Open-type TBM rock burst jamming risk early warning method and device

Abstract

The invention discloses an open type TBM rock burst jamming risk early warning method. The method comprises the steps that the running state of a TBM is divided into a normal tunneling state, a jamming early warning state and a jamming state; screening out key operation parameters; obtaining a data set; the data set comprises input data and output data; dividing the data set into a training set and a test set; preprocessing the training set to obtain a preprocessed training set; a weight adaptive algorithm is adopted, the preprocessed training set is used for training, and a risk early warning model is obtained; and the risk early warning model is utilized to perform early warning on the open-type TBM rockburst jamming risk. According to the method, a transition state between a normal tunneling state and a jamming state of the TBM is defined as a jamming early warning state, and a jamming risk early warning problem is converted into a jamming early warning state recognition problem; accurate early warning of the open-type TBM rockburst jamming risk is realized by using the running data of the TBM, additional deployment of a monitoring sensor is not needed, the reliability is high, and the method has important academic and engineering values.

Description

Technical Field

[0001] The present invention belongs to the field of machine learning and TBM technology, and specifically relates to an open TBM rock burst and jamming risk warning method and device. Background Art

[0002] Compared with the drilling and blasting method, the hard rock tunnel boring machine (TBM) has the advantages of safety, high efficiency, and environmental protection. It is being widely used in transportation, water conservancy, municipal engineering, and mining tunnels. However, when encountering adverse geology, such as high ground stress, fault fracture zone, and weak surrounding rock, it is easy to cause a jamming accident caused by the TBM cutterhead or shield being stuck. According to statistics, among the many construction accidents of TBM, jamming is the most frequent and causes the greatest economic losses. According to the classification of geological conditions, TBM jams can be divided into four categories: jamming of the shield caused by large deformation of soft rock compression, jamming of the cutterhead / shield caused by fault fracture zone, jamming of the shield caused by the sliding of dangerous rock mass, and jamming of the shield caused by rock burst / fragmentation / plate cracking. Among them, jamming caused by rock burst occurs quickly, is highly concealed, and is very rare. If the jamming accident caused by rock burst can be warned in advance, it can buy time for the on-site construction personnel to take countermeasures, which can greatly alleviate the adverse effects of jamming and even avoid the occurrence of jamming accidents. Therefore, early warning of the risk of TBM rockburst and jamming is of great engineering significance.

[0003] There are two main methods for early warning of TBM jam risk: numerical simulation method and data-driven method. The former uses simulation software to simulate the TBM excavation process, proposes jam criteria and predicts the risk of TBM jam, while the latter uses data-driven algorithms to establish a mapping model between jam influencing factors and jam risk. Numerical simulation methods are difficult to accurately simulate the actual excavation process and have poor real-time performance, while data-driven methods can timely predict TBM jam risks based on field data and are gaining more and more attention. Existing data-driven research rarely focuses on open TBM jams caused by rock bursts, and generally requires additional deployment of special sensors, making it difficult to ensure the reliability of the monitoring system. Jam accidents in open TBMs are relatively rare, and there are fewer corresponding studies, and jams caused by rock bursts are even rarer. The shield of an open TBM can be approximately radially expanded and contracted, which can overcome the risk of shield jams caused by surrounding rock convergence deformation or fragmentation and expansion rupture to a certain extent, but the shield has a certain range of expansion and contraction, and exceeding this range will still cause the shield to be jammed. The process of an open TBM jam caused by rock burst occurs very quickly. It usually takes only one or two days from normal excavation to jam, which places high demands on the anticipation and real-time nature of jam warning.

[0004] Therefore, there is an urgent need for a method for early warning of rock burst jamming for open TBMs that can be used in TBM on-site construction. Summary of the invention

[0005] In view of the many problems existing in the current TBM stuck risk warning methods, the present invention provides an open TBM rock burst stuck risk warning method and device. The method provided in this application can realize TBM stuck risk monitoring and warning using only TBM operation monitoring data, and converts the stuck risk warning problem into the problem of identifying the transition state between normal excavation and stuck state, proposes a category weight adaptive KNN algorithm (CWA-KNN) and establishes an open TBM rock burst stuck risk warning model. The method described in this application has high reliability and accuracy, has important academic and engineering value, and can be accurately applied to TBM operations.

[0006] The technical solutions provided by this application are as follows:

[0007] In a first aspect, the present application provides an open TBM rockburst risk warning method, comprising the following steps:

[0008] The operating status of the TBM is divided into three states: normal excavation state, machine jam warning state and machine jam state;

[0009] Screen out key operating parameters;

[0010] Obtaining a data set; the data set includes input data and output data, wherein the input data is a time series statistical feature of a key operating parameter, and the output data is an operating state of a corresponding TBM;

[0011] Divide the dataset into training and testing sets;

[0012] Preprocessing the training set to obtain a preprocessed training set;

[0013] Adopting weight adaptive algorithm, using pre-processed training set for training, to obtain risk warning model;

[0014] The risk warning model is used to warn of the risk of rockburst and jamming of open TBM.

[0015] In one possible implementation, the starting moment of the jamming machine warning state is within two days before the jamming machine occurs, when the thrust gradually increases to 90% of the rated value and has a trend of continuing to increase, and the torque begins to have a downward trend.

[0016] In one possible implementation, the key operating parameters include thrust, torque, cutter head speed, net excavation speed, 1# top shield cylinder rodless chamber pressure, 2# top shield cylinder rodless chamber pressure, left shield cylinder rodless chamber pressure, left wedge cylinder rodless chamber pressure, right shield cylinder rodless chamber pressure, right wedge cylinder rodless chamber pressure, top shield displacement, left shield displacement and right shield displacement.

[0017] In a possible implementation, the method for obtaining the time series statistical characteristics of the key operating parameters includes the following steps:

[0018] The collected TBM operation data is divided into multiple data segments according to the day and the operation status;

[0019] Filtering data in a tunneling working state from the multiple data segments as a first data set;

[0020] Eliminate the data rows containing missing values ​​in the first data set, use the isolation forest algorithm to detect and delete outliers in the selected data segments, and obtain a second data set;

[0021] Normalizing the second data set to obtain a normalized data set;

[0022] The time series statistical characteristics of key operating parameters are extracted from the normalized data set; the time series statistical characteristics of the key operating parameters include maximum value, minimum value, mean value, standard deviation, root mean square, kurtosis factor and skewness.

[0023] In a possible implementation, the method of preprocessing the training set to obtain the preprocessed training set includes the following steps:

[0024] For the samples of machine warning status and machine status in the training set, a multi-scale segmentation sampling algorithm is used to increase the sample size;

[0025] The normal excavation state samples in the training set are under-sampled using the Tomek Links algorithm.

[0026] In a possible implementation, the method of using a weight adaptive algorithm to train with a preprocessed training set to obtain a risk warning model includes the following steps:

[0027] Initialize the category weights w and the number of nearest neighbors k for the three operating states of TBM, and initialize the hyperparameters of the heuristic algorithm;

[0028] Use the particle swarm optimization algorithm formula to update the particle velocity v w 、v k , update the particle's displacement parameters w and k, and perform the nearest rounding operation on k [k];

[0029] Based on the training set and w, [k], the sum of the weighted distances of samples in each category is calculated. The category with the largest sum of weighted distances is the final category of the target point. The fitness of the model on the test set is verified using the cross-validation algorithm.

[0030] By comparing the fitness values, the optimal sample of the population in each iteration is determined;

[0031] Determine the optimal sample of the population by comparing the fitness values;

[0032] Iterate until the optimal sample of the population converges or reaches the specified iteration number;

[0033] Get the optimal category weight w and the number of nearest neighbors k.

[0034] In a second aspect, the present application provides an open-type TBM rockburst and jamming risk warning device, comprising:

[0035] The state classification module is used to classify the operation state of the TBM into three states: normal excavation state, machine jam warning state and machine jam state;

[0036] A screening module is used to screen out key operating parameters;

[0037] An acquisition module, used to acquire a data set; the data set includes input data and output data, wherein the input data is a time series of key operating parameters, and the output data is an operating state of a corresponding TBM;

[0038] Data partitioning module, used to divide the data set into training set and test set;

[0039] A preprocessing module, used for preprocessing the training set to obtain a preprocessed training set;

[0040] A model acquisition module is used to adopt a weight adaptive algorithm and use a preprocessed training set for training to obtain a risk warning model;

[0041] The risk warning module is used to use the risk warning model to warn of the risk of rock burst and jamming of open TBM.

[0042] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the aforementioned open TBM rockburst and jamming risk warning method when executing the program.

[0043] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the aforementioned open TBM rockburst and sticking risk warning method.

[0044] In a fifth aspect, the present application provides a computer program product, including a computer program, characterized in that when the computer program is executed by a processor, it implements the aforementioned open TBM rockburst and sticking risk warning method.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. The method of the present invention defines the transition state between the normal excavation state and the stuck state as the stuck warning state, creatively converting the stuck warning problem into the stuck warning state identification problem. Compared with the existing method, the method can warn the occurrence of stuck machine more in advance, thus buying time for taking countermeasures on site;

[0047] 2. The method described in the present invention only needs to utilize the parameters monitored by the TBM's own monitoring system (wherein thrust, torque, cutter head speed, net excavation speed are commonly selected parameters for monitoring machine jams, and this application also additionally selects open TBM shield cylinder pressure and shield displacement as input parameters) to achieve early warning of machine jam disasters without the need to additionally deploy some special monitoring sensors. The method has strong practicality, wide applicability, and good reliability and economy.

[0048] 3. The method described in the present invention proposes a category weight adaptive k-nearest neighbor algorithm (CWA-KNN), which optimizes the output category so that the model can focus on the recognition accuracy of the key state of the card machine warning state, which is conducive to improving the recognition accuracy.

[0049] 4. Compared with the numerical simulation method, the method of the present invention has good real-time performance and can accurately predict the machine jam caused by rock burst in real time during TBM operation.

[0050] 5. The method of the present invention can realize early warning of rock burst and jamming risk of open TBM. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 is a flow chart of the method of the present invention;

[0053] Figure 2 The performance of the KNN and CWA-KNN models on the test set at different data processing stages of the embodiment of the present invention;

[0054] Figure 3 A schematic diagram of the structure of an open-type TBM rockburst and jamming risk warning device provided by the present invention;

[0055] Figure 4 This is a schematic structural diagram of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0057] refer to Figure 1 , a risk warning method for rock burst and jamming of an open TBM, comprising the following steps:

[0058] Step 100, dividing the operation state of the TBM into three states: normal excavation state, machine jam warning state and machine jam state.

[0059] In step S100, the changes in operating parameters before and after the open TBM jam (blocking shield) caused by rock burst are analyzed, and the operating state of the TBM is divided into normal excavation state, jam warning state and jam state. Among them, the jam warning state is a transition state between the normal excavation state and the jam state, thereby converting the jam risk warning problem into the jam warning state identification problem.

[0060] Furthermore, in step 1, the starting time of the machine jam warning state (the ending time of the machine jam warning state is the time when the machine jam occurs) is within two days before the machine jam occurs (the machine jam caused by rock burst occurs very quickly, and it usually takes only two or three days from normal excavation to machine jam), the moment when the thrust gradually increases to about 90% of the rated value and has a trend of continuing to increase, and the torque begins to have a downward trend.

[0061] Step 200, screening out key operating parameters.

[0062] In one possible implementation method, 13 key operating parameters are screened out, including: thrust, torque, cutter head speed, net excavation speed, rodless chamber pressure of 1# top shield cylinder, rodless chamber pressure of 2# top shield cylinder, rodless chamber pressure of left shield cylinder, rodless chamber pressure of left wedge cylinder, rodless chamber pressure of right shield cylinder, rodless chamber pressure of right wedge cylinder, top shield displacement, left shield displacement and right shield displacement.

[0063] It should be noted that in step S200, the parameters monitored by the TBM's own monitoring system can be used to realize the machine jam warning. Among the above key operating parameters, thrust, torque, cutter head speed, and net excavation speed are the parameters commonly selected for monitoring the machine jam. This application also selects the open TBM shield cylinder pressure (including 1# top shield cylinder rodless chamber pressure, 2# top shield cylinder rodless chamber pressure, left shield cylinder rodless chamber pressure, left wedge cylinder rodless chamber pressure, right shield cylinder rodless chamber pressure, and right wedge cylinder rodless chamber pressure) and shield displacement (top shield displacement, left shield displacement, and right shield displacement) as key operating parameters. The reason for selecting the open TBM shield cylinder pressure and shield displacement is that through a large amount of data analysis and research, it is found that the shield cylinder pressure fluctuation before the machine jams will be more violent and the shield displacement has a tendency to gradually decrease, and these two parameters have a relatively high correlation with the judgment of the machine jam warning state. Furthermore, the present application can only be used to predict the jamming of the machine through the changes in the above data, without the need to deploy some additional special monitoring sensors, which is more reliable, economical and practical. If some additional special monitoring sensors are deployed, the operation will be extremely difficult, time-consuming and labor-intensive in actual TBM operations.

[0064] Step S300, obtaining a data set; the data set includes input data and output data, wherein the input data is the time series statistical characteristics of key operating parameters, and the output data is the operating status of the corresponding TBM.

[0065] In a possible implementation, in step S300, the method for obtaining the time series statistical characteristics of key operating parameters includes the following steps:

[0066] The collected TBM operation data is divided into multiple data segments according to the day and the operation status;

[0067] Filtering data in a tunneling working state from the multiple data segments as a first data set;

[0068] Eliminate the data rows containing missing values ​​in the first data set, use the isolation forest algorithm to detect and delete outliers in the selected data segments, and obtain a second data set;

[0069] Normalizing the second data set to obtain a normalized data set;

[0070] The time series statistical characteristics of key operating parameters are extracted from the normalized data set; the time series statistical characteristics of the key operating parameters include maximum value, minimum value, mean value, standard deviation, root mean square, kurtosis factor and skewness.

[0071] Furthermore, the method for extracting time series statistical features of key operating parameters from the normalized data set comprises the following steps:

[0072] For a time series data, extract its maximum value, minimum value, mean value and standard deviation; at the same time, calculate the root mean square, kurtosis factor and skewness of the sequence according to the root mean square, kurtosis factor and skewness calculation formulas.

[0073] In a possible implementation, the category labels of the normal excavation state, the jammed machine warning state, and the jammed machine state are 0, 1, and 2, respectively.

[0074] Step S400, dividing the data set into a training set and a test set.

[0075] In a possible implementation, in step S400, the data from the left tunnel is used to construct the original training set, and the data from the right tunnel is used to construct the test set. The original training set is used for subsequent unbalanced data processing and rockburst risk warning model training, and the test set is used to verify the effectiveness of the model.

[0076] Step S500, preprocessing the training set to obtain a preprocessed training set.

[0077] In a possible implementation, in step S500, the method of preprocessing the training set to obtain the preprocessed training set includes the following steps:

[0078] For the samples of machine warning status and machine status in the training set, a multi-scale segmentation sampling algorithm is used to increase the sample size;

[0079] The normal excavation state samples in the training set are under-sampled using the Tomek Links algorithm.

[0080] Furthermore, the steps of the multi-scale segmentation algorithm are as follows:

[0081] (1) Evenly divide the data segment into N1 sub-segments;

[0082] (2) Evenly divide the data segment into N2 sub-segments, where N2>N1;

[0083] (3) Evenly divide the data segment into N3 sub-segments, where N3>N2;

[0084] (4) Repeat the above steps until the number of sub-segments Nm reaches the set number;

[0085] (5) For all sub-segment data (the number is N1+N2+…+Nm), extract their time series statistical features to increase the sample size.

[0086] Specifically, the multi-scale segmentation sampling algorithm segments the original data into 2, 3, 5, and 10 parts in turn, and then generates samples from each sub-segment.

[0087] Furthermore, the basic steps of the Tomek Links algorithm are: 1) calculate the distance between each majority class sample and the minority class sample based on the Euclidean distance; 2) identify the sample closest to the minority class sample from the majority class sample; 3) delete the majority class sample in Tomeklinks to achieve downsampling of the majority class sample; 4) repeat steps 2) and 3) until the specified number of iterations is reached.

[0088] It should be noted that the original data is generally extremely unbalanced, where the sample sizes of normal excavation state, jam warning state, and jam state in the original training set are 419, 6, and 21, respectively, and the sample sizes of normal excavation state, jam warning state, and jam state in the test set are 444, 6, and 19, respectively. This step can make the data types of the preprocessed training set tend to be balanced.

[0089] Step S600, adopting a weight adaptive algorithm, using the preprocessed training set for training, to obtain a risk warning model.

[0090] In a possible implementation, step S500 includes the following steps:

[0091] S510, initializing the category weights w and the number of nearest neighbors k of the three TBM operating states, and initializing the heuristic algorithm hyperparameters;

[0092] S520, using the particle swarm optimization algorithm formula to update the particle velocity v w 、v k , update the particle's displacement parameters w and k, and perform the nearest rounding operation on k [k];

[0093] S530, based on the training set and w, [k], calculate the sum of weighted distances of samples of each category, and the category with the largest sum of weighted distances is the final category of the target point; use a cross-validation algorithm to verify the fitness of the model on the test set;

[0094] S540, determining the optimal particle of the population in each iteration by comparing the fitness values;

[0095] S550, determining the optimal particle of the population by comparing the fitness values;

[0096] S560, iterating until the optimal sample of the population converges or reaches a specified iteration number;

[0097] S570, obtaining the optimal category weight w and the number of nearest neighbors k.

[0098] Furthermore, in step 530, the calculation formula of the sum of weighted distances is:

[0099]

[0100] Among them, D 总 is the sum of weighted distances of the target point’s i-th class of neighboring samples, D i,j is the corrected distance of the jth neighbor sample of the i-th category of the target point, w i is the category weight of the i-th sample, i represents the sample category, and j represents the number of the neighboring point of a certain category; It is the weight between the target point and the neighboring point in the KNN algorithm, which is generally the same value or the inverse of the Euclidean distance.

[0101] Furthermore, in step 5, the fitness function is defined as the classification performance F1 of the model on the test set.

[0102] The following is an explanation with reference to specific embodiments.

[0103] In this embodiment, the steps in S500 are as follows:

[0104] S510, initialize the category weight w, w = [w1, w2, w3]. w1, w2, w3 are the category weights of normal excavation state, jammed machine warning state and jammed machine state respectively, the initial category weights are 1 respectively, that is, w1 = 1, w2 = 1, w3 = 1 and the number of nearest neighbors k (an integer is selected arbitrarily between 3 and 25, such as k = 12), initialize the number of particles in the particle swarm (set to 100), and the number of iterations (set to 50);

[0105] S520, updating the particle velocity v according to the particle swarm optimization algorithm formula w 、v k , update the particle's displacement w, k, and perform the nearest rounding operation on k [k];

[0106] S530, based on the training set and w, [k], the sum of weighted distances of samples of each category is calculated, the category with the largest sum of weighted distances is the final category of the target point, and the fitness of the model on the test set is verified by using a cross-validation algorithm;

[0107] S540, determining the optimal particle of the population in each iteration by comparing the fitness values;

[0108] S550, determining the optimal particle of the population by comparing the fitness values;

[0109] S560, repeating steps 2 to 5 until reaching a specified number of iterations;

[0110] S570, obtain the optimal category weight w = [0.89, 1.88, 1.39] and the number of nearest neighbors k = 7.

[0111] It should be noted that the category weight adaptive k-nearest neighbor algorithm (CWA-KNN) described in this application is optimized for the output category (the weight w of the three operating states of the TBM), so that the model can focus on the recognition accuracy of the key state of the card machine warning state, which is conducive to improving the recognition accuracy of the card machine warning state.

[0112] Figure 2 The performance of KNN and CWA-KNN models on the test set at different data processing stages provided by the embodiment of the present invention. Original, MSSS and TomekLinks respectively represent that the model is trained on the original training set, training set 1 (the original training set is processed by the multi-scale segmentation sampling algorithm (MSSS)) and training set 2 (training set 1 is processed by the TomekLinks algorithm). The results show that: 1) at different data processing stages, the F1 of the CWA-KNN model on the test set is higher than that of the KNN model, with an average increase of 0.024; 2) after being processed by the MSSS and TomekLinks algorithms, the F1 of the CWA-KNN model on the test set is improved by 0.132 and 0.164 respectively compared with the original training set; 3) the F1 of the open TBM rockburst risk warning method based on TBM operation data and the CWA-KNN algorithm is as high as 0.875 on the test set.

[0113] The method described in the embodiment of the present application was used in actual TBM operations to successfully implement machine jam warning 2 days before the machine jam, thus achieving early intervention and effectively reducing the risk of construction delays and TBM damage.

[0114] The open TBM rockburst and stuck risk warning device provided by the present invention is described below. The open TBM rockburst and stuck risk warning device described below and the open TBM rockburst and stuck risk warning method described above can be referenced to each other.

[0115] Figure 3 is a schematic diagram of the structure of an open TBM rockburst risk warning device provided by an embodiment of the present invention, such as Figure 3 As shown, it includes: a state division module 31, a screening module 32, an acquisition module 33, a data division module 34, a preprocessing module 35, a model acquisition module 36 and a risk warning module 37, wherein:

[0116] A state classification module 31 is used to classify the operation state of the TBM into three states: normal excavation state, machine jam warning state and machine jam state;

[0117] A screening module 32, for screening out key operating parameters;

[0118] An acquisition module 33 is used to acquire a data set; the data set includes input data and output data, wherein the input data is a time series of key operating parameters, and the output data is an operating state of the corresponding TBM;

[0119] A data partitioning module 34 is used to partition the data set into a training set and a test set;

[0120] A preprocessing module 35 is used to preprocess the training set to obtain a preprocessed training set;

[0121] The model acquisition module 36 is used to adopt a weight adaptive algorithm and use the pre-processed training set for training to obtain a risk warning model;

[0122] The risk warning module 37 is used to use the risk warning model to warn of the risk of rock burst and jamming of the open TBM.

[0123] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the open TBM rockburst risk warning method.

[0124] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0125] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the open TBM rockburst and sticking risk warning methods provided by the above methods.

[0126] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the open TBM rockburst and sticking risk warning method provided by the above methods.

[0127] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0128] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for early warning of rock burst and jamming risk of open TBM, characterized in that: The following steps are involved: The operating status of the TBM is divided into three states: normal excavation state, machine jam warning state and machine jam state; Screen out key operating parameters; Obtaining a data set; the data set includes input data and output data, wherein the input data is a time series statistical feature of a key operating parameter, and the output data is an operating state of a corresponding TBM; Divide the dataset into training and testing sets; Preprocessing the training set to obtain a preprocessed training set; Adopting weight adaptive algorithm, using pre-processed training set for training, to obtain risk warning model; The risk warning model is used to warn of the risk of rockburst and jamming of open TBM.

2. The open TBM rockburst and stuck risk early warning method according to claim 1 is characterized in that: The starting time of the machine jam warning state is within two days before the machine jam occurs, when the thrust gradually increases to 90% of the rated value and has a trend of continuing to increase, and the torque begins to have a downward trend.

3. The open TBM rockburst and stuck risk early warning method according to claim 1 is characterized in that: The key operating parameters include thrust, torque, cutter head speed, net excavation speed, 1# top shield cylinder rodless chamber pressure, 2# top shield cylinder rodless chamber pressure, left shield cylinder rodless chamber pressure, left wedge cylinder rodless chamber pressure, right shield cylinder rodless chamber pressure, right wedge cylinder rodless chamber pressure, top shield displacement, left shield displacement and right shield displacement.

4. The open TBM rockburst and jamming risk early warning method according to claim 1 is characterized in that: The method for obtaining the time series statistical characteristics of the key operating parameters comprises the following steps: The collected TBM operation data is divided into multiple data segments according to the day and the operation status; Filtering data in a tunneling working state from the multiple data segments as a first data set; Eliminate the data rows containing missing values ​​in the first data set, use the isolation forest algorithm to detect and delete outliers in the selected data segments, and obtain a second data set; Normalizing the second data set to obtain a normalized data set; The time series statistical characteristics of key operating parameters are extracted from the normalized data set; the time series statistical characteristics of the key operating parameters include maximum value, minimum value, mean value, standard deviation, root mean square, kurtosis factor and skewness.

5. The open TBM rockburst and stuck risk early warning method according to claim 1, characterized in that: The method of preprocessing the training set to obtain the preprocessed training set comprises the following steps: For the samples of machine warning status and machine status in the training set, a multi-scale segmentation sampling algorithm is used to increase the sample size; The normal excavation state samples in the training set are under-sampled using the Tomek Links algorithm.

6. The open TBM rockburst and stuck risk early warning method according to claim 1, characterized in that: The method of using a weight adaptive algorithm to train with a preprocessed training set to obtain a risk warning model includes the following steps: Initialize the category weights w and the number of nearest neighbors k of the three operating states of TBM, and initialize the hyperparameters of the heuristic algorithm; Use the particle swarm optimization algorithm formula to update the particle velocity v w 、v k , update the particle's displacement parameters w and k, and perform the nearest rounding operation on k [k]; Based on the training set and w, [k], the sum of the weighted distances of samples in each category is calculated. The category with the largest sum of weighted distances is the final category of the target point. The fitness of the model on the test set is verified using the cross-validation algorithm. By comparing the fitness values, the optimal sample of the population in each iteration is determined; Determine the optimal sample of the population by comparing the fitness values; Iterate until the optimal sample of the population converges or reaches the specified iteration number; Get the optimal category weight w and the number of nearest neighbors k.

7. An open TBM rockburst and jam risk warning device, characterized in that: include: The state classification module is used to classify the operation state of the TBM into three states: normal excavation state, machine jam warning state and machine jam state; A screening module is used to screen out key operating parameters; An acquisition module, used to acquire a data set; the data set includes input data and output data, wherein the input data is a time series of key operating parameters, and the output data is an operating state of a corresponding TBM; Data partitioning module, used to divide the data set into training set and test set; A preprocessing module, used for preprocessing the training set to obtain a preprocessed training set; A model acquisition module is used to adopt a weight adaptive algorithm and use a preprocessed training set for training to obtain a risk warning model; The risk warning module is used to use the risk warning model to warn of the risk of rock burst and jamming of open TBM.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the open TBM rockburst and sticking risk warning method as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the open TBM rockburst and sticking risk warning method as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the open TBM rockburst and sticking risk warning method as described in any one of claims 1 to 6 is implemented.

Images