Rock burst prediction method and device of roadway and electronic equipment

Abstract

The invention discloses a roadway rock burst prediction method and device and electronic equipment. The method comprises the steps that a target roadway in a mine is monitored, micro-seismic monitoring data are obtained, and the micro-seismic monitoring data are used for representing the degree of a vibration event of a coal seam and a rock stratum located around the target roadway; based on the micro-seismic monitoring data, rock burst monitoring indexes of the target roadway are determined, the rock burst monitoring indexes are used for representing precursor characteristics and evolution rules of rock burst in the target roadway, and the rock burst is formed by accumulation of internal energy of a coal seam and internal energy of a rock stratum in the vibration event process; the impact strength grade of the rock burst is predicted based on the rock burst monitoring indexes through a prediction model, and the prediction model is obtained by training a neural network model through rock burst monitoring index samples and impact strength grade samples. The technical problem that the prediction accuracy of the rock burst of the roadway is low is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine safety monitoring, and in particular to a method, device and electronic equipment for predicting rock burst in a tunnel. Background Art

[0002] At present, rock burst is an extremely serious dynamic disaster in mines. When rock burst occurs in mine tunnels, it will cause serious damage to the coal and rock layers in the tunnels, and even cause gas explosions and coal dust explosions, directly threatening the lives of personnel. Therefore, it is of great significance to predict rock burst, take preventive measures in time, and reduce the damage caused by rock burst.

[0003] In the related art, in the above prediction process, statistical theories and mathematical methods can be used to predict the rock burst in the tunnel, such as Gaussian distribution, Poisson distribution, time series analysis, etc. However, since the above prediction method uses a small amount of information and is simple in calculation, it cannot effectively and accurately reflect the dynamic changes and impact danger of the underground strata in the mine, making the above method only suitable for a rough prediction of rock burst disasters. Therefore, there is still a technical problem of low accuracy in predicting rock burst in the tunnel.

[0004] Currently, no effective solution has been proposed for the above technical problems. Summary of the invention

[0005] The embodiments of the present invention provide a method, device and electronic device for predicting rock burst in a tunnel, so as to at least solve the technical problem of low accuracy in predicting rock burst in a tunnel.

[0006] According to one aspect of an embodiment of the present invention, a method for predicting rock burst in a tunnel is provided. The method may include: monitoring a target tunnel in a mine to obtain microseismic monitoring data, wherein the microseismic monitoring data is used to indicate the degree of vibration events in the coal seams and rock formations around the target tunnel; determining a rock burst monitoring index for the target tunnel based on the microseismic monitoring data, wherein the rock burst monitoring index is used to indicate the precursor characteristics and evolution law of rock burst in the target tunnel, and the rock burst is formed by the accumulation of internal energy of the coal seam and the internal energy of the rock formation during the vibration event; using a prediction model, based on the rock burst monitoring index, predicting the impact intensity level of the rock burst, wherein the prediction model is obtained by training a neural network model using rock burst monitoring index samples and rock burst intensity level samples, and the impact intensity level is positively correlated with the danger level of the rock burst.

[0007] Optionally, the rock burst monitoring index includes at least one of the following: a first vibration index, a second vibration index, an energy release index, an energy level change index, an average energy level index, a missing value, the number of microseismic events, the logarithm of microseismic energy release, the microseismic event rate and the logarithm of the cumulative microseismic energy release rate, wherein the first vibration index is used to indicate the relationship between the energy level and frequency of vibration events, the second vibration index is used to indicate the target energy microseismic events in the vibration events within the mining range of the working face in the mine, the energy release index is used to indicate the occurrence time, space and intensity of the vibration events; the energy level change index is used to indicate the energy level change characteristics of the vibration events in the time series; the average energy level index is used to indicate the average energy level status of the vibration events.

[0008] Optionally, based on the microseismic monitoring data, the rock burst monitoring index of the target tunnel is determined, including: based on the microseismic monitoring data, recording data records corresponding to the rock burst monitoring index through a preset calculation window and a sliding step, wherein the data records include time information of the vibration event, the rock burst monitoring index and identification information of the preset calculation window; the method also includes: merging the data records according to the number of preset calculation windows to obtain a data record set; obtaining the number of records recording the rock burst monitoring index, and merging the impact intensity levels based on the number of records to obtain an impact intensity level set.

[0009] Optionally, the method also includes: standardizing the data record set obtained at the historical moment, and normalizing the standardized data record set; using the normalized data record set as the rock burst monitoring index sample, and using the rock burst intensity level set obtained at the historical moment as the rock burst intensity level sample; dividing the rock burst monitoring index sample and the rock burst intensity level sample according to a preset ratio to obtain a training set and a test set; inputting the training set into the neural network model, training the neural network model using the training set, and inputting the training set into the trained neural network model, testing the trained neural network model using the test set; in response to the accuracy of the test set being higher than the accuracy threshold, using the trained neural network model as a prediction model.

[0010] Optionally, the data record set obtained at the historical moment is standardized, including: determining the maximum data record and the minimum data record from different data records in the data record set; based on the maximum data record and the minimum data record, standardizing the different data records in the data record set to obtain the standardized data record set; the method also includes: based on the impact intensity level, standardizing the energy data corresponding to the vibration event to obtain the standardized energy data; based on the impact intensity level sample, standardizing the energy data sample corresponding to the vibration event that occurred at the historical moment to obtain the standardized energy data sample.

[0011] Optionally, before inputting the training set into the neural network model and using the training set to train the neural network model, the method also includes: determining the excellent value of the rock burst monitoring index samples based on the first average correlation between the rock burst monitoring index samples and the impact intensity level samples in the training set and using the second average correlation between the rock burst monitoring index samples; determining the rock burst monitoring index samples whose excellent values ​​are greater than the excellent threshold as a rock burst monitoring index sample subset; and using the rock burst monitoring index sample subset and the corresponding impact intensity level samples as the training set to be input into the neural network model.

[0012] Optionally, the training set is input into the neural network model, and the neural network model is trained using the training set, including: inputting the training set into the neural network model, and training the neural network model using a genetic algorithm in combination with the training set.

[0013] Optionally, after the training set is input into the trained neural network model and the trained neural network model is tested using the test set, the method further includes: obtaining a test error of the trained neural network model using the test set, and the fitness corresponding to the hyperparameters of the trained neural network model, wherein the hyperparameters include at least the population size, number of iterations, crossover probability and mutation probability of the genetic algorithm; in response to the test error exceeding the test error threshold, selecting a hyperparameter with a fitness higher than the fitness threshold for copying, and performing a crossover operation and a mutation operation on the hyperparameter; in response to the number of iterations of performing the crossover operation and the mutation operation exceeding the iteration number threshold, determining the target hyperparameter; based on the target hyperparameter, determining the target weight and target threshold of the neural network model to obtain a prediction model.

[0014] According to another aspect of an embodiment of the present invention, a device for predicting rock burst in a tunnel is also provided. The device may include: a monitoring unit, which is used to monitor a target tunnel in a mine and obtain microseismic monitoring data, wherein the microseismic monitoring data is used to indicate the degree of vibration events occurring in coal seams and rock formations around the target tunnel; a determination unit, which is used to determine rock burst monitoring indicators of the target tunnel based on the microseismic monitoring data, wherein the rock burst monitoring indicators are used to indicate the precursor characteristics and evolution laws of rock burst in the target tunnel, and the rock burst is formed by the accumulation of internal energy of the coal seam and the internal energy of the rock formation during the vibration event; a prediction unit, which is used to predict the impact intensity level of the rock burst based on the rock burst monitoring indicators using a prediction model, wherein the prediction model is obtained by training a neural network model using rock burst monitoring indicator samples and rock burst intensity level samples, and the impact intensity level is positively correlated with the danger level of the rock burst.

[0015] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute a method for predicting rock burst in a tunnel.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for predicting rock burst in a tunnel according to the present invention.

[0017] According to another aspect of an embodiment of the present invention, a processor is provided, which is used to run a program, wherein when the program is run, the method for predicting rock burst in a tunnel according to an embodiment of the present invention is executed.

[0018] According to another aspect of the present invention, a computer program product is provided, which includes a computer program, and when executed by a processor, the computer program implements the method for predicting rock burst in a tunnel according to the embodiment of the present application.

[0019] In an embodiment of the present invention, if rock burst is to be predicted, microseismic monitoring data can be obtained by monitoring the target tunnel in the mine to indicate the degree of vibration events in the coal seams and rock formations around the target tunnel; based on the microseismic monitoring data, the rock burst monitoring index of the target tunnel can be determined; the rock burst monitoring index samples and the impact intensity level samples can be used to train the neural network model to obtain a prediction model, and the impact intensity level of the rock burst can be predicted based on the rock burst monitoring index. In this embodiment, by inputting the rock burst monitoring index into the prediction model, the precursor characteristics and evolution law of the rock burst can be effectively grasped, and a more accurate impact intensity level can be output, thereby achieving the technical effect of improving the accuracy of the prediction of the rock burst in the tunnel, and solving the technical problem of low accuracy in the prediction of the rock burst in the tunnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0021] Figure 1 is a flow chart of a method for predicting rock burst in a tunnel according to an embodiment of the present invention;

[0022] Figure 2 It is a schematic diagram of a prediction model for rock burst multi-index fusion in a steeply inclined and extremely thick coal seam excavation tunnel according to an embodiment of the present invention;

[0023] Figure 3 is a schematic diagram of an energy level-frequency distribution curve according to an embodiment of the present invention;

[0024] Figure 4 is a flow chart of a neural network model optimized by a genetic algorithm according to an embodiment of the present invention;

[0025] Figure 5 It is a schematic diagram of a device for predicting rock burst in a tunnel according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] According to an embodiment of the present invention, an embodiment of a method for predicting rock burst pressure in a tunnel is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0029] Figure 1 is a flow chart of a method for predicting rock burst in a tunnel according to an embodiment of the present invention. Figure 1 As shown, the method may include the following steps:

[0030] Step S102, monitoring the target tunnel in the mine to obtain microseismic monitoring data.

[0031] In the technical solution provided in step S102 of the present invention, the microseismic monitoring data is used to indicate the extent of the vibration events occurring in the coal seams and rock strata around the target tunnel. Microseismic monitoring data may also be referred to as microseismic data or source data. The target tunnel may be a steeply inclined and thick coal seam excavation tunnel. If the extent of the vibration event is small, it may be referred to as a microvibration or microseismic event.

[0032] In this embodiment, if it is necessary to predict the impact intensity of rock burst in a tunnel in a mine, the target tunnel in the mine can be monitored to obtain microseismic monitoring data.

[0033] For example, seismographs, microseismic sensors and other equipment can be used to monitor target tunnels in mines to obtain microseismic monitoring data. That is, the above equipment can be used to monitor and record the vibration activity of underground coal seams and rock formations to obtain microseismic monitoring data, such as the source location, energy level / microseismic energy level of microseismic events, source mechanism and other data.

[0034] It should be noted that the above-mentioned equipment for monitoring the target tunnel in the mine and the microseismic detection data obtained are only examples and are not specifically limited here. As long as it is a device that can be used to monitor the microseismic monitoring data of the target tunnel, and the data that can reflect the degree of vibration events in the coal seams and rock formations around the tunnel, they are all within the protection scope of the embodiments of the present invention.

[0035] Alternatively, microseismic monitoring data is particularly important in the monitoring process of steeply inclined and thick coal seam excavation tunnels, because the stress distribution and release pattern of the rock formation under the geological conditions of the steeply inclined and thick coal seam excavation tunnels are more complex and irregular, and vibration events may be more frequent and the energy release is more intense. By accurately and comprehensively collecting microseismic monitoring data and using microseismic monitoring data, the activity state of the coal and rock formations around the target tunnel can be determined, possible geological disasters can be predicted, and the excavation safety of the tunnel can be ensured.

[0036] Step S104, determining the rock burst monitoring index of the target tunnel based on the microseismic monitoring data.

[0037] In the technical solution provided in the above step S104 of the present invention, after monitoring the target tunnel in the mine and obtaining the microseismic monitoring data, the rock burst monitoring index of the target tunnel can be determined based on the microseismic monitoring data. Among them, the rock burst monitoring index is used to represent the precursor characteristics and evolution law of the rock burst in the target tunnel, and can be used as the main influencing factor of the rock burst. The rock burst is formed by the accumulation of internal energy of the coal seam and the internal energy of the rock formation during the vibration event, that is, the rock burst can be the result of the accumulation and sudden release of internal energy of the rock formation or the coal seam.

[0038] In this embodiment, the rock burst monitoring index can be determined based on the microseismic monitoring data. The determined rock burst monitoring index can fully and accurately reflect the dynamic changes of the rock and coal seams around the target tunnel, and provide strong data support for the prediction of the rock burst intensity level, thereby guiding the mine to take effective anti-rock burst measures and ensure the safety of mine operations.

[0039] Optionally, the microseismic data collected during the process of mining coal or other mineral resources by digging tunnels to gradually go deeper into the mine can be used as source data to calculate multiple rock burst monitoring indicators, identify precursor features closely related to the occurrence of rock burst, such as magnitude distribution, energy level change frequency, energy dynamic changes, etc., and obtain the internal evolution law of the rock formation before the occurrence of rock burst, such as identifying the triggering conditions of rock burst, evaluating the frequency and energy distribution pattern of rock burst, etc.

[0040] Step S106, using the prediction model, based on the rock burst monitoring index, predicting the rock burst intensity level.

[0041] In the technical solution of step S106 of the present invention, after determining the rock burst monitoring index of the target tunnel based on the microseismic monitoring data, the prediction model can be used to predict the impact intensity level of the rock burst based on the rock burst monitoring index. Among them, the prediction model can be obtained by training a neural network model using rock burst monitoring index samples and rock burst intensity level samples, and the impact intensity level is positively correlated with the degree of danger of rock burst. The neural network model can be a back propagation (BP) neural network model. The impact intensity level can be called an impact hazard level or impact type, such as micro-shock, weak shock, medium shock and strong shock. It should be noted that the above-mentioned impact intensity level division method and division results are only for illustrative purposes and are not specifically limited here.

[0042] In this embodiment, the rock burst monitoring index may be input into the prediction model, and the prediction model may be used to output the rock burst type based on the pre-learned association between the rock burst monitoring index and the rock burst intensity level.

[0043] Optionally, different impact types can correspond to different anti-impact measures. For example, if the impact intensity level of rock burst is predicted to be a strong impact danger level, mining activities can be suspended, personnel can be evacuated, and measures such as stress release or tunnel reinforcement can be implemented; if the impact intensity level of rock burst is predicted to be a weak impact danger level, monitoring can be strengthened and observation can be continued.

[0044] Optionally, the prediction model can extract key features from the rock burst monitoring indicators, that is, the prediction model can extract indicators or combinations that have a significant impact on the predicted rock burst level and can significantly characterize the dynamic activities inside the rock formation from the rock burst detection indicators. For example, energy release characteristics, event frequency characteristics, energy classification characteristics, rock burst monitoring indicators, etc., so as to predict the impact intensity level of rock burst and provide real-time danger prediction for the mine. The above-mentioned machine learning-based tunnel rock burst prediction method can not only improve the accuracy and timeliness of the prediction, but also adapt to complex and changeable geological conditions, and provide strong support for the prediction of rock burst in steeply inclined and thick coal seam excavation tunnels.

[0045] In the above steps S102 to S106 of the present application, if the rock burst is to be predicted, the target tunnel in the mine can be monitored to obtain microseismic monitoring data to indicate the degree of vibration events in the coal seams and rock formations around the target tunnel; based on the microseismic monitoring data, the rock burst monitoring index of the target tunnel can be determined; the prediction model obtained by training the neural network model using the rock burst monitoring index samples and the impact intensity level samples can be used to predict the impact intensity level of the rock burst based on the rock burst monitoring index. In this embodiment, by inputting the rock burst monitoring index into the prediction model, the precursor characteristics and evolution laws of the rock burst can be effectively grasped, and a more accurate impact intensity level can be output, thereby achieving the technical effect of improving the accuracy of the prediction of the rock burst in the tunnel, and solving the technical problem of low accuracy in the prediction of the rock burst in the tunnel.

[0046] The above method of this embodiment is further introduced below.

[0047] As an optional implementation method, the rock burst monitoring index includes at least one of the following: a first vibration index, a second vibration index, an energy release index, an energy level change index, an average energy level index, a missing value, the number of microseismic events, the logarithm of microseismic release energy, the microseismic event rate and the logarithm of the cumulative microseismic energy release rate, wherein the first vibration index is used to indicate the relationship between the energy level and frequency of a vibration event, the second vibration index is used to indicate the target energy microseismic event in the vibration event within the mining range of the working face in the mine, the energy release index is used to indicate the occurrence time, space and intensity of the vibration event; the energy level change index is used to indicate the energy level change characteristics of the vibration event in the time series; the average energy level index is used to indicate the average energy level status of the vibration event.

[0048] Optionally, the first vibration index, that is, the b value, can be determined by the following formula:

[0049]

[0050] Where m can be used to represent the total number of microseismic energy levels; N i It can be used to indicate the total number of microseisms at the i-th energy level.

[0051] Optionally, the second vibration index, that is, the A(b) value, can be determined by the following formula:

[0052]

[0053] Wherein, b can be used to represent the b value within the calculation window; M i It can be used to indicate the energy level of microseismic events.

[0054] Optionally, the energy release index, that is, the mine seismic activity S value, can be determined by the following formula:

[0055]

[0056] Among them, N and M i They can be used to indicate the total number of microseismic events (number of microseismic events) and the energy level of occurrence respectively; M can be used to indicate the maximum energy.

[0057] Optionally, the energy level change index, that is, the algorithm complexity AC value, can be determined by the following formula:

[0058]

[0059] Among them, n can be used to represent the number of energy level changes in a specific time period; M can be expressed by M max -M min +1 OK, M max It can be used to indicate the maximum energy level of a microseismic event; M min It can be used to represent the minimum energy level of a microseismic event.

[0060] Optionally, the average energy level index, that is, the equivalent energy level parameter EEM value, can be determined by the following formula:

[0061]

[0062] Among them, m* can be used to represent the normalized energy level of microseismic events; M* can be used to represent the average normalized energy level of m*.

[0063] Optionally, the missing value can be determined by the following formula:

[0064]

[0065] Among them, M and M0 can be used to represent the average energy level and the initial energy level of microseismic events in the area within the time period, respectively; n can be used to represent the total number of microseismic events within the statistical range.

[0066] Optionally, the number of microseismic events can be represented by N; the logarithm of microseismic energy release can be represented by E; the microseismic event rate can be represented by NR; and the logarithm of the microseismic cumulative energy release rate can be represented by ER.

[0067] In this embodiment, the number of microseismic events can reflect the frequency of microfracture activities inside the rock formation. The logarithm of microseismic energy release can quantify the intensity of energy release inside the rock formation. The microseismic event rate can reflect the intensity and stability of dynamic activities inside the rock formation. The logarithm of the microseismic cumulative energy release rate can reflect the logarithmic change rate of the cumulative energy released by microseismic events over a period of time, and can provide dynamic information on energy accumulation and release inside the rock formation. Through the above four rock burst monitoring indicators, a multi-dimensional monitoring system can be constructed to comprehensively capture the characteristics of dynamic activities inside the rock formation, which has a good indicative effect on the occurrence of rock burst.

[0068] As an optional implementation method, step S104, based on microseismic monitoring data, determines the rock burst monitoring index of the target tunnel, including: based on the microseismic monitoring data, through a preset calculation window and a sliding step, recording the data record corresponding to the rock burst monitoring index, wherein the data record includes the time information of the vibration event, the rock burst monitoring index and the identification information of the preset calculation window; the method also includes: merging the data records according to the number of preset calculation windows to obtain a data record set; obtaining the number of records recording the rock burst monitoring index, and merging the impact intensity levels based on the number of records to obtain an impact intensity level set.

[0069] In this embodiment, in the process of determining the rock burst monitoring index of the target tunnel based on microseismic monitoring data, data records corresponding to the rock burst monitoring index can be recorded based on the microseismic monitoring data through a preset calculation window and a sliding step, and the data records include time information of the vibration event, rock burst monitoring index and identification information of the preset calculation window; the data records can also be merged according to the number of preset calculation windows to obtain a data record set; the number of records recording the rock burst monitoring index is obtained, and the impact intensity levels are merged based on the number of records to obtain an impact intensity level set.

[0070] Optionally, the preset calculation window can be divided according to days, and the data record set can be expressed by the following formula:

[0071] w i =(id,date,e1,e2,e3,e4,e5,e6,e7,e8,e9,e 10 )

[0072] Among them, id can be used to represent identification information; date can be used to represent time information (date and time); e1 to e10 It can be used to indicate rock burst monitoring indicators.

[0073] Thus, we can obtain the time series data divided into n windows, which can be expressed by the following formula:

[0074] Window=(w_0,w_1,,...,w_{n-1})

[0075] Optionally, a label set of each corresponding impact strength index G is established, that is, an impact strength grade set, which can be expressed by the following formula:

[0076] G=(t0,t1,t2,......,t D )

[0077] Where D can be used to represent the number of overall statistical days, that is, the number of records; t i It can be used to represent the impact intensity level label of the i-th day in the time window sequence.

[0078] As an optional implementation method, the method also includes: standardizing the data record set obtained at the historical moment, and normalizing the standardized data record set; using the normalized data record set as the rock burst monitoring index sample, and using the rock burst intensity level set obtained at the historical moment as the rock burst intensity level sample; dividing the rock burst monitoring index sample and the rock burst intensity level sample according to a preset ratio to obtain a training set and a test set; inputting the training set into the neural network model, training the neural network model using the training set, and inputting the training set into the trained neural network model, testing the trained neural network model using the test set; in response to the accuracy of the test set being higher than the accuracy threshold, using the trained neural network model as a prediction model.

[0079] In this embodiment, the data record set obtained at the historical moment can be standardized, and the standardized data record set can be normalized; the normalized data record set can be used as a rock burst monitoring index sample, and the rock burst intensity level set obtained at the historical moment can be used as an rock burst intensity level sample; the rock burst monitoring index sample and the rock burst intensity level sample can be divided according to a preset ratio to obtain a training set and a test set; the training set can be input into the neural network model, the neural network model can be trained using the training set, and the training set can be input into the trained neural network model, and the trained neural network model can be tested using the test set; the trained neural network model can be used as a prediction model when the accuracy of the test set is higher than the accuracy threshold.

[0080] Optionally, through standardization and normalization processing, the problem of imbalance in neural network model training caused by different magnitudes of data record sets can be avoided, and the generalization ability and prediction accuracy of the neural network model can be improved.

[0081] Optionally, the normalized data record set is used as the rock burst monitoring index sample, and the rock burst intensity level set obtained at the historical moment is used as the rock burst intensity level sample. The above samples can be divided into a training set and a test set according to a preset ratio (such as 8:2). The training set can be used to train the neural network model so that the neural network model can learn the relationship between the rock burst monitoring index and the rock burst intensity level, and the test set can be used to verify the prediction performance of the neural network model. When the accuracy of the test set is higher than the accuracy threshold, it means that the neural network model has a high accuracy in predicting the rock burst intensity level on unseen data, and the trained neural network model can be used as a prediction model.

[0082] As an optional implementation method, the data record set obtained at the historical moment is standardized, including: determining the maximum data record and the minimum data record from different data records in the data record set; based on the maximum data record and the minimum data record, standardizing the different data records in the data record set to obtain the standardized data record set; the method also includes: based on the impact intensity level, standardizing the energy data corresponding to the vibration event to obtain the standardized energy data; based on the impact intensity level sample, standardizing the energy data sample corresponding to the vibration event that occurred at the historical moment to obtain the standardized energy data sample.

[0083] In this embodiment, in the process of standardizing the data record set acquired at the historical moment, the maximum data record and the minimum data record can be determined from the different data records in the data record set; the different data records in the data record set can be standardized based on the maximum data record and the minimum data record to obtain the standardized data record set. In the process of standardizing the energy data samples after the data is standardized, the energy data corresponding to the vibration event can be standardized based on the impact intensity level to obtain the standardized energy data; the energy data samples corresponding to the vibration event occurring at the historical moment can be standardized based on the impact intensity level samples to obtain the standardized energy data samples. Among them, the energy data can be single energy data.

[0084] Optionally, the data record set obtained at the historical moment is standardized, that is, each sequence indicator and single energy data is standardized, which can reduce the magnitude difference between the indicators and improve the overall expressiveness of the energy data. The target energy value of each sequence indicator set is normalized and the indicator is converted into a relative value, which can eliminate the difference between positive and negative indicators, thereby improving the accuracy of the prediction. Data standardization can be performed using the following formula:

[0085] Positive indicators:

[0086] Negative indicators:

[0087] Bidirectional indicators:

[0088] Among them, R i Can be used to represent the i-th value of a sequence; R max Can be used to represent the peak value of all data in the sequence; R min It can be used to represent the valley value of all data in the sequence, that is, the minimum data record; R max Can be used to represent the calculation window R i The maximum value of the sequence, that is, the largest data record; R avg It can be used to calculate the average value of the window indicator series.

[0089] Optionally, the energy data corresponding to the vibration event can be standardized for processing the energy data of a single microseismic event. The energy value of the highest energy event occurring on that day can be used as the level of impact rock pressure monitoring and early warning, and the corresponding single energy data can be standardized according to the interval divided by the microseismic energy level.

[0090] As an optional implementation method, before the training set is input into the neural network model and the neural network model is trained using the training set, the method also includes: determining the excellent value of the rock burst monitoring index samples based on the first average correlation between the rock burst monitoring index samples and the impact intensity level samples in the training set and using the second average correlation between the rock burst monitoring index samples; determining the rock burst monitoring index samples whose excellent values ​​are greater than the excellent threshold as a rock burst monitoring index sample subset; and using the rock burst monitoring index sample subset and the corresponding impact intensity level samples as the training set to be input into the neural network model.

[0091] In this embodiment, before the training set is input into the neural network model and the neural network model is trained using the training set, the excellence value of the rock burst monitoring index samples can be determined based on the first average correlation between the rock burst monitoring index samples and the impact intensity level samples in the training set and the second average correlation between the rock burst monitoring index samples; the rock burst monitoring index samples whose excellence value is greater than the excellence threshold can be determined as the rock burst monitoring index sample subset; the rock burst monitoring index sample subset and the corresponding impact intensity level samples can be used as the training set to be input into the neural network model. Among them, the rock burst monitoring index samples can be features, and the impact intensity level samples can be target variables / classes. The excellence threshold can be the maximum merit value, and the sample subset can be a feature subset.

[0092] Optionally, before inputting the neural network model training, a feature selection algorithm (Correlation-based Feature Selection, referred to as CFS) can be introduced to select a subset of rock burst monitoring indicator samples. The CFS algorithm selects features based on the correlation between features and the correlation between features and target variables. The heuristic equation can be determined by the following formula:

[0093]

[0094] Among them, Meit s It can be used to represent the heuristic excellence value of the feature subset S containing k features, that is, the excellence value of the rock burst monitoring indicator sample; It can be used to represent the feature-class average correlation, that is, the first average correlation; It can be used to represent the feature-feature average correlation, that is, the second average correlation.

[0095] If Meit s The larger the value, the greater the average correlation between the feature and the category, and the smaller the average redundancy between the features. s The smaller the value, the smaller the average correlation between the feature and the category, and the greater the average redundancy between features.

[0096] Optionally, the CFS algorithm reads the data set by loading data; calculates the excellence values ​​of all features; selects the feature with the maximum excellence value as the feature subset; selects the feature subset step by step through iteration; calculates the merit value of the current feature subset, and when the merit value of the current feature subset is greater than the maximum merit value, updates the maximum merit value and the current feature subset; when the excellence value decreases, removes the feature and traverses the next feature subset; checks whether all feature subsets have been processed, and if so, returns the selected feature subset. The CFS algorithm can be used to obtain the excellence values ​​of different feature subsets, and select a suitable subset of rock burst monitoring indicator samples and the corresponding rock burst intensity grade samples as training sets. Since the rock burst monitoring indicators of different mines have different sensitivities to rock burst prediction, the CFS algorithm can be used to select rock burst monitoring indicators suitable for different mines, thereby improving the generalization and prediction capabilities of the neural network model.

[0097] As an optional implementation method, the training set is input into the neural network model, and the neural network model is trained using the training set, including: inputting the training set into the neural network model, and training the neural network model using a genetic algorithm combined with the training set.

[0098] In this embodiment, in the process of inputting the training set into the neural network model and using the training set to train the neural network model, the training set can be input into the neural network model, and the neural network model can be trained using a genetic algorithm combined with the training set.

[0099] Alternatively, genetic algorithms can effectively process high-dimensional parameter spaces and find better neural network model configurations by simulating the evolutionary process. In the complex problem of rock burst prediction, genetic algorithms can help neural network models better adapt to changing geological conditions and improve the generalization ability and prediction accuracy of neural network models.

[0100] As an optional implementation mode, after the training set is input into the trained neural network model and the trained neural network model is tested using the test set, the method also includes: obtaining the test error of testing the trained neural network model using the test set, and the fitness corresponding to the hyperparameters of the trained neural network model, wherein the hyperparameters include at least the population size, number of iterations, crossover probability and mutation probability of the genetic algorithm; in response to the test error exceeding the test error threshold, selecting the hyperparameters with fitness higher than the fitness threshold for copying, and performing crossover and mutation operations on the hyperparameters; in response to the number of iterations of performing the crossover and mutation operations exceeding the iteration number threshold, determining the target hyperparameters; based on the target hyperparameters, determining the target weights and target thresholds of the neural network model to obtain a prediction model.

[0101] In this embodiment, after the training set is input into the trained neural network model and the trained neural network model is tested using the test set, the test error of the trained neural network model using the test set and the fitness corresponding to the hyperparameters of the trained neural network model can be obtained, wherein the hyperparameters include at least the population size, number of iterations, crossover probability and mutation probability of the genetic algorithm; when the test error exceeds the test error threshold, the hyperparameters with fitness higher than the fitness threshold can be selected for copying, and the crossover and mutation operations can be performed on the hyperparameters; when the number of iterations of performing the crossover and mutation operations exceeds the iteration number threshold, the target hyperparameters can be determined; based on the target hyperparameters, the target weights and target thresholds of the neural network model can be determined to obtain a prediction model.

[0102] Optionally, a genetic algorithm can be used to tune the hyperparameters of the neural network model. The steps of optimizing the BP neural network model using the genetic algorithm are as follows:

[0103] Import data and divide it into training set and test set;

[0104] Define the structure of the neural network model, including the number of input features, the number of hidden layer neurons, and the number of output features. Define the hyperparameters of the genetic algorithm, including population size, number of iterations, crossover probability, and mutation probability;

[0105] Encode the weights and thresholds of the neural network model corresponding to each individual to initialize the population;

[0106] The genetic algorithm is iterated by crossover, mutation, calculating the fitness of each new individual, selecting individuals with better performance to generate the next generation population, and repeating the above steps until the predetermined number of iterations is reached;

[0107] The BP neural network model is trained with hyperparameters optimized by genetic algorithm, and the performance of the optimized neural network model on the test set is evaluated.

[0108] Through the above steps, the genetic algorithm can optimize the weight and bias of the BP neural network model, improve the prediction performance of the neural network model, and obtain the optimal network model for prediction based on the rock burst index data of tunneling tunnels in different steeply inclined coal seams.

[0109] In an embodiment of the present invention, if rock burst is to be predicted, microseismic monitoring data can be obtained by monitoring the target tunnel in the mine to indicate the degree of vibration events in the coal seams and rock formations around the target tunnel; based on the microseismic monitoring data, the rock burst monitoring index of the target tunnel can be determined; the rock burst monitoring index samples and the impact intensity level samples can be used to train the neural network model to obtain a prediction model, and the impact intensity level of the rock burst can be predicted based on the rock burst monitoring index. In this embodiment, by inputting the rock burst monitoring index into the prediction model, the precursor characteristics and evolution law of the rock burst can be effectively grasped, and a more accurate impact intensity level can be output, thereby achieving the technical effect of improving the accuracy of the prediction of the rock burst in the tunnel, and solving the technical problem of low accuracy in the prediction of the rock burst in the tunnel.

[0110] The technical solution of the embodiment of the present invention is illustrated below in conjunction with preferred implementation modes.

[0111] At present, in order to ensure the safe and efficient excavation of steeply inclined and extremely thick coal seam tunnels, microseismic, ground sound and other methods are used to monitor various indicators of rock burst during the excavation process. By analyzing and summarizing various physical indicators to evaluate the degree of danger of rock burst, early warning indicators are provided for the excavation tunnel, and pressure relief measures are implemented in a timely manner. Based on microseismic monitoring data, machine learning methods are introduced into the field of rock mechanics, and intelligent models are integrated to study dynamic disaster problems, forming a data-driven intelligent early warning method that can effectively extract the characteristics of disaster precursor data and achieve effective early warning.

[0112] In the existing technology, on the one hand, there are few studies on rock burst warning for steeply inclined and thick coal seams, and most of them are based on nearly horizontal or gently inclined coal seams. For the extremely complex geological conditions of coal rock occurrence in steeply inclined and thick coal seams, there are few studies on the mining of rock burst precursor features and intelligent prediction. The existing early warning technology has a limited monitoring range, insufficient time and space resolution, and cannot capture rapidly changing or local information and cover all potential rock burst danger areas. On the other hand, the existing technology is not sufficient for the extraction of rock burst precursor information, and the generalization ability of the early warning model is poor. For example, the precursor information extracted by the existing early warning technology contains noise, which makes the signal-to-noise ratio low and difficult to accurately extract useful information. In addition, the existing early warning technology model is sensitive to noise or outliers, and some important precursor information cannot be extracted or there is confusing information, so the early warning of potential dangers is not timely or accurate, resulting in false alarms or missed alarms in practical applications. The early warning model may be affected by environmental factors, and the performance of data in different time periods or different locations is quite different, lacking robustness, resulting in low reliability of the early warning results.

[0113] The present invention aims at the intelligent prediction problem of rock burst in tunnels with steep inclination and thick coal seams. By revealing the precursor characteristics and evolution laws contained in multiple physical indicators of rock burst, a multi-indicator evolution data set of rock burst danger occurrence is constructed. Based on the machine learning algorithm, a multivariate data-driven intelligent prediction model of rock burst in tunnels with steep inclination and thick coal seams is constructed to obtain the sensitivity weights of each indicator under tunneling in steep inclination and thick coal seams, realize rock burst danger warning applicable to tunnels with steep inclination and thick coal seams, and timely take pressure relief measures to ensure safe tunneling of the tunnel. Thus, the technical effect of improving the accuracy of the prediction of rock burst in the tunnel is achieved, and the technical problem of the accuracy of the prediction of rock burst in the tunnel is solved.

[0114] The embodiments of the present invention are further described below.

[0115] In an embodiment of the present invention, based on microseismic monitoring data, 10 physical information indicators with strong sensitivity to rock burst pressure in steeply inclined and thick coal seam excavation tunnels were selected for data collation and analysis, and multiple rock burst pressure monitoring indicators were calculated using microseismic data during the historical excavation of the mine as source data.

[0116] Optionally, the first vibration index, that is, the b value, can be determined by the following formula:

[0117]

[0118] Where m can be used to represent the total number of microseismic energy levels; N i It can be used to indicate the total number of microseisms at the i-th energy level.

[0119] Optionally, the second vibration index, that is, the A(b) value, can be determined by the following formula:

[0120]

[0121] Wherein, b can be used to represent the b value within the calculation window; M i It can be used to indicate the energy level of microseismic events.

[0122] Optionally, the energy release index, that is, the mine seismic activity S value, can be determined by the following formula:

[0123]

[0124] Among them, N and M i They can be used to indicate the total number of microseismic events and the energy level of occurrence respectively; M can be used to indicate the maximum energy.

[0125] Optionally, the energy level change index, that is, the algorithm complexity AC value, can be determined by the following formula:

[0126]

[0127] Among them, n can be used to represent the number of energy level changes in a specific time period; M can be expressed by M max -M min +1 OK, M max It can be used to indicate the maximum energy level of a microseismic event; M min It can be used to represent the minimum energy level of a microseismic event.

[0128] Optionally, the average energy level index, that is, the equivalent energy level parameter EEM value, can be determined by the following formula:

[0129]

[0130] Among them, m* can be used to represent the normalized energy level of microseismic events; M* can be used to represent the average normalized energy level of m*.

[0131] Optionally, the missing value can be determined by the following formula:

[0132]

[0133] Among them, M and M0 can be used to represent the average energy level and the initial energy level of microseismic events in the area within the time period, respectively; n can represent the total number of microseismic events within the statistical range.

[0134] Optionally, the number of microseismic events can be represented by N; the logarithm of microseismic energy release can be represented by E; the microseismic event rate can be represented by NR; and the logarithm of the microseismic cumulative energy release rate can be represented by ER.

[0135] Optionally, each indicator has dynamic variability and diversity, and the indicators have significant correlation with each other, and are highly sensitive to the predicted impact hazard level. The integration of multiple indicators to comprehensively evaluate the development trend of impact hazards can effectively make up for the shortcomings of limited data monitoring range and limited early warning capability of a single indicator, and effectively improve the early warning capability. According to the magnitude-frequency curve during the excavation of steeply inclined and thick coal seams, the critical value of the impact hazard is divided. The maximum impact hazard level with a prediction time of one day can be selected as the prediction value to timely warn of potential dangers. It has good prediction potential for impact hazards under the geological conditions of steeply inclined and thick coal seams, and can effectively improve the situation of limited monitoring range and short early warning time of existing technologies.

[0136] Optionally, the calculation window and sliding step can be divided according to the number of days, and the energy level given labels can be re-divided. Taking the calculation window as 5 days and the sliding step as 1 day as an example, a data set can be constructed.

[0137] Optionally, to construct a time series data set for each indicator, the window value of each indicator can be obtained and recorded by sliding the calculation window. Assume that the data record calculated by the i-th sliding window is w i , which can be expressed as:

[0138] w i =(id,date,e1,e2,e3,e4,e5,e6,e7,e8,e9,e 10 )

[0139] Among them, id can be used to represent identification information; date can be used to represent time information (date and time); e1 to e 10 It can be used to indicate rock burst monitoring indicators.

[0140] Thus, we can get the time series data divided into n windows:

[0141] Window=(w_0,w_1,,...,w_{n-1})

[0142] Then, a label set of the impact hazard index G corresponding to each precursor pattern sequence set is established, that is, the impact intensity level set, which can be expressed by the following formula:

[0143] G=(t0,t1,t2,......,t D )

[0144] Where D can be used to represent the number of overall statistical days, that is, the number of records; t i Can be used to represent the impact intensity level label of a certain day in the time window series.

[0145] Before inputting the neural network model training, the CFS algorithm can be introduced to select a subset of rock burst monitoring indicator samples. The selection of prediction indicators is the key to affecting the accuracy of rock burst prediction. A reasonable combination of prediction indicators can reduce the model's budget and learning difficulty and improve the model's generalization ability. The 10 indicator sequences calculated are used as feature subsets, and a heuristic method is used to evaluate the excellence of the feature subsets. The model is repeatedly trained and the least important features are eliminated until the specified number of features is reached or the best model performance is achieved. When introducing the CFS algorithm to optimize the feature indicator combination, CFS selects features based on the correlation between features and the correlation between features and target variables. The heuristic equation can be determined by the following formula:

[0146]

[0147] Among them, Meit s , which can be used to represent the heuristic excellence value of the feature subset S containing k features, that is, the excellence value of the rock burst monitoring indicator sample; It can be used to represent the feature-class average correlation, that is, the first average correlation; It can be used to represent the feature-feature average correlation, that is, the second average correlation.

[0148] If Meit s The larger the value, the greater the average correlation between the feature and the category, and the smaller the average redundancy between the features. s The smaller the value, the smaller the average correlation between the feature and the category, and the greater the average redundancy between features.

[0149] Optionally, the CFS algorithm reads the data set by loading data; calculates the excellence values ​​of all features; selects the feature with the maximum excellence value as the feature subset; selects the feature subset step by step through iteration; calculates the merit value of the current feature subset, and when the merit value of the current feature subset is greater than the maximum merit value, updates the maximum merit value and the current feature subset; when the excellence value decreases, removes the feature and traverses the next feature subset; checks whether all feature subsets have been processed, and if so, returns the selected feature subset. The CFS algorithm can be used to obtain the excellence values ​​of different feature subsets, and select a suitable subset of rock burst monitoring indicator samples and the corresponding rock burst intensity grade samples as training sets. Since the rock burst monitoring indicators of different mines have different sensitivities to rock burst prediction, the CFS algorithm can be used to select rock burst monitoring indicators suitable for different mines, thereby improving the generalization and prediction capabilities of the neural network model.

[0150] In order to solve the problems of insufficient extraction of impact precursor information and poor generalization ability of early warning models, the BP neural network model is used to learn the data, and the error and prediction conditions of multiple machine learning models are comprehensively compared. The BP neural network model is the optimal prediction model for steeply inclined and thick coal seam excavation tunnels. Figure 2 is a schematic diagram of a prediction model for rock burst multi-index fusion in a steeply inclined and thick coal seam tunneling tunnel according to an embodiment of the present invention, such as Figure 2 As shown, the following steps are included:

[0151] Step S202, normalizing the input samples according to the main influencing factors of rock burst pressure to obtain a data set for rock burst pressure prediction.

[0152] In this embodiment, the label levels of various physical indicators and single energy of the input samples can be calculated according to the main influencing factors of rock burst pressure, and then the input samples are normalized to generate a data set for rock burst pressure prediction.

[0153] Step S204, performing model training through a neural network topology structure.

[0154] In this embodiment, the data set can be divided into a training set and a test set in a ratio of 8:2, and model training is performed through a neural network topology structure.

[0155] The microseismic energy level-frequency situation during the mine excavation is used to divide the levels and assign labels. Taking the energy level-frequency distribution curve of the historical excavation of +400B3 in Wudong Coal Mine as an example, Figure 3 is a schematic diagram of an energy level-frequency distribution curve according to an embodiment of the present invention, such as Figure 3 As shown in the figure, two energy level offset points with obvious characteristics can be observed from the distribution curve, namely (3.25, 1.75) and (3.65, 0.78). At the offset points, the number of microseismic events exceeding the energy level background value decreases, and the discreteness is enhanced. It can be used as the critical value for judging the impact hazard level of the working face, and the energy critical value of weak impact hazard is set to Er = 103.25 = 1778 J. The energy critical value of medium impact hazard is Ez = 103.65 = 3981 J. According to the actual situation on site, Eq = 104 = 10000 J is set as the energy critical value of strong impact hazard.

[0156] Optionally, for tunneling, the possibility of rock burst danger is very high when an event with energy greater than the fifth power occurs. Therefore, the rock burst danger level can be divided according to the energy value, and the energy critical value E of weak rock burst danger can be set. r =10 4 =10000J. Medium impact energy threshold E z =5×10 4 =50000J, set the critical value of strong impact danger energy E q =10 5 =10000J is the energy critical value for strong impact hazard.

[0157] Step S206: predict the impact type using the model.

[0158] In this embodiment, the impact types may be classified into micro impact, weak impact, medium impact, and strong impact.

[0159] Optionally, standardize each sequence indicator and single microseismic energy to reduce the magnitude difference between the indicators and improve the overall performance of the data. Normalize the target energy value of each sequence indicator set, convert the indicator into a relative value, eliminate the difference between positive and negative indicators, and thus improve the accuracy of model prediction. Data standardization can be performed using the following formula:

[0160] Positive indicators:

[0161] Negative indicators:

[0162] Bidirectional indicators:

[0163] Among them, R i Can be used to represent the i-th value of a sequence; R max Can be used to represent the peak value of all data in the sequence; R min It can be used to represent the valley value of all data in the sequence, that is, the minimum data record; R max Can be used to represent the calculation window R i The maximum value of the sequence, that is, the largest data record; R avg It can be used to calculate the average value of the window indicator series.

[0164] Optionally, the single microseismic energy data is also processed, and the energy value of the highest energy event occurring on the day is used as the level of rock burst monitoring and early warning, and the corresponding single energy data is standardized according to the intervals divided by the microseismic energy level. Table 1 is a single energy standardization table of an embodiment of the present invention. As shown in Table 1, different energy intervals correspond to different energy levels, and the energy standardization results are all different.

[0165] Table 1 Single energy standardization table

[0166]

[0167] In an embodiment of the present invention, the input samples are normalized according to the main influencing factors of rock burst pressure to obtain a data set for rock burst pressure prediction, and the model is trained through a neural network topology structure, so that the model is used to predict the rock burst type. This can achieve the training of an excellent model under the geological conditions of steeply inclined and thick coal seams, reduce data differences, and improve the reliability of rock burst type prediction, thereby achieving the technical effect of improving the accuracy of prediction of rock burst pressure in the roadway and solving the technical problem of low accuracy in prediction of rock burst pressure in the roadway.

[0168] Figure 4 is a flow chart of a neural network model optimized by a genetic algorithm according to an embodiment of the present invention, such as Figure 4 As shown, the following steps are included:

[0169] Step S401, determining the structure of the neural network.

[0170] In this embodiment, data can be imported, divided into training sets and test sets, and the structure of the neural network can be defined, including the number of input features, the number of hidden layer neurons, and the number of output features. The hyperparameters of the genetic algorithm can be defined, including population size, number of iterations, crossover probability, and mutation probability.

[0171] Step S402, encode the neural network weights and thresholds to obtain an initial population.

[0172] In this embodiment, the weight and threshold of the neural network corresponding to each individual are encoded to initialize the population.

[0173] Step S403, obtaining weights and thresholds, assigning the weights and thresholds to the newly created BP network, using training samples to train the network, and using the training samples to test the network and test the error.

[0174] In this embodiment, weights and thresholds can be obtained through decoding, and the weights and thresholds can be assigned to the newly created BP network. The network can be trained using training samples, and the network can be tested using the training samples to test errors.

[0175] Step S404: crossover, mutation, and fitness calculation.

[0176] In this embodiment, in the genetic algorithm, crossover, mutation and fitness calculation are three core operations, which work together in the iterative process of the genetic algorithm. By iteratively performing crossover, mutation and fitness calculation, the genetic algorithm can find a neural network model with better performance in the search space, especially when the search space is very large and has multimodal characteristics, which is more effective than the traditional gradient descent method. In the application of rock burst warning, the genetic algorithm can help the neural network model better adapt to complex geological conditions and improve the accuracy and timeliness of the warning.

[0177] Step S405, genetic algorithm iteration.

[0178] In this embodiment, through crossover, mutation, and calculation of the fitness of each new individual, individuals with better performance may be selected to generate the next generation population, and the above steps are repeated until a predetermined number of iterations is reached.

[0179] Step S406: whether the iteration termination condition is met.

[0180] In this embodiment, it is determined whether the iteration termination condition is satisfied. If so, step S407 is executed; otherwise, step S403 is executed.

[0181] Step S407: decoding to obtain appropriate neural network weights and thresholds.

[0182] In this embodiment, when the iteration termination condition is met, decoding is performed to obtain appropriate neural network weights and thresholds.

[0183] In an embodiment of the present invention, by optimizing the neural network model through a genetic algorithm, it is possible to effectively process a high-dimensional, multi-modal search space, find a rock burst prediction model that can adapt to complex geological conditions, improve the efficiency of mine safety management and prediction, achieve the technical effect of improving the accuracy of the prediction of rock burst in the tunnel, and solve the technical problem of the accuracy of the prediction of rock burst in the tunnel.

[0184] According to an embodiment of the present invention, a device for predicting rock burst in a tunnel is also provided. It should be noted that the device for predicting rock burst in a tunnel can be used to execute the method for predicting rock burst in a tunnel in the above embodiment.

[0185] Figure 5 Schematic diagram of a device for predicting rock burst in a tunnel according to an embodiment of the present invention. Figure 5 As shown, the device 500 for predicting rock burst in the tunnel may include: a monitoring unit 502 , a determination unit 504 and a prediction and determination unit 506 .

[0186] The monitoring unit 502 is used to monitor the target tunnel in the mine and obtain microseismic monitoring data, wherein the microseismic monitoring data is used to indicate the extent of vibration events occurring in the coal seams and rock strata surrounding the target tunnel.

[0187] The determination unit 504 is used to determine the rock burst monitoring index of the target tunnel based on the microseismic monitoring data, wherein the rock burst monitoring index is used to represent the precursor characteristics and evolution law of the rock burst in the target tunnel. The rock burst is formed by the accumulation of internal energy of the coal seam and the internal energy of the rock formation during a vibration event.

[0188] The prediction unit 506 is used to use the prediction model to predict the impact intensity level of the rock burst based on the rock burst monitoring index, wherein the prediction model is obtained by training a neural network model using rock burst monitoring index samples and rock burst intensity level samples, and the impact intensity level is positively correlated with the danger level of the rock burst.

[0189] In the embodiment of the present invention, the target tunnel in the mine is monitored by the monitoring unit 502 to obtain microseismic monitoring data, wherein the microseismic monitoring data is used to indicate the degree of vibration events in the coal seams and rock formations around the target tunnel. The rock burst monitoring index of the target tunnel is determined by the determination unit 504 based on the microseismic monitoring data, wherein the rock burst monitoring index is used to indicate the precursor characteristics and evolution law of the rock burst in the target tunnel, and the rock burst is formed by the accumulation of the internal energy of the coal seam and the internal energy of the rock formation during the vibration event. The rock burst is predicted by the prediction unit 506 using a prediction model based on the rock burst monitoring index, wherein the prediction model is obtained by training a neural network model using rock burst monitoring index samples and rock burst intensity level samples, and the rock burst intensity level is positively correlated with the danger level of the rock burst. Thus, the technical effect of improving the accuracy of the prediction of the rock burst in the tunnel is achieved, and the technical problem of the accuracy of the prediction of the rock burst in the tunnel is solved.

[0190] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute a method for predicting rock burst in a tunnel.

[0191] According to an embodiment of the present invention, a computer-readable storage medium is further provided, the storage medium comprising a stored program, wherein the program executes the method for predicting rock burst in a tunnel in the above embodiment.

[0192] According to an embodiment of the present invention, a processor is further provided, and the processor is used to run a program, wherein when the program is run, the method for predicting rock burst in the tunnel in the above embodiment is executed.

[0193] The embodiment of the present application also provides a computer program product. Optionally, in the present embodiment, the computer program product may include a computer program, and when the computer program is executed by a processor, the method for predicting rock burst in a tunnel of the embodiment of the present application is implemented.

[0194] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0195] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0196] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0197] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0198] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or 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 for 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, Read-Only Memory (ROM), Random Access Memory (RAM), mobile hard disk, magnetic disk or optical disk, etc., various media that can store program codes.

[0199] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for predicting rock burst in a tunnel, characterized in that: include: Monitoring a target tunnel in a mine to obtain microseismic monitoring data, wherein the microseismic monitoring data is used to indicate the extent of vibration events occurring in coal seams and rock strata surrounding the target tunnel; Based on the microseismic monitoring data, determining a rock burst monitoring index of the target tunnel, wherein the rock burst monitoring index is used to represent the precursor characteristics and evolution law of the rock burst in the target tunnel, and the rock burst is formed by the accumulation of internal energy of the coal seam and the internal energy of the rock formation during the occurrence of the vibration event; The prediction model is used to predict the impact intensity level of the rock burst based on the rock burst monitoring index, wherein the prediction model is obtained by training a neural network model using rock burst monitoring index samples and rock burst intensity level samples, and the impact intensity level is positively correlated with the degree of danger of the rock burst.

2. The method according to claim 1, characterized in that The rock burst monitoring index includes at least one of the following: a first vibration index, a second vibration index, an energy release index, an energy level change index, an average energy level index, a missing value, the number of microseismic events, the logarithm of microseismic release energy, the microseismic event rate and the logarithm of the cumulative microseismic energy release rate, wherein the first vibration index is used to indicate the relationship between the energy level and frequency of the vibration event, the second vibration index is used to indicate the target energy microseismic event in the vibration event within the mining range of the working face in the mine, the energy release index is used to indicate the occurrence time, space and intensity of the vibration event; the energy level change index is used to indicate the energy level change characteristics of the vibration event in the time series; the average energy level index is used to indicate the average energy level status of the vibration event.

3. The method according to claim 1, characterized in that Determining the rock burst monitoring index of the target tunnel based on the microseismic monitoring data includes: Based on the microseismic monitoring data, a data record corresponding to the rock burst monitoring index is recorded through a preset calculation window and a slip step, wherein the data record includes time information of the occurrence of the vibration event, the rock burst monitoring index and identification information of the preset calculation window; The method further comprises: merging the data records according to the number of the preset calculation windows to obtain a data record set; The number of records recording the rock burst monitoring index is obtained, and the rock burst intensity levels are merged based on the number of records to obtain an rock burst intensity level set.

4. The method according to claim 3, characterized in that The method further comprises: Standardizing the data record set obtained at the historical moment, and normalizing the standardized data record set; The normalized data record set is used as the rock burst monitoring indicator sample, and the rock burst intensity level set obtained at the historical moment is used as the rock burst intensity level sample; Dividing the rock burst monitoring index samples and the rock burst intensity grade samples according to a preset ratio to obtain a training set and a test set; Inputting the training set into the neural network model, training the neural network model using the training set, inputting the training set into the trained neural network model, and testing the trained neural network model using the test set; In response to the accuracy of testing the test set being higher than an accuracy threshold, the trained neural network model is used as the prediction model.

5. The method according to claim 4, characterized in that The data record set obtained at the historical moment is standardized, including: Determining a maximum data record and a minimum data record from different data records in the data record set; Based on the maximum data record and the minimum data record, standardize different data records in the data record set to obtain the standardized data record set; The method further includes: based on the impact intensity level, performing standardization processing on energy data corresponding to the vibration event to obtain standardized energy data; Based on the impact intensity level samples, the energy data samples corresponding to the vibration events occurring at the historical moments are standardized to obtain the energy data samples after the standardized data.

6. The method according to claim 4, characterized in that Before inputting the training set into the neural network model and using the training set to train the neural network model, the method further includes: Based on the first average correlation between the rock burst monitoring indicator samples and the impact intensity level samples in the training set, and using the second average correlation between the rock burst monitoring indicator samples, determining the excellent value of the rock burst monitoring indicator samples; Determine the rock burst monitoring indicator samples whose excellent values ​​are greater than the excellent threshold as a rock burst monitoring indicator sample subset; The rock burst monitoring index sample subset and the corresponding rock burst intensity level samples are used as a training set to be input into the neural network model.

7. The method according to claim 6, characterized in that Inputting the training set into the neural network model, and using the training set to train the neural network model, comprises: The training set is input into the neural network model, and the neural network model is trained by combining the training set with a genetic algorithm.

8. The method according to claim 7, characterized in that After inputting the training set into the trained neural network model and testing the trained neural network model using the test set, the method further includes: Obtaining a test error of the trained neural network model using the test set, and a fitness corresponding to a hyperparameter of the trained neural network model, wherein the hyperparameters include at least a population size, a number of iterations, a crossover probability, and a mutation probability of the genetic algorithm; In response to the test error exceeding the test error threshold, selecting the hyperparameter whose fitness is higher than the fitness threshold for replication, and performing a crossover operation and a mutation operation on the hyperparameter; In response to the number of iterations of executing the crossover operation and the mutation operation exceeding an iteration number threshold, determining a target hyperparameter; Based on the target hyperparameters, target weights and target thresholds of the neural network model are determined to obtain the prediction model.

9. A device for predicting rock burst in a tunnel, characterized in that: include: A monitoring unit, used to monitor a target tunnel in a mine and obtain microseismic monitoring data, wherein the microseismic monitoring data is used to indicate the extent of vibration events occurring in coal seams and rock strata surrounding the target tunnel; a determination unit, configured to determine a rock burst monitoring index of the target tunnel based on the microseismic monitoring data, wherein the rock burst monitoring index is used to represent the precursor characteristics and evolution law of the rock burst in the target tunnel, and the rock burst is formed by the accumulation of internal energy of the coal seam and internal energy of the rock formation during the occurrence of the vibration event; A prediction unit is used to use a prediction model to predict the impact intensity level of the rock burst based on the rock burst monitoring index, wherein the prediction model is obtained by training a neural network model using rock burst monitoring index samples and rock burst intensity level samples, and the impact intensity level is positively correlated with the degree of danger of the rock burst.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the method for predicting rock burst in a tunnel as claimed in any one of claims 1 to 8.

Images