TBM construction permission and risk dynamic assessment and machine jamming early warning method and system
Through the combination of entropy weight method, fuzzy comprehensive evaluation method and Bi-LSTM model, a safety evaluation system for TBM construction was established, which solved the problems of inaccurate risk assessment of card machines and lagging early warning response in the existing technology, and achieved complex geology Improvement of TBM construction safety and economic benefits under conditions.
Patent Information
- Application Number
- CN202411330615.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-24
- Publication Date
- 2025-01-14
AI Technical Summary
In the prior art, the lack of systematic and accurate machine risk assessment and early warning methods during TBM construction process leads to delays in construction progress and safety hazards, especially in complex geological conditions, which is difficult to respond quickly.
The entropy weight method and fuzzy comprehensive evaluation method are used to combine the Bi-LSTM model. By obtaining geological data and operating parameters, a safety evaluation system is established, safety pre-evaluation and real-time dynamic evaluation are carried out before construction, safety level is adjusted dynamically and early warning signals are issued.
It improves the accuracy of TBM card machine risk assessment and timely warning, ensures the safety and economic benefits of the construction process, and is suitable for tunnel construction scenarios under complex geological conditions.
Smart Images
Figure CN119313146A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of TBM construction, and in particular relates to a TBM construction permit and risk dynamic assessment and machine jam warning method and system. Background Art
[0002] Tunnel boring machine (TBM) is one of the key equipment widely used in modern tunnel construction. With its efficient and safe excavation ability, it has an irreplaceable position in the construction of long-distance and deep-buried tunnels. However, TBMs often face complex and changeable geological conditions during construction, especially in weak surrounding rock or high ground stress environments, which are prone to machine jams. TBM jams not only cause construction delays, but may also cause equipment failures, and in serious cases even affect the safety and economic benefits of the entire project.
[0003] In the existing technology, the assessment and early warning methods of TBM jam risk mainly rely on experience judgment or simple parameter monitoring, which lacks systematicness and accuracy. These methods often fail to detect potential risks in a timely manner, resulting in a lack of sufficient foresight during the construction process. Although some progress has been made in the field of data-driven and machine learning in recent years, due to the complexity of data processing and real-time requirements, an effective method that can be widely used in actual projects has not yet been formed.
[0004] In addition, current technical means usually do not make full use of pre-construction safety survey data and the ever-changing real-time data during the construction process for dynamic adjustment and optimization. This makes it difficult for the existing early warning system to respond quickly when dealing with emergencies and complex working conditions, which in turn affects the safe operation of TBMs.
[0005] In view of the above problems, there is an urgent need for a comprehensive method that integrates safety pre-evaluation and real-time monitoring and evaluation to improve the accuracy of TBM jam risk assessment and the timeliness of early warning. Summary of the invention
[0006] The present invention aims to provide a TBM stuck risk dynamic assessment and early warning method to solve the problems existing in the prior art of unclear construction permit determination, insufficient accuracy of TBM stuck risk assessment and delayed early warning response.
[0007] The technical solution of the present invention is as follows:
[0008] A method for TBM construction permit and risk dynamic assessment and machine jam warning, the method comprising:
[0009] Obtain geological data and engineering design plans for the TBM tunnel construction area;
[0010] Based on the geological data and engineering design plan of the tunnel construction area, the safety level of the construction area is preliminarily determined by the entropy weight method and the fuzzy comprehensive evaluation method;
[0011] The operating parameters generated by the TBM during the construction process are obtained, and the operating parameters are input into the trained early warning model to further determine the safety level of the construction area.
[0012] Furthermore, the operating parameters include but are not limited to cycle footage, thrust pressure, total thrust, cutter head speed, cutter head torque, penetration, excavation speed and gripper shoe pressure.
[0013] Furthermore, the engineering design solution includes construction methods and equipment configuration.
[0014] Furthermore, the geological data of the tunnel construction area includes but is not limited to lithology, surrounding rock grade and special geological conditions.
[0015] Furthermore, based on the geological data and engineering design plan of the tunnel construction area, the safety level of the construction area is preliminarily determined by the entropy weight method and the fuzzy comprehensive evaluation method as follows:
[0016] Based on the collected geological data and engineering design plan, a safety evaluation system table is established. Each construction team conducts an initial safety assessment based on the data from the on-site exploration, and the entropy weight method is used to calculate the weight of each safety evaluation index;
[0017] The weights are applied to the fuzzy comprehensive evaluation model to calculate the safety pre-evaluation score before construction;
[0018] The obtained safety pre-evaluation score is analyzed and compared with the set safety level standard to preliminarily judge the safety level of the construction area. If the safety pre-evaluation score is lower than the set safety standard, the construction plan needs to be further adjusted or additional safety measures need to be taken; if the safety pre-evaluation score meets the safety standard, the next step of dynamic evaluation is entered.
[0019] Furthermore, the early warning model is trained and optimized based on the Bi-LSTM model, specifically:
[0020] The Adam optimization algorithm is used to optimize the Bi-LSTM model for training. The learning rate is adjusted to improve the convergence speed and prediction accuracy of the model. The k-fold cross-validation method is used to ensure the generalization ability and robustness of the model on different data sets.
[0021] Furthermore, the operating parameters generated by the TBM during the construction process are obtained, and the operating parameters are input into the trained early warning model to further determine the safety level of the construction area:
[0022] Continuously collect and monitor the real-time operating parameters of the TBM, input them into the trained early warning model, predict the future values of the total thrust and cutter torque, and calculate the residual rate through MSE, RMSE, and residual distribution map;
[0023] According to the changes in the residual rate, the previously calculated safety pre-assessment score is dynamically adjusted. When it drops to the specified standard value, a warning signal is issued, suggesting adjustments to the construction plan or the adoption of protective measures.
[0024] The present invention also provides a TBM construction permit and risk dynamic assessment and machine jam warning system, the system comprising:
[0025] The data acquisition module is used to obtain geological data of the TBM tunnel construction area, engineering design schemes, and operating parameters generated by the TBM during the construction process;
[0026] The safety level judgment module is used to judge the safety level of the construction area.
[0027] Compared with the prior art, the present invention has the following advantages:
[0028] The present invention comprehensively utilizes advanced technical means such as geological condition preliminary investigation, entropy weight method, fuzzy comprehensive evaluation method, neural network model training and real-time monitoring. It is particularly suitable for tunnel construction scenarios under complex geological conditions and proposes a new method for TBM construction evaluation. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] The accompanying drawings generally illustrate various embodiments by way of example and not limitation, and together with the description and claims, serve to illustrate the embodiments of the invention. Where appropriate, the same reference numerals are used throughout the drawings to refer to the same or similar parts. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the present apparatus or method.
[0030] Figure 1 An overall flow chart of the implementation method provided by the embodiment of the present invention;
[0031] Figure 2 A tunnel characteristic diagram provided by an embodiment of the present invention;
[0032] Figure 3 This is the TBM safety performance evaluation index system of the embodiment of the present invention;
[0033] Figure 4 The weight diagram of each indicator of the entropy weight method calculated by the embodiment of the present invention;
[0034] Figure 5 A thermodynamic coefficient diagram provided for an embodiment of the present invention;
[0035] Figure 6A neural network structure diagram provided for an embodiment of the present invention;
[0036] Figure 7 A schematic diagram of a 5-fold cross validation provided by an embodiment of the present invention;
[0037] Figure 8 A neural network training diagram provided by an embodiment of the present invention;
[0038] Fig. 9 The card machine data prediction diagram provided in the embodiment of the present invention is used to verify the model. DETAILED DESCRIPTION
[0039] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0040] The embodiment of the present invention provides a TBM machine jam risk dynamic assessment and early warning method, which specifically includes:
[0041] S1 Safety Pre-investigation and Data Collection:
[0042] Before TBM excavation begins, collect and analyze geological data and engineering design plans for the tunnel construction area. The geological data include but are not limited to lithology, surrounding rock grade, special geological conditions, etc., and the engineering design plan includes construction methods and equipment configuration. Collect data on multiple operating parameters generated by TBM during construction, including but not limited to cycle footage, thrust pressure, total thrust, cutterhead speed, cutterhead torque, penetration, excavation speed, gripper pressure, etc., to provide a basis for subsequent analysis and modeling.
[0043] S2 processes data:
[0044] Based on the collected geological data and engineering design plan, a safety evaluation system table is established, and each construction team conducts an initial safety assessment based on the data from the on-site exploration. The entropy weight method is used to calculate the weight of each safety evaluation index to reduce the subjective influence, including standardization, specific gravity calculation, information entropy calculation, difference coefficient calculation and final weight determination. Then, the fuzzy comprehensive evaluation method is used to conduct a comprehensive evaluation of various indicators before construction, and a preliminary safety pre-evaluation score is obtained to determine the risk level before construction, providing a basis for subsequent dynamic evaluation.
[0045] The key steps are as follows:
[0046] (1) Assume that the data set involved in the evaluation is T = {X1, X2, …, X m} The original data matrix is formed as follows: ij The value of the jth indicator of the i-th sample, so the following matrix X is formed:
[0047]
[0048] (2) In order to eliminate the impact of different dimensions of each evaluation index on the safety evaluation, the decision matrix needs to be standardized to form a standardized matrix Y = {Y1, Y2, …, Y m} as follows ij is the value of the jth indicator of the i-th sample. First, each indicator is standardized. Different types of indicators need to be standardized in different ways:
[0049]
[0050] Therefore, the standardized matrix is as follows:
[0051]
[0052] (3) Calculate X under a certain indicator j The proportion of the i-th sample in the index is the proportion of the j-th sample in the index and the entropy value E formula is as follows:
[0053]
[0054] Where n is the number of samples, 0≤E j ≤1
[0055] (4) Calculate the weight by calculating the difference coefficient, D j is the coefficient of variation W j Weight
[0056]
[0057] The fuzzy comprehensive evaluation method is used to comprehensively evaluate various indicators before construction to obtain a preliminary safety pre-evaluation score. This safety pre-evaluation score is used to determine the risk level before construction begins and provide a basis for subsequent dynamic evaluation. The main steps are as follows:
[0058] (1) Establish the factor set U = {u1,u2,…,u n}, divide the factor set into several groups according to the comments, that is, U = {U1, U2, ..., U i}, so that U = ∪U i ,U i ∩U j =φ(i≠j)i,j=1,2,…,7, it is called U={U1,U2,…,U i} is the first-level factor set.
[0059] (2) Let the first-level factor set U = {U1,U2,…,Ui The weight of} is A={A1,A2,…,A i};
[0060] (3) Set the second-level factor set The weight of i=1,2,…,7,n i The number of secondary indicator factors contained in each primary factor;
[0061] (4) The indicator evaluation set uses fuzzy language to uniformly express the level evaluation of the target problem. The constructed evaluation set is: V = {v1, v2, …, v n}Where V is the indicator evaluation set of the target problem; v i (i=1,2,…,n) is the fuzzy grade evaluation of each indicator of the target problem, and there exists v i ∈[0,1], when v i The closer the value is to 1, the higher the fuzzy evaluation level of the indicator.
[0062] (5) Let the judgment set V = {v1, v2, …, v5}. First, the second-level factor set n j The single factor evaluation matrix is:
[0063]
[0064] Using the fuzzy weighted average method, the fuzzy comprehensive evaluation of the i-th factor is:
[0065] B i =A i ·R i =(b i1 ,b i2 ,…,b i5 )
[0066] (6) The second-level comprehensive evaluation is then performed on the first-level factor set U = {U1, U2, ..., U i} for comprehensive evaluation, the evaluation matrix is a first-level fuzzy comprehensive evaluation matrix
[0067]
[0068] (7) Using the fuzzy weighted average type, the second-level fuzzy comprehensive evaluation is:
[0069] B=A·R=(b1,b2,…,b5)
[0070] (8) Establish a comprehensive evaluation score matrix and assign scores to the comments in turn (F). The final score formula is as follows:
[0071] Z=B*F
[0072] Analysis of S3 security pre-evaluation results:
[0073] The safety pre-evaluation scores obtained based on the entropy weight method and the fuzzy comprehensive evaluation method are analyzed, and compared with the set safety level standards to preliminarily determine the safety level of the construction area. If the safety pre-evaluation score is lower than the set safety standard, the construction plan needs to be further adjusted or additional safety measures need to be taken; if the safety pre-evaluation score meets the safety standard, the next step of dynamic evaluation is entered.
[0074] S4 Bi-LSTM model training and optimization:
[0075] A neural network model based on bidirectional long short-term memory (Bi-LSTM) was designed and constructed to predict the key operating parameters of TBM.
[0076] This model uses the Adam optimizer, and its default learning rate is 0.001. The model architecture includes a BI-LSTM layer and a fully connected layer. To prevent overfitting, the model adds a 0.1 Dropout layer after the BI-LSTM layer and the fully connected layer. The input data consists of two parts: the excavation speed, cutterhead torque, cutterhead speed, and total thrust through the Bi-LSTM layer, and the uniaxial compressive strength, lithology, and special geology through a fully connected layer and after Flatten processing, the features are extracted through the fully connected layer. Then the features of the two parts are spliced, and the cutterhead torque and total thrust are predicted respectively through the output of two fully connected layers. In terms of data partitioning, the model divides the data set into a training set and a test set, with the test set accounting for 20%. This model uses an Early Stopping callback function during training to monitor the validation loss and stop training when the validation loss no longer improves, thereby preventing overfitting. During the training process, 80% of the training data and 20% of the validation data were used for model validation.
[0077] S5 real-time monitoring and dynamic evaluation:
[0078] During the actual TBM construction process, various operating parameters are monitored in real time, and real-time data is input into the trained Bi-LSTM model to dynamically predict the propulsion force and cutter torque of the TBM. The difference between the predicted value and the actual value of the evaluation model is calculated by calculating the residual rate, and the previously calculated safety pre-evaluation score is dynamically adjusted according to the change of the real-time residual rate, and the safety level of the construction area is updated in real time. If the safety score drops significantly, the system will automatically issue a warning signal to indicate the possible risk of machine jamming, and recommend adjusting the construction plan or taking protective measures.
[0079] Example 1
[0080] Before TBM excavation, detailed geological exploration and data collection were carried out in the Xinjiang tunnel construction area. Key geological data such as lithology, surrounding rock grade and special geological conditions were collected. Figure 2 During the excavation process, key operating parameters such as cycle footage, thrust pressure, total thrust, cutter head speed, cutter head torque, penetration, excavation speed, and gripper shoe pressure are monitored and recorded in real time.
[0081] Based on the collected data, a systematic safety evaluation system was established, as shown in 3. All data were standardized to eliminate dimensional differences and ensure the comparability of various indicators. The entropy weight method was used to calculate the weight of each safety evaluation indicator, as shown in Figure 4 The calculation of information entropy reduces the subjective influence and ensures the scientific nature of the evaluation results. Subsequently, the fuzzy comprehensive evaluation method is used to comprehensively evaluate various indicators before construction to obtain a preliminary safety pre-evaluation score. When the safety pre-evaluation score reaches the required score, construction is allowed.
[0082] In the actual construction process, the collected operating parameters are analyzed by Pearson correlation coefficient and redundant features are removed, such as Figure 5 . According to the collected parameters and analysis, build a neural network structure, such as Figure 6 . Cross-validation and early stopping measures are used to improve the accuracy and generalization ability of the model and reduce overfitting, such as Figure 7 . The optimized and trained Bi-LSTM neural network model is input to dynamically predict key parameters such as propulsion force and cutter torque. Figure 8 The training results of the model are displayed, verifying the accuracy and rationality of the model. Fig. 9 , is the data for real-time prediction of the cutter head and total thrust of the unknown segment, from Fig. 9 In (d), it is found that the safety score keeps decreasing when the 39th cycle is reached, and the safety score continues to decrease when the 40th cycle is reached. Therefore, the model believes that the machine will be stuck at the 41st cycle ahead. The system automatically triggers an early warning and recommends taking appropriate protective measures or adjusting the construction plan. By mutual verification with the phase field data, it is found that this method corresponds to the actual situation.
[0083] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A method for TBM construction permit and risk dynamic assessment and machine jam warning, characterized in that: The method comprises: Obtain geological data and engineering design plans for the TBM tunnel construction area; Based on the geological data and engineering design plan of the tunnel construction area, the safety level of the construction area is preliminarily determined by the entropy weight method and the fuzzy comprehensive evaluation method; The operating parameters generated by the TBM during the construction process are obtained, and the operating parameters are input into the trained early warning model to further determine the safety level of the construction area.
2. The TBM construction permit and risk dynamic assessment and machine jam warning method according to claim 1 is characterized in that: The operating parameters include but are not limited to cycle footage, thrust pressure, total thrust force, cutter head speed, cutter head torque, penetration, excavation speed and support shoe pressure.
3. The TBM construction permit and risk dynamic assessment and machine jam warning method according to claim 1 is characterized in that: The engineering design solution includes construction methods and equipment configuration.
4. The TBM construction permit and risk dynamic assessment and machine jam warning method according to claim 1 is characterized in that: The geological data of the tunnel construction area include but are not limited to lithology, surrounding rock grade and special geological conditions.
5. The TBM construction permit and risk dynamic assessment and machine jam warning method according to claim 1 is characterized in that: Based on the geological data and engineering design plan of the tunnel construction area, the safety level of the construction area is preliminarily determined by the entropy weight method and fuzzy comprehensive evaluation method as follows: Based on the collected geological data and engineering design plan, a safety evaluation system table is established. Each construction team conducts an initial safety assessment based on the data from the on-site exploration, and the entropy weight method is used to calculate the weight of each safety evaluation index; The weights are applied to the fuzzy comprehensive evaluation model to calculate the safety pre-evaluation score before construction; The obtained safety pre-evaluation score is analyzed and compared with the set safety level standard to preliminarily judge the safety level of the construction area. If the safety pre-evaluation score is lower than the set safety standard, the construction plan needs to be further adjusted or additional safety measures need to be taken; if the safety pre-evaluation score meets the safety standard, the next step of dynamic evaluation is entered.
6. The TBM construction permit and risk dynamic assessment and machine jam warning method according to claim 1 is characterized in that: The early warning model is based on Bi-LSTM model training and optimization, specifically: The Bi-LSTM model is optimized and trained using the Adam optimization algorithm. The learning rate is adjusted to improve the convergence speed and prediction accuracy of the model. The early stopping callback function is used to monitor the validation loss and stop training when the validation loss no longer improves, thereby preventing overfitting.
7. The TBM construction permit and risk dynamic assessment and machine jam warning method according to claim 1 is characterized in that: The operating parameters generated by the TBM during the construction process are obtained, and the operating parameters are input into the trained early warning model to further determine the safety level of the construction area: Continuously collect and monitor the real-time operating parameters of the TBM, input them into the trained early warning model, predict the future values of the total thrust and cutter torque, and calculate the residual rate through MSE, RMSE, and residual distribution map; According to the changes in the residual rate, the previously calculated safety pre-assessment score is dynamically adjusted. When it drops to the specified standard value, a warning signal is issued to adjust the construction plan or take protective measures.
8. A TBM construction permit and risk dynamic assessment and machine jam warning system, characterized in that: The system comprises: The data acquisition module is used to obtain geological data of the TBM tunnel construction area, engineering design schemes, and operating parameters generated by the TBM during the construction process; The safety level judgment module is used to judge the safety level of the construction area.