBM construction under complex geological conditions. The present invention is particularly suitable for TBM construction under complex geological conditions such as weak surrounding rock and fault fracture zones. Through the real-time dynamic evaluation system, the risk of surrounding rock inadaptability can be identified in advance, equipment jams and other construction accidents can be avoided, and construction efficiency and safety can be significantly improved.

[0053] Specifically, the present invention provides a risk dynamic assessment method based on TBM operation parameters, comprising the following steps:

[0054] Step 1: Obtain and continuously collect the operating parameter data of each cycle of the TBM (tunnel boring machine) during normal excavation, including excavation speed, cutter head torque and total propulsion force, and establish a complete database;

[0055] Step 2: Process the operating parameter data by using the coefficient of variation method, calculate the coefficient of variation of each operating parameter, and use the TOPSIS method (topic ideal solution ranking method) to comprehensively evaluate the data of each cycle and generate the corresponding evaluation score; the coefficient of variation method is used to calculate the degree of dispersion of each operating parameter and provide the weight basis of each parameter to input into the TOPSIS evaluation model. The TOPSIS method is used to construct the optimal solution and the worst solution, and calculate the distance to the positive ideal solution and the negative ideal solution based on the data of each cycle, so as to determine the evaluation score of each cycle.

[0056] Step 3, construct and train the first neural network model, taking the TBM's excavation speed, cutterhead torque and total thrust as input and the evaluation score as output, and optimize the weight parameters of the neural network; the first neural network model is trained through the three characteristics of excavation speed, cutterhead torque and total thrust to generate the corresponding surrounding rock adaptability evaluation score.

[0057] Step 4, construct and train a second neural network model, using historical data to predict the future values ​​of three characteristics of the TBM: tunneling speed, cutterhead torque, and total thrust, as input to the first neural network model;

[0058] Step 5: Input the tunneling speed, cutterhead torque, and total propulsion force predicted by the second neural network model into the first neural network model to dynamically output the surrounding rock adaptability evaluation score of the TBM in real time; the second neural network predicts the tunneling speed, cutterhead torque, and total propulsion force in future TBM operations through time series analysis of historical operation data. The second neural network model dynamically adjusts the prediction parameters through the error feedback mechanism and inputs the prediction results into the first neural network model to achieve dynamic evaluation of TBM adaptability.

[0059] Step 6: During the construction process, the actual and predicted operating parameters of the TBM are input in real time to dynamically adjust the excavation parameters of the TBM and realize dynamic evaluation of the adaptability of the surrounding rock. The real-time application of the neural network model can realize dynamic adjustment of construction parameters under complex geological conditions by continuously inputting and outputting the excavation speed, cutter torque and total thrust of the TBM, thereby improving construction safety and efficiency.

[0060] Furthermore, the operating parameter data of each cycle includes excavation speed, cutter head torque and total propulsion force.

[0061] Furthermore, in step 2, the calculation formula of the coefficient of variation is:

[0062] ;

[0063] is the coefficient of variation of the ith operating parameter, is the standard deviation of the ith operating parameter, is the mean of the ith operating parameter. By calculating the coefficient of variation, we can measure the volatility of each parameter and provide a weight basis for the subsequent TOPSIS method.

[0064] Furthermore, the TOPSIS method is used to comprehensively evaluate the data of each cycle and generate corresponding evaluation scores, including the following steps:

[0065] Step 2.1: Standardize the TBM operating parameters. The standardization formula is:

[0066] ;

[0067] is the normalized value of the ith parameter at the jth cycle,

[0068] is the original value of the ith parameter at the jth loop;

[0069] Step 2.2: Calculate the positive and negative ideal solutions, the formula is as follows:

[0070] ;

[0071] ;

[0072] is the maximum value after normalization, is the minimum value after standardization; the profitability index is as large as possible, and the cost index is as small as possible.

[0073] Step 2.3: Calculate the Euclidean distance of each cycle's data to the positive ideal solution and the negative ideal solution respectively:

[0074] ;

[0075] ;

[0076] Define the distance between the i-th evaluation object and the maximum value as , define the distance between the i-th evaluation object and the minimum value as .

[0077] Step 2.4: Calculate the comprehensive evaluation score for each cycle using the following formula:

[0078] ;

[0079] Fraction It is used to indicate the adaptability score of the surrounding rock in each cycle. The closer the value is to 1, the more ideal it is, and the closer the value is to 0, the worse the adaptability of the surrounding rock is.

[0080] like Figure 1 , Figure 2 , step 3 includes the following steps:

[0081] Step 3.1: The excavation speed, cutterhead torque and total thrust of the TBM are used as the input of the network. The network structure includes two hidden layers, each layer includes 128 neurons, and the ReLU activation function is used to enhance the model's ability to handle nonlinear features. In order to ensure that the output value is a continuous evaluation score, the output layer uses a linear activation function, allowing the output to be any real number;

[0082] Step 3.2: The neural network is designed to complete the regression task, and its goal is to predict the scores related to the adaptability of the surrounding rock based on the input operating parameters. In order to improve the training efficiency and prevent overfitting, the network adopts the Adam optimization algorithm, the initial learning rate is set to 0.001, and the generalization ability of the model is further improved by introducing L2 regularization. The loss function selects the mean square error (MSE) to measure the difference between the predicted value and the true value;

[0083] Step 3.3: During the training of the first neural network model, the historical data set is divided into a training set and a validation set. The network weights and biases are optimized through repeated iterations to gradually improve the prediction performance.

[0084] Step 3.4: After training, the first neural network model is applied in a real-time environment to output the surrounding rock adaptability evaluation score according to the current operating parameters of the TBM, thereby providing important decision support for parameter adjustment during the construction process.

[0085] Furthermore, step 5 includes the following steps:

[0086] Step 5.1: The input of the second neural network model is the historical operating parameters of the TBM in the past several time steps, including the excavation speed, cutterhead torque and total thrust;

[0087] Step 5.2: The second neural network model adopts the structure of LSTM network, including one LSTM hidden layer, including 64 neurons, which can capture the time dependency in the input sequence. The LSTM hidden layer is followed by a fully connected layer to output the future operation parameters (excavation speed, cutter head torque and total thrust), and a linear activation function is used to make the output a continuous value.

[0088] Step 5.3: The network is trained using the Adam optimizer, with an initial learning rate of 0.0005, and a Dropout mechanism is added with a dropout rate of 0.2 to reduce the risk of overfitting. The loss function is the mean square error (MSE), which is used to measure the difference between the model's predicted value and the actual value. During the training process, the data is divided into a training set and a validation set, and the model weights and bias parameters are optimized through repeated iterations;

[0089] Step 5.4: After training is completed, the second neural network model predicts future parameters based on the input historical operation data, providing a basis for dynamic adjustment of TBM operation;

[0090] Step 5.4: Input the predicted future operating parameters into the first neural network to achieve real-time dynamic update of TBM adaptability evaluation.

[0091] The embodiments described above are only descriptions of the preferred modes of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations, modifications, and substitutions made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A dynamic risk assessment method based on TBM operation parameters, characterized in that: The following steps are involved: Step 1: Obtain and continuously collect the operating parameter data of each cycle of the TBM during normal excavation, including the excavation speed, cutter head torque and total propulsion force, and establish a complete database; Step 2: Process the operating parameter data by using the coefficient of variation method, calculate the coefficient of variation of each operating parameter, and use the TOPSIS method to comprehensively evaluate the data of each cycle to generate the corresponding evaluation score; Step 3: Build and train the first neural network model, taking the TBM’s excavation speed, cutterhead torque, and total thrust as inputs and the evaluation score as outputs to optimize the weight parameters of the neural network. Step 4, construct and train a second neural network model, using historical data to predict the future values ​​of three characteristics of the TBM: tunneling speed, cutterhead torque, and total thrust, as input to the first neural network model; Step 5, inputting the tunneling speed, cutter head torque, and total propulsion force predicted by the second neural network model into the first neural network model to dynamically output the surrounding rock adaptability evaluation score of the TBM in real time; Step 6: During the construction process, the actual and predicted operating parameters of the TBM are input in real time to dynamically adjust the excavation parameters of the TBM and realize dynamic evaluation of the surrounding rock adaptability; In step 2, the coefficient of variation is calculated as: ； is the coefficient of variation of the ith operating parameter, is the standard deviation of the ith operating parameter, is the mean of the ith operating parameter; In step 2, the TOPSIS method is used to comprehensively evaluate the data of each cycle and generate corresponding evaluation scores, including the following steps: Step 2.1: Standardize the TBM operating parameters. The standardization formula is: ； is the normalized value of the ith parameter at the jth cycle, is the original value of the ith parameter at the jth loop; Step 2.2: Calculate the positive and negative ideal solutions, the formula is as follows: ； in, is the maximum value after normalization, is the minimum value after standardization; Step 2.3: Calculate the Euclidean distance of each cycle's data to the positive ideal solution and the negative ideal solution respectively: ； Among them, the distance between the i-th evaluation object and the maximum value is defined as , define the distance between the i-th evaluation object and the minimum value as ; Step 2.4: Calculate the comprehensive evaluation score for each cycle using the following formula: ； Fraction It is used to indicate the adaptability score of the surrounding rock in each cycle. The closer the value is to 1, the more ideal it is, and the closer it is to 0, the worse the adaptability of the surrounding rock is. Step 3 includes the following steps: Step 3.1: The excavation speed, cutter head torque and total thrust of the TBM are used as inputs of the first neural network model. The network structure of the first neural network model includes two hidden layers, each layer includes 128 neurons; Step 3.2: The first neural network model uses the Adam optimization algorithm, the initial learning rate is set to 0.001, and the loss function selects the mean square error to measure the difference between the predicted value and the true value; Step 3.3: During the training of the first neural network model, the historical data set is divided into a training set and a validation set. The network weights and biases are optimized through repeated iterations to gradually improve the prediction performance. Step 3.4: After training, the first neural network model is applied in a real-time environment to output the surrounding rock adaptability evaluation score based on the current operating parameters of the TBM; Step 5 includes the following steps: Step 5.1: The input of the second neural network model is the historical operating parameters of the TBM in the past several time steps, including the excavation speed, cutterhead torque and total thrust; Step 5.2: The second neural network model adopts the structure design of LSTM network, including one LSTM hidden layer, including 64 neurons, and one fully connected layer after the LSTM hidden layer, outputting future operating parameters, and using linear activation function to make the output a continuous value; Step 5.3: The second neural network model is trained using the Adam optimizer. The initial learning rate is set to 0.0005. The loss function is the mean square error, which is used to measure the difference between the model's predicted value and the actual value. During the training process, the data is divided into a training set and a validation set. The model weights and bias parameters are optimized through repeated iterations. Step 5.4: After training is completed, the second neural network model predicts future parameters based on the input historical operation data, providing a basis for dynamic adjustment of TBM operation; Step 5.4: Input the predicted future operating parameters into the first neural network to achieve real-time dynamic update of TBM adaptability evaluation.

2. The method for dynamic risk assessment based on TBM operation parameters according to claim 1 is characterized in that: In step 1, the operating parameter data of each cycle includes the excavation speed, cutter head torque and total propulsion force.

Images