Rock burst case data-driven tunnel rock burst dynamic susceptibility evaluation method

Abstract

The invention discloses a tunnel rockburst dynamic susceptibility evaluation method driven by rockburst case data. During rockburst susceptibility evaluation, an incomplete or inaccurate database can reduce the accuracy of a susceptibility evaluation result. The method comprises the following steps: 1, collecting rockburst case data; 2, calculating the similarity of numerical factor indexes; 3, correcting the similarity of the category-type indexes; 4, evaluating the rockburst susceptibility; and 5, performing dynamic updating on the rockburst susceptibility based on data driving. According to the method, on-site rockburst susceptibility evaluation is more and more accurate through continuous increase of rockburst cases, and disaster-causing information factor characteristics can be extracted; meanwhile, the problem that the factor weight is manually given by a traditional statistical model is avoided, and the weight is prevented from being subject to subjective factors and sample data images.

Description

Technical Field

[0001] The present invention belongs to the technical field of rockburst susceptibility evaluation, and in particular relates to a tunnel rockburst dynamic susceptibility evaluation method driven by rockburst case data. Background Art

[0002] With the continuous development of infrastructure projects, tunnels are buried deeper and deeper, and underground rock projects are advancing to 2000m. Under the influence of deep stress, rock bursts are very likely to occur during the excavation of hard rock sections, causing casualties and property losses. At present, rock bursts, as one of the world's problems, have not been effectively and thoroughly solved. A large number of scholars have conducted in-depth research on rock burst warning. At present, there are two types of rock burst prediction methods. The first type is artificial intelligence rock burst prediction methods based on machine learning with a large amount of rock burst sample data training, such as neural network method, support vector machine method, etc. This type of method has high prediction accuracy when the data is sufficient, but its generalization ability and interpretability are poor, and the data volume of the training set is high; the second type is hierarchical analysis method, gray theory method, etc. Among them, the second type of method does not require a large amount of sample data training and is convenient and fast to calculate. Compared with the first type of artificial intelligence rock burst prediction method, it has certain advantages, but some key steps rely on expert knowledge bases and need to be determined manually. For example, the determination of factor weights in the hierarchical analysis method needs to rely on expert experience and lacks interpretability. In addition, these methods are very dependent on the quality of rock burst data. In the evaluation of rockburst susceptibility, an incomplete or inaccurate database will weaken the statistical laws or learning characteristics of the data-driven model and reduce the accuracy of the susceptibility evaluation results. During the tunneling process, as the excavation continues, the improvement of rockburst case data can more fully and accurately characterize the rockburst characteristic information, and the judgment of susceptibility is closer to reality. Therefore, it is urgent to carry out rockburst susceptibility evaluation driven by rockburst case data. Summary of the invention

[0003] In order to make up for the shortcomings of the prior art, the present invention provides a tunnel rockburst dynamic susceptibility evaluation method driven by rockburst case data to solve the defects of traditional machine early warning models, such as high data volume requirements and statistical models being greatly influenced by human subjectivity.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is:

[0005] A method for evaluating dynamic rockburst susceptibility of a tunnel driven by rockburst case data, characterized in that it comprises the following steps:

[0006] Step 1: rockburst case data collection;

[0007] Step 2: Calculation of similarity of numerical factor indicators;

[0008] Step 3: Similarity correction of category indicators;

[0009] Step 4: rockburst susceptibility assessment;

[0010] Step 5: Dynamic update of rockburst susceptibility based on data-driven.

[0011] Furthermore, in step one, data collection includes determining statistical granularity, collecting numerical indicator factors, collecting category indicator factors, and labeling targeted results.

[0012] In step 1, the statistical granularity is determined as follows:

[0013] In order to ensure the accuracy and comparability of the data, the statistical granularity is set to each excavation cycle.

[0014] In step 1, the numerical index factors collected include burial depth, ground stress, rock strength, ground stress direction, tunnel axis excavation direction, joint dip angle, and TSP longitudinal wave velocity.

[0015] In step one, category indicator factors collected include lithology, excavation method, hole size, and excavation footage.

[0016] Furthermore, in the targeted result annotation, rockburst events are divided into four levels: no rockburst, slight rockburst, moderate rockburst and severe rockburst.

[0017] Further, step 2 is specifically as follows:

[0018] By calculating the similarity of numerical factor indicators, different excavation cycles are ranked by similarity, including determining the semantic distance calculation method, numerical factor selection, and similar sequence generation.

[0019] Furthermore, in step 2, similar sequences are generated as follows:

[0020] The fuzzy similarity priority ratio method is used to calculate the similarity sequence between the excavation cycle to be analyzed and the excavated rockburst case database. Assume that the database consists of a set A = {H1, H2, …, Hn} consisting of n excavated units;

[0021] make Compared with the unit to be excavated Ho, any excavated unit Hi, Hj is called the excavated unit for comparison, then the fuzzy similarity priority ratio matrix R is as shown in formula (1):

[0022] R=(r ij ) n×n , r ij ∈[0,1](i,j=1,2,…,n)

[0023] Where: R is the fuzzy similarity priority ratio matrix; rij is the fuzzy similarity priority ratio between the excavated comparison unit Hi and the unit to be excavated Ho; n is the total number of excavated units in the database;

[0024] The Hamming distance is used to characterize the similarity of excavation cycles. The calculation method of the Hamming distance under the K factor is as follows:

[0025]

[0026] Where: is the target slope unit H o Compared with the slope unit H i Semantic distance under k factor; is the target slope unit H o The state value under the k factor; For comparison, the slope unit H i The state value under the k factor; if the k factor is the TSP longitudinal wave velocity, then D is the velocity difference between the two excavation units;

[0027] According to formula (2), the fuzzy similarity priority ratio r of Hi and Hj to the target slope unit Ho under the k factor is calculated. ij k :

[0028]

[0029] It can be seen from formula (2) that The larger the value, the smaller the excavated unit H is under factor k. i With the slope unit to be excavated H o The similarity is higher than that of H j With H o The larger the value, the smaller the excavation unit fuzzy similarity priority ratio matrix under the k factor is constructed. As shown in formula (3):

[0030]

[0031] The similarity priority ratio matrix is ​​continuously entered into the horizontal cut set to determine the similarity degree, and finally a similar finite sequence for the unit to be excavated is obtained:

[0032] S o ={S o1 , S o2 , S o3 , S o4 , S o5}.

[0033] Further, step three is specifically as follows:

[0034] Assume that the similarity priority sequence of the cyclic unit Ho to be excavated calculated by the numerical factor is:

[0035] S o ={S o1 , S o2 , So3 , S o4 , S o5}

[0036] In the unit to be excavated H o In the similar priority sequence of o1 Represents the slope unit H to be excavated o The most similar excavated unit, S o5 Indicates that H o The most dissimilar excavated unit. The method of using the category factor to modify the similarity priority sequence of the excavated unit Ho is as follows: when the properties of the slope unit to be compared are different from those of Ho under a certain category factor, the slope unit is moved to the end of the similarity sequence So, and if they are the same, the original position remains unchanged; for example, S o2 and S o4 Different from the properties of Ho under a certain categorical factor, other slope units are the same, then the modified sequence of So is:

[0037] S o ={S o1 , S o3 , S o5 , S o2 , S o4}

[0038] After the above corrections, an absolutely similar sequence of units to be excavated is finally obtained.

[0039] Further, step four is specifically as follows:

[0040] By setting the similarity threshold, the absolutely similar sequence is divided into similar units and non-similar units. The similarity threshold I can be expressed by the ratio of rockburst units to the total number of excavation units in the database: I = [n / N]

[0041] Where n is the total number of excavation units in the database; N is the number of rockburst units currently collected in the database; and by counting the number of excavation units with rockburst as the target indicator in the prone area, the rockburst prone area is subdivided into high, medium and low prone areas.

[0042] Beneficial effects of the present invention:

[0043] 1) The reliance of traditional machine learning algorithms on sample data often becomes a bottleneck in their application. In many cases, obtaining sufficient high-quality sample data is time-consuming and expensive. The newly developed prediction model significantly reduces the reliance on sample data by introducing more flexible and intelligent data processing technology. This not only improves the applicability of the model in data-scarce situations, but also makes the updating and adjustment of the model more efficient in a rapidly changing environment;

[0044] 2) The problem of assigning subjective factors in conventional statistical methods easily leads to deviation and uncertainty in the results. By introducing a data-driven approach, the model can automatically identify the similarities of various indicator factors, thereby reducing human interference when judging the susceptibility of rockbursts. This objective evaluation method not only improves the accuracy of the prediction, but also enhances the reliability of the model, allowing engineers to conduct rockburst risk assessment more scientifically.

[0045] 3) The model's clarity and interpretability in the physical sense are also a major advantage. Through scientific indicator selection and factor analysis, the model can clearly show the impact of various factors on rockburst susceptibility. This interpretability not only helps engineers understand the prediction results, but also provides strong support for subsequent decision-making, allowing engineering managers to more confidently formulate corresponding prevention and treatment measures.

[0046] 4) It has the ability to classify according to typical rockburst cases, so that different rockburst types can be counted and analyzed. This classification capability not only helps to identify potential rockburst risks, but also provides direction for future research and promotes in-depth understanding and research of rockburst phenomena.

[0047] 5) With the continuous improvement of the rockburst database, the early warning capability of the model is also continuously enhanced. By continuously accumulating and updating rockburst case data, the model can learn more rockburst performance characteristics and laws, so that when faced with new cases, it can conduct risk assessment more quickly and accurately. This dynamic learning feature enables the prediction model to always maintain a high level of accuracy, providing a stronger guarantee for engineering safety.

[0048] 6) The model is sensitive to new types of rockbursts and can quickly identify the differences between new rockburst cases and existing cases in the database. This feature means that engineers can take preventive measures in advance to reduce potential risks before new types of rockbursts are widely known.

[0049] 7) The generalization ability of the model also provides broad possibilities for its application prospects. Regardless of rockburst tunnels with different lithology and different excavation methods, the model can be used for unified analysis. By building the model once, it can operate effectively under various conditions, greatly reducing duplication of work and time costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is an operation flow chart of the present invention. DETAILED DESCRIPTION

[0051] The present invention is described in detail below in conjunction with specific implementation modes.

[0052] The present invention can determine potential rock burst high-prone areas through rock burst cases, and establish a dynamic rock burst susceptibility model driven by rock burst case data. The existing rock burst warning model based on machine learning has poor generalization ability. The model established in one tunnel is not applicable to another tunnel. When there are fewer rock burst cases in tunnels and the data is incomplete, it is even more difficult to establish a warning model through machine learning. With the increasing number of rock burst cases, the assessment of rock burst susceptibility on site will become more and more accurate, and the characteristics of disaster-causing information factors can be extracted. At the same time, it avoids the problem of artificially given factor weights in traditional statistical models, and avoids the weights being affected by subjective factors and sample data images.

[0053] like Figure 1 As shown, the present invention comprises the following steps:

[0054] Step 1: Collection of rockburst case data; The collection of rockburst case data not only lays the foundation for the training and verification of subsequent models, but also provides the necessary empirical support for understanding the occurrence mechanism of rockburst. Through the systematic collection of multiple numerical indicator factors and category indicator factors, a comprehensive and accurate database is constructed to improve the effectiveness of the prediction model. Data collection includes statistical granularity determination, numerical indicator factor collection, category indicator factor collection, and targeted result annotation, specifically:

[0055] (I) The statistical granularity is determined as follows:

[0056] In order to ensure the accuracy and comparability of the data, the statistical granularity is set to each excavation cycle (each blasting footage for blasting method and every 10m footage for TBM).

[0057] (II) Collection of numerical indicator factors includes:

[0058] Numerical index factors are quantitative descriptions of rockburst events, usually including the following key parameters:

[0059] 1) Buried depth: Buried depth refers to the vertical distance between the rock surface and the excavation surface. This factor directly affects the stress state of the rock mass and its compressive capacity. The greater the buried depth, the greater the ground stress the rock mass is subjected to, which may increase the risk of rock burst.

[0060] 2) Geostress: Geostress is the natural stress generated inside the rock mass, usually caused by the weight of the stratum and the surrounding environmental conditions. Understanding the state and changes of geostress can help us better assess the potential risk of rock burst.

[0061] 3) Rock strength: The compressive, tensile and shear strength of rock are important indicators for determining rock mass stability. These strength parameters affect the ability of rock mass to withstand external pressure during excavation. The lower the rock strength, the higher the possibility of rock burst.

[0062] 4) Direction of geostress: The direction of geostress plays an important role in the occurrence of rock bursts. Different directions of geostress will cause different degrees of deformation and damage to the rock mass. Therefore, accurate determination of the direction of geostress is crucial.

[0063] 5) Tunnel axis excavation direction: The relationship between the tunnel axis excavation direction and the ground stress direction will also affect the stability of the rock mass. Reasonable selection of excavation direction can effectively reduce the risk of rock burst.

[0064] 6) Joint inclination and dip: Joints are cracks or weak planes in the rock mass, and their inclination and dip will directly affect the mechanical properties of the rock mass. The existence of joints is often one of the key factors that cause rock bursts, so it is very important to record their information in detail.

[0065] 7) TSP longitudinal wave velocity: The longitudinal wave velocity of TSP (seismic wave) is an important indicator for evaluating the properties of rock materials, and can reflect the elastic characteristics and crack conditions of the rock mass. The change in wave velocity can be used to judge the integrity and mechanical properties of the rock mass, and then infer the possibility of rock burst.

[0066] (III) Category indicator factor collection includes:

[0067] Category indicator factors are used to classify and characterize rockburst events. These factors include:

[0068] 1) Lithology: Lithology describes the mineral composition and physical properties of the rock mass. Different lithologies have different sensitivities to rockbursts, so detailed records of lithology information are the basis for understanding rockburst phenomena.

[0069] 2) Excavation method: Different excavation methods (such as blasting, mechanical excavation, etc.) have different effects on the rock mass. Reasonable selection of excavation method can effectively reduce the risk of rock burst.

[0070] 3) Hole diameter size: The size of the hole diameter directly affects the stress distribution of the rock mass. The larger the hole diameter, the higher the risk of rock mass damage. Therefore, the collection of hole diameter information is very important for rock burst analysis.

[0071] 4) Excavation footage: Excavation footage refers to the length of excavation completed within a specific period of time. The speed of the footage will affect the stress state of the rock mass and the occurrence of rock bursts, so the footage data of each excavation should be recorded in detail.

[0072] (IV) Targeted Results

[0073] In the targeted result annotation, the definition of targeted results is crucial in the final stage of data collection. Rockburst events are classified into four levels: no rockburst, slight rockburst, moderate rockburst, and severe rockburst.

[0074] Step 2: Calculation of similarity of numerical factor indicators;

[0075] By calculating the similarity of numerical factor indicators, different excavation cycles are ranked by similarity, including determining the semantic distance calculation method, numerical factor selection, and similar sequence generation.

[0076] 1) The concept of semantic distance

[0077] Semantic distance refers to the similarity measure of two samples (in this case, excavation cycles) in a multidimensional feature space. The core idea is to analyze the features of each excavation cycle, calculate their differences in feature dimensions, and then judge their similarity. In rockburst prediction, the smaller the semantic distance, the more similar the conditions of the two excavation cycles are, and the more comparable the risk of rockburst occurrence is.

[0078] 2) Selection of numerical factors

[0079] When calculating semantic distance, we first need to select appropriate numerical factors. These factors include burial depth, ground stress, rock strength, ground stress direction, tunnel axis excavation direction, joint dip angle, and TSP longitudinal wave velocity. These factors can fully reflect the characteristics of the excavation cycle and provide a reliable basis for calculating semantic distance.

[0080] 3) The specific steps of similar sequence generation are as follows:

[0081] The fuzzy similarity priority ratio method is used to calculate the similarity sequence between the excavation cycle to be analyzed and the excavated rockburst case database. Assume that the database consists of a set A = {H1, H2, …, Hn} consisting of n excavated units;

[0082] make Compared with the unit to be excavated Ho, any excavated unit Hi, Hj is called the excavated unit for comparison, then the fuzzy similarity priority ratio matrix R is as shown in formula (1):

[0083] R=(r ij ) n×n , r ij ∈[0,1](i,j=1,2,…,n)

[0084] Where: R is the fuzzy similarity priority ratio matrix; rij is the fuzzy similarity priority ratio between the excavated comparison unit Hi and the unit to be excavated Ho; n is the total number of excavated units in the database;

[0085] The Hamming distance is used to characterize the similarity of excavation cycles. The calculation method of the Hamming distance under the K factor is as follows:

[0086]

[0087] Where: is the target slope unit ho Compared with the slope unit H i Semantic distance under k factor; is the target slope unit H o The state value under the k factor; For comparison, the slope unit H i The state value under the k factor; if the k factor is the TSP longitudinal wave velocity, then D is the velocity difference between the two excavation units;

[0088] According to formula (2), the fuzzy similarity priority ratio r of Hi and Hj to the target slope unit Ho under the k factor is calculated. ij k :

[0089]

[0090] It can be seen from formula (2) that The larger the value, the smaller the excavated unit H is. i With the slope unit to be excavated H o The similarity is higher than that of H j With H o The larger the value, the smaller the excavation unit fuzzy similarity priority ratio matrix under the k factor is constructed. As shown in formula (3):

[0091]

[0092] The similarity priority ratio matrix is ​​continuously entered into the horizontal cut set to determine the similarity degree, and finally a similar finite sequence for the unit to be excavated is obtained:

[0093] S o ={S o1 , S o2 , S o3 , S o4 , S o5}.

[0094] Step 3: Similarity correction of category indicators;

[0095] Step three is as follows:

[0096] The category index plays a decisive role in controlling the similarity. When the similarity of the numerical index is very high, but the category index is different, it is impossible to consider that the two excavation cycles are similar. This is mainly because the category index determines the difference in type more. For example, if the lithology and excavation method are different, the rockburst evaluation mechanism is completely different. Therefore, it is necessary to carry out the similarity correction of the category index based on the calculation results of the similarity results of the numerical factor index, so as to determine the final similarity sequence. It is assumed that the similarity priority sequence of the cycle unit Ho to be excavated calculated by the numerical factor is:

[0097] S o ={S o1 , S o2 , S o3 , S o4 , S o5}

[0098] In the unit to be excavated H o In the similar priority sequence of o1 Represents the slope unit H to be excavated o The most similar excavated unit, S o5 Indicates that H o The most dissimilar excavated unit. The method of using the category factor to modify the similarity priority sequence of the excavated unit Ho is as follows: when the properties of the slope unit to be compared are different from those of Ho under a certain category factor, the slope unit is moved to the end of the similarity sequence So, and if they are the same, the original position remains unchanged; for example, S o2 and S o4 Different from the properties of Ho under a certain categorical factor, other slope units are the same, then the modified sequence of So is:

[0099] S o ={S o1 , S o3 , S o5 , S o2 , S o4}

[0100] After the above corrections, an absolutely similar sequence of units to be excavated is finally obtained.

[0101] Step 4: rockburst susceptibility assessment (how to divide high, medium and low susceptibility areas);

[0102] Step 4 is as follows:

[0103] By setting the similarity threshold, the absolutely similar sequence is divided into similar units and non-similar units. The similarity threshold I can be expressed by the ratio of rockburst units to the total number of excavation units in the database: I = [n / N]

[0104] In the formula: n is the total number of excavation units in the database; N is the number of rockburst units currently collected in the database; and by counting the number of excavation units with the target index of rockburst in the prone area (rockburst similarity frequency), the rockburst prone area is subdivided into high, medium and low prone areas. The larger the rockburst similarity frequency, the more similar the unit to be excavated is to the rockburst units, that is, the higher the rockburst susceptibility of the unit to be excavated; conversely, the lower the rockburst susceptibility.

[0105] Step 5: Dynamic update of rockburst susceptibility based on data-driven.

[0106] The data-driven model is based on a large amount of historical data, using machine learning and statistical analysis methods to identify and capture potential patterns in the data. These patterns can reflect the occurrence mechanism of rockbursts, thus providing a scientific basis for engineering practice. As the database expands, the model can access richer case data, making the model training process more comprehensive and the accuracy of predictions improved. With the real-time collection and storage of new excavation cycle data, the content of the database is constantly enriched. Data obtained from multiple engineering projects and geological environments can not only cover different types of rockburst cases, but also capture a variety of scenarios in which rockbursts occur. This rich data foundation provides more samples for model learning, thereby improving the model's ability to identify statistical laws.

[0107] Dynamic susceptibility assessment refers to the real-time analysis and evaluation of rockburst susceptibility while the rockburst database is constantly updated. Specifically, as new excavation cycle data flows in, the model automatically adjusts its parameters and structure to adapt to the new data characteristics. This dynamic update process enables the model to reflect the current rockburst risk situation in real time.

[0108] The implementation of dynamic susceptibility assessment usually includes the following steps:

[0109] (1) Data collection: After each excavation cycle, relevant numerical and categorical factor data are collected in a timely manner and integrated into the rockburst database.

[0110] (2) Data preprocessing: Clean, filter, and normalize new data to ensure data quality and consistency. The preprocessed data will be used as the basis for model updates.

[0111] (3) Model training and updating: Using the updated database, the data-driven model is retrained. By introducing new excavation cycle data, the model can learn the latest rockburst characteristics and laws, thereby adjusting its prediction ability.

[0112] (4) Susceptibility assessment: Based on the updated model, the current rockburst susceptibility is assessed. By calculating the similarity between the new excavation cycle and other cycles in the database, the model will determine the possibility of rockburst under the current excavation conditions and give corresponding warnings.

[0113] (5) Result analysis and feedback: The results of the rockburst susceptibility assessment are compared with the actual monitoring data to evaluate the prediction accuracy of the model. On this basis, the similarity threshold parameters are further adjusted to continuously optimize the prediction performance.

[0114] The content of the present invention is not limited to the embodiments listed, and any equivalent changes made to the technical solution of the present invention by ordinary technicians in this field after reading the description of the present invention are covered by the claims of the present invention.

Claims

1. A method for evaluating dynamic rockburst susceptibility of tunnels driven by rockburst case data, characterized by: The following steps are involved: Step 1: rockburst case data collection; Step 2: Calculation of similarity of numerical factor indicators; Step 3: Similarity correction of category indicators; Step 4: rockburst susceptibility assessment; Step 5: Dynamic update of rockburst susceptibility based on data-driven.

2. The method for evaluating dynamic rockburst susceptibility of tunnels driven by rockburst case data according to claim 1, characterized in that: In step one, data collection includes determining statistical granularity, collecting numerical indicator factors, collecting category indicator factors, and labeling targeted results.

3. The method for dynamic rockburst susceptibility assessment of tunnels driven by rockburst case data according to claim 2 is characterized by: In step 1, the statistical granularity is determined as follows: In order to ensure the accuracy and comparability of the data, the statistical granularity is set to each excavation cycle.

4. The method for dynamic rockburst susceptibility assessment of tunnels driven by rockburst case data according to claim 3 is characterized by: In step 1, the numerical index factors collected include burial depth, ground stress, rock strength, ground stress direction, tunnel axis excavation direction, joint dip angle, and TSP longitudinal wave velocity.

5. The method for evaluating dynamic rockburst susceptibility of tunnels driven by rockburst case data according to claim 4 is characterized by: In step one, category indicator factors collected include lithology, excavation method, hole size, and excavation footage.

6. The method for dynamic rockburst susceptibility assessment of tunnels driven by rockburst case data according to claim 5, characterized in that: In the targeted result annotation, rockburst events are divided into four levels: no rockburst, slight rockburst, moderate rockburst and severe rockburst.

7. The method for evaluating dynamic rockburst susceptibility of tunnels driven by rockburst case data according to claim 6 is characterized by: Step 2 is as follows: By calculating the similarity of numerical factor indicators, different excavation cycles are ranked by similarity, including determining the semantic distance calculation method, numerical factor selection, and similar sequence generation.

8. The method for evaluating dynamic rockburst susceptibility of tunnels driven by rockburst case data according to claim 7, characterized in that: In step 2, similar sequences are generated as follows: The fuzzy similarity priority ratio method is used to calculate the similarity sequence between the excavation cycle to be analyzed and the excavated rockburst case database. Assume that the database consists of a set A = {H1, H2, …, Hn} consisting of n excavated units; make Compared with the unit to be excavated Ho, any excavated unit Hi, Hj is called the excavated unit for comparison, then the fuzzy similarity priority ratio matrix R is as shown in formula (1): R=(r ij ) n×n ,r ij ∈[0,1](i,j=1,2,…,n) Where: R is the fuzzy similarity priority ratio matrix; rij is the fuzzy similarity priority ratio between the excavated comparison unit Hi and the unit to be excavated Ho; n is the total number of excavated units in the database; The Hamming distance is used to characterize the similarity of excavation cycles. The calculation method of the Hamming distance under the K factor is as follows: Where: is the target slope unit H o Compared with the slope unit H i Semantic distance under k factor; is the target slope unit H o The state value under the k factor; For comparison, the slope unit H i The state value under the k factor; if the k factor is the TSP longitudinal wave velocity, then D is the velocity difference between the two excavation units; According to formula (2), the fuzzy similarity priority ratio r of Hi and Hj to the target slope unit Ho under the k factor is calculated. ij k : It can be seen from formula (2) that The larger the value, the smaller the excavated unit H is. i With the slope unit to be excavated H o The similarity is higher than that of H j With H o The larger the value, the smaller the excavation unit fuzzy similarity priority ratio matrix under the k factor is constructed. As shown in formula (3): The similarity priority ratio matrix is ​​continuously entered into the horizontal cut set to determine the similarity degree, and finally a similar finite sequence for the unit to be excavated is obtained: S o ={S o1 ,S o2 ,S o3 ,S o4 ,S o5 }。 9. The method for evaluating dynamic rockburst susceptibility of tunnels driven by rockburst case data according to claim 8, characterized in that: Step three is as follows: Assume that the similarity priority sequence of the cyclic unit Ho to be excavated calculated by the numerical factor is: S o ={S o1 ,S o2 ,S o3 ,S o4 ,S o5 } In the unit to be excavated H o In the similar priority sequence of o1 Represents the slope unit H to be excavated o The most similar excavated unit, S o5 Indicates that H o The most dissimilar excavated unit. The method of using the category factor to modify the similarity priority sequence of the excavated unit Ho is as follows: when the properties of the slope unit to be compared are different from those of Ho under a certain category factor, the slope unit is moved to the end of the similarity sequence So, and if they are the same, the original position remains unchanged; for example, S o2 and S o4 Different from the properties of Ho under a certain categorical factor, other slope units are the same, then the modified sequence of So is: S o ={S o1 ,S o3 ,S o5 ,S o2 ,S o4 } After the above corrections, an absolutely similar sequence of units to be excavated is finally obtained.

10. The method for dynamic rockburst susceptibility assessment of tunnels driven by rockburst case data according to claim 9, characterized in that: Step 4 is as follows: By setting the similarity threshold, the absolutely similar sequence is divided into similar units and non-similar units. The similarity threshold I can be expressed by the ratio of rockburst units to the total number of excavation units in the database: I = [n / N] Where n is the total number of excavation units in the database; N is the number of rockburst units currently collected in the database; and by counting the number of excavation units with rockburst as the target indicator in the prone area, the rockburst prone area is subdivided into high, medium and low prone areas.

Images