TBM surrounding rock type identification and operation control method

Abstract

The application discloses a TBM surrounding rock type identification and operation control method. The method comprises the following steps: obtaining historical characteristic data of a TBM and performing data cleaning to construct a training set; establishing a surrounding rock type identification model and inputting the training set into training; constructing a rotation speed operation table and a pushing speed operation table; establishing a rotation speed joint model and a pushing speed joint model; inputting real-time collected characteristic data of the TBM into the identification model after data cleaning to output a surrounding rock type; extracting a cutter disc rotation speed and a pushing speed from the rotation speed operation table and the pushing speed operation table and inputting the cutter disc rotation speed and the pushing speed into the joint model to output a joint cutter disc rotation speed and a joint pushing speed, and performing operation control on the TBM. The method can significantly improve the accuracy of surrounding rock type identification, intelligently adjust parameters of the TBM under different surrounding rock conditions, improve tunneling efficiency and safety, and provide an intelligent and comprehensive solution for operation control, thereby bringing more efficient and safe construction practice to the field of tunnel engineering and promoting intelligent development of the TBM technology.

Description

Technical Field

[0001] This invention relates to a TBM operation control method, specifically to a method for identifying and controlling the surrounding rock type of a TBM. Background Technology

[0002] The type of surrounding rock has a significant impact on the performance of tunnel boring machines (TBMs) in engineering construction. Therefore, during tunneling, operators need to adjust the TBM's operating parameters according to the current surrounding rock conditions to ensure both efficiency and safety. By observing the tunneling load data recorded by sensors, operators can promptly understand the TBM's performance under different surrounding rock types and make corresponding adjustments. Accurate identification of the surrounding rock type is a crucial step in achieving intelligent adjustment of operating parameters. One current research focus is on methods for identifying surrounding rock types. These methods comprehensively consider information from multiple aspects, including geological features and operating parameters, and employ statistical and machine learning techniques to establish accurate and comprehensive identification models. Through these models, operators can obtain accurate identification results of the current surrounding rock type in real-time monitoring, providing a basis for subsequent operating parameter decisions.

[0003] After identifying the surrounding rock type, the decision-making system needs to further select appropriate operating parameters to adapt to the current surrounding rock conditions. This involves optimizing the TBM's tunneling load based on the characteristics of the surrounding rock type. Different surrounding rock types may exert different forces on the TBM, therefore, operating parameters such as the cutterhead feed rate and cutterhead rotation speed need to be adjusted according to specific circumstances. Such adjustments to operating parameters aim to ensure that the TBM can tunnel stably and efficiently under various surrounding rock conditions. Several decision-making schemes already exist to guide TBM tunneling, emphasizing the application of actual engineering data. Through the analysis of actual data, the decision-making system can more accurately understand the TBM's performance under different surrounding rock types and adjust operating parameters accordingly. This data-driven approach not only improves the accuracy of the decision-making system but also enhances its applicability in different engineering scenarios.

[0004] The main challenges faced by TBMs in tunneling through surrounding rock stem from the complexity and variability of the surrounding rock types. Current research on surrounding rock identification faces several key issues, one of which is the insufficient sample size, making accurate identification of rock types difficult. Furthermore, current analysis primarily relies on laboratory data, which is not only time-consuming and labor-intensive but also unsuitable for real-time guidance of the TBM tunneling process. Regarding operational parameter decision-making, the main challenges lie in the complexity of model structures and algorithms. Current research often employs highly complex models and algorithms that may perform well only under specific geological conditions, while their universality in other scenarios remains to be verified. Therefore, further research is needed to simplify and optimize operational parameter decision-making models to improve their practicality under different geological conditions. Summary of the Invention

[0005] To address the problems existing in the background technology, this invention provides a method for identifying and controlling the surrounding rock type of a TBM (Tunnel Boring Machine). The method proposed in this invention achieves accurate identification of the surrounding rock type based on multi-source actual engineering data, and simultaneously realizes intelligent decision-making for operating parameters, providing a reliable reference for the working status of the TBM.

[0006] The technical solution adopted in this invention is:

[0007] The TBM surrounding rock type identification and operation control method of the present invention includes:

[0008] 1) Obtain historical feature data with high correlation between tunnel boring machine (TBM) and different surrounding rock types. After cleaning, obtain a series of historical stable segment data. Combine the series of historical stable segment data and their corresponding surrounding rock types to construct a training set.

[0009] 2) Establish a TBM surrounding rock type identification model. By inputting the training set into the TBM surrounding rock type identification model for training, a complete TBM surrounding rock type identification model can be obtained.

[0010] 3) Based on the surrounding rock type and the tunneling load data of the tunnel boring machine (TBM), construct a speed operation table based on the relationship between surrounding rock type, cutterhead torque, total thrust and cutterhead speed, and construct a thrust speed operation table based on the relationship between surrounding rock type, cutterhead torque, total thrust and thrust speed.

[0011] 4) Establish a combined speed model and a combined thrust model.

[0012] 5) Use sensors to collect feature data of the tunnel boring machine (TBM) in real time, and perform the same data cleaning process as in step 1) to obtain several real-time stable segment data. Input the real-time stable segment data into the complete TBM surrounding rock type identification model. After processing by the complete TBM surrounding rock type identification model, output the surrounding rock type currently being excavated by the TBM, thereby realizing the identification of the surrounding rock type of the TBM.

[0013] 6) Based on the real-time cutterhead torque and real-time total thrust of the tunnel boring machine (TBM) under the current surrounding rock type, extract the first and second cutterhead speeds from the speed operation table, and extract the first and second propulsion speeds from the propulsion speed operation table. Input the first and second cutterhead speeds into the speed joint model, which outputs the joint cutterhead speed of the TBM. Input the first and second propulsion speeds into the propulsion speed joint model, which outputs the joint propulsion speed of the TBM. Control the cutterhead speed and propulsion speed of the TBM at the next moment based on the joint cutterhead speed and joint propulsion speed to achieve the operation control of the TBM.

[0014] In step 1), the historical characteristic data of the tunnel boring machine (TBM) that are highly correlated with different surrounding rock types include the cutterhead rotation speed n, propulsion speed v, cutterhead torque T, and total propulsion force F of the TBM. Each subset of data consisting of cutterhead rotation speed n, propulsion speed v, cutterhead torque T, and total propulsion force F corresponds to a type of surrounding rock R that the TBM is tunneling in.

[0015] In step 1), data cleaning specifically involves sequentially removing non-tunneling state data, removing abnormal state data, dividing the tunneling segment and removing shorter tunneling segment data, separating stable tunneling stage data, and smoothing the data.

[0016] The non-tunneling state data to be removed includes data on the TBM's stop and step-change states during operation, as the TBM has no interaction with the surrounding rock at these times and needs to be removed; data where the cutterhead torque T is 0, as the cutterhead speed, feed rate, and total feed force are all zero at this time, and this data is removed accordingly; and abnormal state data to be removed includes data where the TBM's feed rate v is greater than the abnormal threshold th. ab The data is due to sensor anomalies. Specifically, the process of dividing the tunneling segment and removing shorter segments involves taking the moment when the TBM's propulsion speed v starts increasing from 0 as the starting point of the segment and the moment when v decreases to 0 as the ending point. Then, segments with durations shorter than the shortest time threshold t are removed. dFor tunneling sections with a travel time of less than 1 second; separating the stable tunneling phase data specifically involves, for each divided tunneling section, the first t of the tunneling section... s Data removal in seconds, starting from the t-th second. s Starting from a second, data for the stable tunneling stage is separated. Finally, the data for each stable tunneling stage is smoothed to obtain its respective historical stable segment data. This historical stable segment data and its corresponding surrounding rock type are then used to construct a training set. In practice, the data is divided into a training set and a test set according to a preset ratio.

[0017] In step 2), the established TBM surrounding rock type identification model is specifically a machine learning model, selected from the random forest algorithm;

[0018] During training, the TBM surrounding rock type identification model uses the cutterhead rotation speed n, feed speed v, cutterhead torque T, and total propulsion force F from the training set as inputs and the surrounding rock type R as output. The number of decision trees for the TBM surrounding rock type identification model is preset, and the model is trained to obtain a complete TBM surrounding rock type identification model.

[0019] The accuracy A of the complete TBM surrounding rock type identification model is evaluated using a test set. Accuracy A is equal to the total number of correctly classified samples in the test set divided by the total number of samples in the test set, as follows:

[0020]

[0021] Where A is the recognition accuracy of the test set, and R is... t R represents a single sample in the test set that is correctly classified.

[0022] When the accuracy A exceeds the preset accuracy threshold, the current TBM surrounding rock type identification model is taken as the final complete TBM surrounding rock type identification model.

[0023] In step 3), based on the surrounding rock type and the tunneling load data of the tunnel boring machine (TBM), the cutterhead torque T under each surrounding rock category is set according to a preset torque value T. stage Increment, the total thrust F is increased according to the preset thrust value F. stageThe cutterhead speed is extracted incrementally within each torque range and each thrust range, and then averaged to obtain the average cutterhead speed within each torque range and each thrust range. Using the range of cutterhead torque T as the x-axis and the average cutterhead speed within the corresponding torque range as the y-axis, the cutterhead speed under different surrounding rock conditions as the cutterhead torque T increases is obtained. Similarly, using the total thrust F as the x-axis and the average cutterhead speed within the corresponding thrust range as the y-axis, the cutterhead speed under different surrounding rock conditions as the total thrust F increases is obtained. Finally, a speed operation table is constructed based on the relationship between surrounding rock type, TBM cutterhead torque, total thrust, and cutterhead speed.

[0024] The cutterhead torque T under the surrounding rock category is set according to the preset torque value T. stage Increment, the total thrust F is increased according to the preset thrust value F. stage The propulsion speed is extracted incrementally within each torque range and each thrust range, and then averaged to obtain the average propulsion speed within each torque range and each thrust range. Using the range of cutterhead torque T as the x-axis and the average propulsion speed within the corresponding torque range as the y-axis, the propulsion speed under different surrounding rock conditions as the cutterhead torque T increases is obtained. Simultaneously, using the total thrust F as the x-axis and the average propulsion speed within the corresponding thrust range as the y-axis, the propulsion speed under different surrounding rock conditions as the total thrust F increases is obtained. Finally, a propulsion speed operation table based on the relationship between surrounding rock type, TBM cutterhead torque, total thrust, and propulsion speed is constructed.

[0025] In step 4), the combined rotational speed model is as follows:

[0026] n r =an1+bn2

[0027] Where, n r denoted as the combined cutterhead speed of the tunnel boring machine (TBM); a and b are the first and second speed coefficients, respectively; n1 and n2 are the first and second cutterhead speeds of the TBM, respectively.

[0028] In step 4), the thrust-speed joint model is as follows:

[0029] v r =cv1+dv2

[0030] Among them, v r denoted as , where is the combined propulsion speed of the tunnel boring machine (TBM); c and d are the first and second propulsion speed coefficients, respectively; v1 and v2 are the first and second propulsion speeds, respectively.

[0031] The beneficial effects of this invention are:

[0032] This invention's method fully utilizes data from real engineering sites, overcoming the limitation of limited sample sizes in current rock formation identification research and improving the reliability of rock type identification. Simultaneously, by combining practical operational data with advanced identification models, it provides a reliable reference for the working status of TBMs, enabling them to adapt more flexibly to different engineering scenarios.

[0033] This invention effectively models the complex nonlinear relationship between surrounding rock type and tunneling parameters, significantly improving the accuracy of surrounding rock type identification. Simultaneously, it enables more intelligent adjustment of the TBM's advance speed and cutterhead rotation speed under different surrounding rock conditions, thereby adapting to the characteristics of the surrounding rock and improving tunneling efficiency and safety. This invention provides an intelligent and comprehensive solution for operation control, bringing more efficient and safer construction practices to the tunnel engineering field and promoting the intelligent development of TBM technology. Attached Figure Description

[0034] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention.

[0035] Figure 2 This is the operating parameter curve for a certain tunneling section in the embodiment. Detailed Implementation

[0036] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The technical features of each embodiment of the present invention can be combined accordingly without conflict.

[0037] like Figure 1 As shown, the TBM surrounding rock type identification and operation control method of the present invention includes:

[0038] 1) Obtain historical feature data with high correlation between tunnel boring machine (TBM) and different surrounding rock types. After cleaning, obtain a series of historical stable segment data. Combine the series of historical stable segment data and their corresponding surrounding rock types to construct a training set.

[0039] In step 1), the historical feature data of the tunnel boring machine (TBM) with high correlation with different surrounding rock types include the cutterhead rotation speed n, propulsion speed v, cutterhead torque T, and total propulsion force F of the TBM. Each subset of data consisting of cutterhead rotation speed n, propulsion speed v, cutterhead torque T, and total propulsion force F corresponds to a type of surrounding rock R that the TBM is tunneling in.

[0040] In step 1), data cleaning specifically involves sequentially removing non-tunneling state data, removing abnormal state data, dividing the tunneling segments and removing shorter tunneling segment data, separating stable tunneling stage data, and smoothing the data.

[0041] The non-tunneling state data to be removed includes data on the TBM's stop and step-change states during operation, as the TBM has no interaction with the surrounding rock at these times and needs to be removed; data where the cutterhead torque T is 0, as the cutterhead speed, feed rate, and total feed force are all zero at this time, and this data is removed accordingly; and abnormal state data to be removed includes data where the TBM's feed rate v is greater than the abnormal threshold th. ab The data is due to sensor anomalies. Specifically, the process of dividing the tunneling segment and removing shorter segments involves taking the moment when the TBM's propulsion speed v starts increasing from 0 as the starting point of the segment and the moment when v decreases to 0 as the ending point. Then, segments with durations shorter than the shortest time threshold t are removed. d For tunneling sections with a travel time of less than 1 second; separating the stable tunneling phase data specifically involves, for each divided tunneling section, the first t of the tunneling section... s Data removal in seconds, starting from the t-th second. s Starting from a second, data for the stable tunneling stage is separated. Finally, the data for each stable tunneling stage is smoothed to obtain its respective historical stable segment data. This historical stable segment data and its corresponding surrounding rock type are then used to construct a training set. In practice, the data is divided into a training set and a test set according to a preset ratio.

[0042] 2) Establish a TBM surrounding rock type identification model. By inputting the training set into the TBM surrounding rock type identification model for training, a complete TBM surrounding rock type identification model can be obtained.

[0043] In step 2), the TBM surrounding rock type identification model is specifically a machine learning model, selected from the random forest algorithm.

[0044] During training, the TBM surrounding rock type identification model uses the cutterhead rotation speed n, feed speed v, cutterhead torque T, and total propulsion force F from the training set as inputs and the surrounding rock type R as output. The number of decision trees for the TBM surrounding rock type identification model is preset, and the model is trained to obtain a complete TBM surrounding rock type identification model.

[0045] The accuracy A of the complete TBM surrounding rock type identification model is evaluated using a test set. Accuracy A is equal to the total number of correctly classified samples in the test set divided by the total number of samples in the test set, as follows:

[0046]

[0047] Where A is the recognition accuracy of the test set, and R is... t R represents a single sample in the test set that is correctly classified.

[0048] When the accuracy A exceeds the preset accuracy threshold, the current TBM surrounding rock type identification model is taken as the final complete TBM surrounding rock type identification model.

[0049] 3) Based on the surrounding rock type and the tunneling load data of the tunnel boring machine (TBM), construct a speed operation table based on the relationship between surrounding rock type, cutterhead torque, total thrust, and cutterhead speed; and construct a thrust speed operation table based on the relationship between surrounding rock type, cutterhead torque, total thrust, and thrust speed.

[0050] In step 3), based on the surrounding rock type and the tunneling load data of the tunnel boring machine (TBM), the cutterhead torque T under each surrounding rock category is set according to the preset torque value T. stage Increment, the total thrust F is increased according to the preset thrust value F. stage The cutterhead speed is extracted incrementally within each torque range and each thrust range, and then averaged to obtain the average cutterhead speed within each torque range and each thrust range. Using the range of cutterhead torque T as the x-axis and the average cutterhead speed within the corresponding torque range as the y-axis, the cutterhead speed under different surrounding rock conditions as the cutterhead torque T increases is obtained. Similarly, using the total thrust F as the x-axis and the average cutterhead speed within the corresponding thrust range as the y-axis, the cutterhead speed under different surrounding rock conditions as the total thrust F increases is obtained. Finally, a speed operation table is constructed based on the relationship between surrounding rock type, TBM cutterhead torque, total thrust, and cutterhead speed.

[0051] The cutterhead torque T under the surrounding rock category is set according to the preset torque value T. stage Increment, the total thrust F is increased according to the preset thrust value F. stage The propulsion speed is extracted incrementally within each torque range and each thrust range, and then averaged to obtain the average propulsion speed within each torque range and each thrust range. Using the range of cutterhead torque T as the x-axis and the average propulsion speed within the corresponding torque range as the y-axis, the propulsion speed under different surrounding rock conditions as the cutterhead torque T increases is obtained. Simultaneously, using the total thrust F as the x-axis and the average propulsion speed within the corresponding thrust range as the y-axis, the propulsion speed under different surrounding rock conditions as the total thrust F increases is obtained. Finally, a propulsion speed operation table based on the relationship between surrounding rock type, TBM cutterhead torque, total thrust, and propulsion speed is constructed.

[0052] 4) Establish a combined speed model and a combined thrust model.

[0053] In step 4), the combined rotational speed model is as follows:

[0054] n r =an1+bn2

[0055] Where, n r denoted as the combined cutterhead speed of the tunnel boring machine (TBM); a and b are the first and second speed coefficients, respectively; n1 and n2 are the first and second cutterhead speeds of the TBM, respectively.

[0056] In step 4), the thrust-speed joint model is as follows:

[0057] v r =cv1+dv2

[0058] Among them, v r denoted as the combined propulsion speed of the tunnel boring machine (TBM); c and d are the first and second propulsion speed coefficients, respectively; v1 and v2 are the first and second propulsion speeds, respectively.

[0059] 5) Use sensors to collect feature data of the tunnel boring machine (TBM) in real time, and perform the same data cleaning process as in step 1) to obtain several real-time stable segment data. Input the real-time stable segment data into the complete TBM surrounding rock type identification model. After processing by the complete TBM surrounding rock type identification model, output the surrounding rock type currently being excavated by the TBM, thereby realizing the identification of the surrounding rock type of the TBM.

[0060] 6) Based on the real-time cutterhead torque and real-time total thrust of the tunnel boring machine (TBM) under the current surrounding rock type, extract the first and second cutterhead speeds from the speed operation table, and extract the first and second propulsion speeds from the propulsion speed operation table. Input the first and second cutterhead speeds into the speed joint model, which outputs the joint cutterhead speed of the TBM. Input the first and second propulsion speeds into the propulsion speed joint model, which outputs the joint propulsion speed of the TBM. Control the cutterhead speed and propulsion speed of the TBM at the next moment based on the joint cutterhead speed and joint propulsion speed to achieve the operation control of the TBM.

[0061] The following detailed description of the specific usage process of the method of the present invention, with reference to specific embodiments, demonstrates the practicality and accuracy of the invention. Data from a tunneling section of a Jilin Yinsong Water Supply Project is used as an example.

[0062] The threshold for abnormal progress speed during the data cleaning process is set to th. ab =150mm / min, removal duration in t d For tunneling sections with a duration of less than 200 seconds, the t-th segment of each tunneling section... s=200 seconds as the starting point, separate the steady phase, and finally obtain the sample dataset for model training.

[0063] Using the cutterhead rotation speed n, feed speed v, cutterhead torque T, and total propulsion force F as inputs, and the surrounding rock type R as output, the number of decision trees was set to 30. 80% of the sample set was randomly selected as the training set and 20% as the test set to train the random forest algorithm and establish a surrounding rock type identification model. After testing, the accuracy of the model on the test set was 98.8%.

[0064] For each subdataset corresponding to each surrounding rock category, torque is calculated according to T. stage =Incrementing by 200kNm, thrust according to F stage =Increment by 1000kN, extract the cutter head rotation speed and feed speed in each range, and perform average processing.

[0065] A histogram was plotted with the range of torque on the x-axis and the average cutterhead rotational speed and average feed rate within the corresponding range on the y-axis. The rotational speed and feed rate were summarized as the cutterhead torque increased under different surrounding rock conditions. Similarly, a histogram was plotted with the range of thrust on the x-axis and the average cutterhead rotational speed and average feed rate within the corresponding range on the y-axis. The rotational speeds n1 and n2, and the feed rates v1 and v2 were then derived.

[0066] The speed decision coefficient is calculated by finding speeds n1 and n2 based on the surrounding rock type, torque, and thrust range. Coefficients a and b are set from 0 to 1 in increments of 0.01, with their sum equal to 1. For each set of coefficient values, the decision speed n is calculated. r The average relative error MRE1 between the actual rotational speed n and the actual rotational speed n is as follows:

[0067]

[0068] Where N is the number of iterations, n i Let be the actual rotational speed in the i-th iteration.

[0069] The set of coefficients with the smallest mean relative error (MRE) is selected as the rotational speed decision coefficients for this type of surrounding rock.

[0070] The thrust decision coefficient is calculated by finding the rotational speeds v1 and v2 based on the surrounding rock type, torque, and thrust range. Coefficients c and d are set from 0 to 1 in increments of 0.01, with their sum equal to 1. For each set of coefficient values, the decision thrust velocity v is calculated. r The MRE2 compared to the actual thrust rate v is as follows:

[0071]

[0072] Among them, v i This represents the actual push speed in the i-th iteration.

[0073] The set of coefficients with the smallest mean relative error (MRE) is selected as the thrust rate decision coefficients for this type of surrounding rock.

[0074] Operational control is achieved based on the identified surrounding rock type and decision coefficients. The final operational parameters for the Jilin Yinsong Water Supply Project are as follows: Figure 2 As shown, the current surrounding rock type is Class IV, the rotation speed is 5.2 r / min, and the pushing speed is 57 mm / min. The operating parameters for other tunneling sections in the project can be compared with those in this embodiment.

[0075] This invention fully utilizes data from real engineering sites, overcoming the limitation of limited sample sizes in current rock formation identification research and improving the reliability of rock type identification. Simultaneously, by combining actual data with advanced identification models, it provides a reliable reference for the operational status of TBMs.

[0076] The surrounding rock type identification model of this invention uses on-site TBM tunneling parameters as input and is constructed based on the random forest algorithm, which improves the accuracy of surrounding rock identification. Operation control makes joint decisions based on tunneling load torque and thrust, providing a reliable reference for TBM operators. This invention fully utilizes data from real engineering sites, has high identification accuracy, and, combined with human experience, can more flexibly adapt to different engineering scenarios.

[0077] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A method for identifying and controlling the surrounding rock type of a TBM, characterized in that, include: Step 1) Obtain historical feature data with high correlation between tunnel boring machine (TBM) and different surrounding rock types. After cleaning, obtain a series of historical stable section data. Construct a training set by combining the series of historical stable section data and their corresponding surrounding rock types. Step 2) Establish a TBM surrounding rock type identification model. Train the TBM surrounding rock type identification model by inputting the training set into the model to obtain a complete TBM surrounding rock type identification model. Step 3) Based on the surrounding rock type and the tunneling load data of the tunnel boring machine (TBM), construct a speed operation table based on the relationship between the surrounding rock type, cutterhead torque T, total thrust F and cutterhead speed n, and construct a push speed operation table based on the relationship between the surrounding rock type, cutterhead torque T, total thrust F and push speed v. Step 4) Establish the combined speed model and the combined thrust model; Step 5) Use sensors to collect feature data of the tunnel boring machine (TBM) in real time, and perform the same data cleaning process as in Step 1) to obtain several real-time stable segment data. Input the real-time stable segment data into a complete TBM surrounding rock type identification model. After processing by the complete TBM surrounding rock type identification model, output the surrounding rock type currently being excavated by the TBM, thereby realizing the identification of the surrounding rock type of the TBM. Step 6) Based on the real-time cutterhead torque T and real-time total thrust F of the tunnel boring machine (TBM) under the current surrounding rock type, extract the first and second cutterhead speeds from the speed operation table, and extract the first and second thrust speeds from the thrust speed operation table. Input the first and second cutterhead speeds into the speed joint model, and the speed joint model outputs the joint cutterhead speed of the TBM. Input the first and second thrust speeds into the thrust speed joint model, and the thrust speed joint model outputs the joint thrust speed of the TBM. Control the cutterhead speed n and thrust speed v of the TBM at the next moment based on the joint cutterhead speed and the joint thrust speed to realize the operation control of the TBM. In step 1), the historical characteristic data of the tunnel boring machine (TBM) with high correlation with different surrounding rock types include the cutterhead rotation speed n, propulsion speed v, cutterhead torque T, and total propulsion force F of the TBM. Each subset of data consisting of cutterhead rotation speed n, propulsion speed v, cutterhead torque T, and total propulsion force F corresponds to a type of surrounding rock R that the TBM is tunneling in. In step 3), based on the surrounding rock type and the tunneling load data of the tunnel boring machine (TBM), the cutterhead torque T under each surrounding rock category is set according to a preset torque value T. stage Increment, the total thrust F is increased according to the preset thrust value F. stage The cutterhead rotation speed n is extracted for each torque range and each thrust range, and then averaged to obtain the average cutterhead rotation speed for each torque range and each thrust range. Using the range of cutterhead torque T as the x-axis and the average cutterhead rotation speed for the corresponding torque range as the y-axis, the cutterhead rotation speed n under different surrounding rock conditions is obtained as the cutterhead torque T increases. Similarly, using the total thrust F as the x-axis and the average cutterhead rotation speed for the corresponding thrust range as the y-axis, the cutterhead rotation speed n under different surrounding rock conditions as the total thrust F increases is obtained. Finally, a rotation speed operation table is constructed based on the relationship between surrounding rock type, TBM cutterhead torque T, total thrust F, and cutterhead rotation speed n. The cutterhead torque T under the surrounding rock category is set according to the preset torque value T. stage Increment, the total thrust F is increased according to the preset thrust value F. stage The propulsion speed v is extracted for each torque range and each thrust range, and then averaged to obtain the average propulsion speed for each torque range and each thrust range. The range of cutterhead torque T is used as the abscissa and the average propulsion speed within the corresponding torque range is used as the ordinate to obtain the propulsion speed v as the cutterhead torque T increases under different surrounding rock conditions. At the same time, the total thrust F is used as the abscissa and the average propulsion speed within the corresponding thrust range is used as the ordinate to obtain the propulsion speed v as the total thrust F increases under different surrounding rock conditions. Finally, a propulsion speed operation table based on the relationship between the cutterhead torque T, total thrust F, and propulsion speed v of the tunnel boring machine (TBM) is constructed. In step 4), the combined rotational speed model is as follows: Where, n r denoted as the combined cutterhead speed of the tunnel boring machine (TBM); a and b are the first and second speed coefficients, respectively; n1 and n2 are the first and second cutterhead speeds of the TBM, respectively. In step 4), the thrust-speed joint model is as follows: Among them, v r denoted as , where is the combined propulsion speed of the tunnel boring machine (TBM); c and d are the first and second propulsion speed coefficients, respectively; v1 and v2 are the first and second propulsion speeds, respectively.

2. The TBM surrounding rock type identification and operation control method according to claim 1, characterized in that: In step 1), data cleaning specifically involves sequentially removing non-tunneling state data, removing abnormal state data, dividing the tunneling segment and removing shorter tunneling segment data, separating stable tunneling stage data, and smoothing the data. The removed non-tunneling state data includes TBM stoppage and step change status data during operation, and data where the cutterhead torque T is 0; the removed abnormal state data includes TBM advance speed v greater than the abnormal threshold th. ab The data is divided into tunneling segments, and shorter segments are removed. Specifically, the moment when the tunnel boring machine (TBM) advance speed v starts to increase from 0 is taken as the starting point of the tunneling segment, and the moment when the advance speed v decreases to 0 is taken as the ending point of the tunneling segment. Then, the segments with a duration less than the shortest time threshold t are removed. d For tunneling sections with a travel time of less than 1 second; separating the stable tunneling phase data specifically involves, for each divided tunneling section, the first t of the tunneling section... s Data removal in seconds, starting from the t-th second. s Starting from a second, the data of the stable tunneling stage is separated. Finally, the data of each stable tunneling stage is smoothed to obtain the historical stable segment data. The historical stable segment data and their corresponding surrounding rock types are used to construct a training set.

3. The TBM surrounding rock type identification and operation control method according to claim 1, characterized in that: In step 2), the established TBM surrounding rock type identification model is specifically a machine learning model, selected from the random forest algorithm; During training, the TBM surrounding rock type identification model uses the cutterhead rotation speed n, feed speed v, cutterhead torque T, and total propulsion force F from the training set as inputs and the surrounding rock type R as output. The number of decision trees for the TBM surrounding rock type identification model is preset, and the model is trained to obtain a complete TBM surrounding rock type identification model.

Images