Displacement-micro-seismic simultaneous-air-space integrated tunnel collapse monitoring and early warning system and method

Through the displacement-microseismic space-attached tunnel collapse monitoring system, the Gaussian process classification model and acoustic and light alarm device are used to solve the monitoring lag and human factors of the tunnel collapse early warning system, and achieve high accuracy and fully automated early warning effect .

CN119091588APending Publication Date: 2024-12-06GUANGXI UNIV
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411188920.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The existing tunnel landslide warning system is delayed in monitoring time and cannot accurately evaluate the overall safety of surrounding rocks. The setting of early warning threshold is affected by human factors, resulting in insufficient warning accuracy.

Method used

The tunnel collapse monitoring and early warning system is adopted with displacement-microseismic simultaneous space-integrated tunnel, and the displacement and microseismic data are obtained in real time through the sensor integrated device, filtering and feature parameter extraction are performed, and early warning judgment is performed using the Gaussian process classification model, and the acousto-optical alarm device provides fast response.

Benefits of technology

It improves the accuracy and reaction time of tunnel collapse warning, reduces casualties and equipment losses, adapts to different geological conditions, and realizes fully automated early warning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119091588A_ABST
    Figure CN119091588A_ABST
Patent Text Reader

Abstract

The invention discloses a displacement-microseismic time-space integrated tunnel collapse monitoring and early warning method, and the method comprises a sensor integration device which is used for carrying out the real-time collection of microseismic waveform data and displacement change data, and transmitting the data to a data processing device. And the data processing device is used for carrying out filtering processing on the original microseismic waveform, calculating the microseismic energy release rate, the fractal dimension, the dominant frequency and the displacement change rate, calculating the group of data by using a trained early warning model, and sending a prediction result to the sound-light alarm device. And the sound-light alarm device is used for responding to the prediction result. According to the tunnel collapse early-warning method, the tunnel collapse early-warning model is established by training the existing characteristic parameters such as the actually measured displacement change rate, the micro-seismic energy release rate, the fractal dimension and the dominant frequency of a certain tunnel through Gaussian process classification, and then real-time automatic early warning is performed by using the early-warning model, so that a guarantee is provided for construction safety.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of displacement monitoring and microseismic monitoring, and specifically relates to a tunnel collapse monitoring and early warning system and method integrating displacement and microseismic space and time, which is suitable for real-time early warning of collapse disasters during tunnel excavation and operation. Technical Background

[0002] With the rapid economic development in the new era, my country's tunnel projects are increasing day by day. At present, my country has become the country with the largest number of tunnel construction and operating mileage in the world. During the construction of tunnels, due to the low strength and poor stability of the tunnel surrounding rock, it is extremely easy to cause engineering geological disasters such as landslides after being disturbed by stress redistribution and construction process. Once a tunnel collapse accident occurs, it often causes huge losses such as casualties of on-site construction personnel, destruction of equipment, and delays in construction period, which has an extremely serious impact on society. Therefore, establishing an efficient tunnel collapse early warning system is of great significance to the safety of tunnel construction and operation.

[0003] At present, the displacement change of surrounding rock is the most intuitive manifestation of tunnel collapse. Therefore, the traditional real-time monitoring of tunnel collapse is mainly based on displacement monitoring, which mainly includes displacement index and change rate index. When the displacement of the tunnel displacement monitoring point is within a reasonable range, it is considered that there is no collapse risk; the change rate index reflects the displacement change of the surrounding rock over a period of time. If the rate value is small, the displacement change is considered normal. However, in the actual monitoring process, the monitoring timing often lags behind the excavation process of the monitoring section, resulting in the monitored displacement being less than the actual displacement of the surrounding rock, that is, there is often a loss of displacement, which reduces the accuracy of the early warning. At the same time, displacement monitoring often only monitors the displacement of a certain monitoring point in real time, and cannot accurately evaluate the overall safety of the surrounding rock. In addition, the warning value setting used by the traditional displacement monitoring method often requires manual setting, and the threshold value usually has certain differences under different geological conditions. Therefore, the accuracy of the collapse warning is greatly interfered by human factors. Microseismic monitoring technology is to monitor the microseismic signals generated by rock fracture in real time, and evaluate the risk of surrounding rock collapse based on the evolution characteristics of the microseismic signals. It has now become an important means of monitoring the stability of hard rock tunnels. However, in the process of landslide, there are often five stages: elastic deformation, plastic deformation, rock fracture, rock loosening and rock collapse. The microseismic monitoring system often detects fewer microseismic signals in the elastic deformation and plastic deformation stages. When more microseismic signals are detected in the rock fracture stage, the landslide may be about to occur, resulting in less time for on-site workers to evacuate the site, thus failing to achieve the goal of reducing casualties. Therefore, combining displacement and microseismic signals to simultaneously monitor the deformation and fracture of the surrounding rock will help improve the accuracy of tunnel landslide warning, greatly increase the response time to disasters, and provide valuable time for the evacuation of tunnel site workers and equipment, compared with using only one monitoring method.

[0004] The tunnel collapse warning threshold is the most important factor affecting the accuracy of the warning. However, when the mechanism of tunnel collapse is unclear and the displacement and microseismic characteristic parameter thresholds of collapse under different geological conditions are unclear, it is extremely difficult to manually set the tunnel collapse warning threshold, and the accuracy of the warning is difficult to meet the engineering needs. Machine learning methods can discover patterns in existing monitoring data, establish an efficient collapse warning model, and continuously improve the prediction and warning performance according to the increase of monitoring data. Therefore, the selection of efficient and feasible machine learning methods for automatic collapse warning can not only realize the automatic warning system without human intervention, but also shorten the collapse risk determination time, which will effectively reduce or avoid the significant losses caused by tunnel collapse. Summary of the invention

[0005] The purpose of the present invention is to solve the problem of inaccurate early warning of tunnel collapse and propose a new early warning method to reduce the economic losses caused by tunnel collapse disasters and reduce casualties of construction personnel. The present invention provides a displacement-microseismic simultaneous and spatial integrated tunnel collapse monitoring and early warning system and a displacement-microseismic simultaneous and spatial integrated tunnel collapse monitoring and early warning method.

[0006] To achieve the above purpose, the present invention adopts the following technical measures:

[0007] In a first aspect, the present invention provides a displacement-microseismic simultaneous and spatial integrated tunnel collapse monitoring and early warning system, comprising a sensor integration device, a data processing device, and an audible and visual alarm device.

[0008] The sensor integrated device is used to acquire displacement data and microseismic waveform data in real time, and send the above data to a data processing device.

[0009] The data processing device is used to filter the microseismic waveform data to extract the microseismic signal generated by rock fracture; then calculate the collected displacement data and microseismic waveform data to obtain the displacement change rate (x x 、x y 、x z 、x 总 ), microseismic energy release rate (MSER), fractal dimension (D t ) and main frequency and other characteristic parameters; it is also used to calculate the above characteristic parameters through a trained prediction model. If the prediction result is a landslide, the prediction result is saved in a data storage device and sent to the sound and light alarm device. Otherwise, the data is saved in the data storage device, and the calculation process of this round ends immediately.

[0010] The sound and light alarm device is used to quickly respond to the tunnel "collapse" early warning results output by the data processing device. The alarm light flashes continuously and the buzzer keeps beeping.

[0011] In a second aspect, the present invention provides a tunnel collapse monitoring and early warning method integrating displacement and microseismic space and time, comprising the following steps:

[0012] Step S1: Drill a hole in the tunnel surrounding rock and fix the sensor integrated device to the bottom of the hole. The integrated device is connected to the data processing device outside the hole through a data transmission cable. After installation, grouting and sealing are performed.

[0013] Step S2: The sensor integration device monitors the tunnel surrounding rock in real time, and transmits the displacement data and microseismic waveform data to the data processing device through the data transmission cable.

[0014] Step S3: The data processing device performs noise filtering on the received microseismic waveform data, removes environmental noise such as blasting and drilling, and extracts relatively pure rock fracture signals.

[0015] Step S4: Extract microseismic characteristic parameters based on the microseismic waveform signal generated by rock fracture, including microseismic energy release rate (MSER), fractal dimension (D t ) and main frequency, and obtain the displacement data when the rock fracture generates the microseismic waveform signal, and calculate the displacement change rate (x x 、x y 、x z 、x 总 ).

[0016] Step S5: embedding the trained displacement-microseismic simultaneous space-time landslide warning model in the data processing device.

[0017] Step S6: The microseismic energy release rate (MSER) ​​and fractal dimension (D t ), main frequency and displacement change rate (x x 、x y 、x z 、x 总 ) and other seven quantitative characteristic parameters are automatically saved in the data storage unit of the data processing device.

[0018] Step S7: The seven quantitative characteristic parameters extracted in step S4 are imported into the trained landslide warning model to determine the landslide risk. If the determination result is landslide, the prediction result is sent to the sound and light alarm device, otherwise the calculation process ends immediately.

[0019] Step S8: If the sound and light alarm device receives the landslide instruction sent by the data processing device, the alarm device responds immediately, the buzzer keeps beeping and the alarm light keeps flashing.

[0020] The step S5 described above adopts the Gaussian process classification (GPC) method to train the early warning model, including the following steps:

[0021] Step 5.1 Select Gaussian process classification as the displacement-microseismic simultaneous and spatial integrated tunnel collapse early warning model, and initialize the displacement-microseismic simultaneous and spatial integrated monitoring tunnel collapse early warning model. Input the existing monitoring data of a tunnel excavation section as sample data, including microseismic energy release rate (MSER), fractal dimension (D t ), main frequency and displacement change rate (x x 、x y 、x z 、x 总 ).

[0022] Step 5.2 The covariance function of the present invention selects the square exponential covariance function.

[0023] Step 5.3: Train the GPC model. Divide the sample data into training and test sets. Based on the 10-fold cross-validation method, use any 9 subsets of the sample as training sets and the remaining 1 subset as validation sets. Assign initial hyperparameters and learn the training samples through the squared exponential covariance function.

[0024] Step 5.4: Determine the posterior model. By maximizing the log-likelihood function of the training sample potential function f, the optimal hyperparameters of the covariance function are obtained, and then the posterior model is determined.

[0025] Step 5.5 trains the early warning model through the training set, obtains the predicted result data of the early warning model by inputting the verification set, compares the predicted result data with the actual result data in the verification set, and calculates the accuracy of the predicted result.

[0026] Step 5.6 compares the accuracy of the above prediction results. If the expected accuracy is met, a model is generated; if not, the training set and test set are randomly re-divided and step 5.3 is repeated.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] 1. The present invention considers displacement and microseismic signals to jointly evaluate the health status of the tunnel's surrounding rock. Compared with the traditional displacement monitoring method, this method can evaluate the health status of the entire surrounding rock by monitoring the microseismic signals generated by rock fracture, avoiding the traditional displacement monitoring method that can only monitor the displacement changes of a certain monitoring point and generate a large number of monitoring blind areas. Compared with the microseismic monitoring method, this method can judge the collapse risk according to the displacement changes during the deformation stage of the surrounding rock collapse process, thereby greatly increasing the emergency response time of the collapse and buying more valuable time for the evacuation of on-site staff and equipment.

[0029] 2. The simultaneous spatial and temporal acquisition of displacement and microseismic data effectively improves the accuracy of landslide warning. Compared with the traditional non-simultaneous spatial and temporal acquisition method, this method can automatically provide data from two monitoring sensors at any time and any spatial position, and there is no time difference between the two sensor data, which avoids the problem of inaccurate landslide warning caused by inconsistent timing of the data collected by the displacement and microseismic sensors, and effectively improves the accuracy of tunnel landslide warning.

[0030] 3. The present invention realizes quantitative early warning of tunnel collapse and has good adaptability to tunnels with different geological conditions. The method selects microseismic energy release rate (MSER), fractal dimension (D t ), main frequency and displacement change rate (x x 、x y 、x z 、x 总 ) and seven other quantitative characteristic parameters. The displacement-microseismic simultaneous space-time landslide warning model was constructed through the existing field monitoring data and the Gaussian process classification model training. The model was used to study the relationship between the occurrence law of tunnel landslides and related parameters, so as to provide real-time and efficient early warning of landslide disasters. In addition, this method obtains the early warning threshold of each parameter through machine learning methods, which solves the defects of manual threshold setting, which is time-consuming and labor-intensive and cannot adapt to different excavation conditions.

[0031] 4. The present invention can realize fully automated early warning of landslides. The method has a high level of automation and intelligence, and the monitoring process does not require the use of transmission equipment such as optical fibers to exchange information with the outside of the tunnel, which greatly reduces the cost of use and also reduces the risk of early warning failure due to transmission equipment failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a structural schematic diagram of the system of the present invention;

[0033] Figure 2 It is a schematic diagram of the structure of the sensor integration device in the present invention;

[0034] Figure 3 This is a schematic diagram of the installation of the sensor integration device in the present invention;

[0035] Figure 4 It is a schematic diagram of a data processing device module of the present invention;

[0036] Figure 5 It is a structural schematic diagram of the sound and light alarm device of the present invention;

[0037] Figure 6 It is a flow chart of the real-time monitoring and early warning method of tunnel collapse integrating displacement and microseismicity in both space and time;

[0038] Figure 7A flow chart for training the displacement-microseismic simultaneous and spatial integrated monitoring tunnel collapse early warning model embedded in the data processing device;

[0039] In the figure: 100 - displacement sensor and microseismic sensor integrated device; 200 - data processing device; 300 - sound and light alarm device. DETAILED DESCRIPTION

[0040] In order to facilitate ordinary technicians in the field to understand and implement the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and examples. It should be understood that the implementation examples described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0041] like Figure 1 As shown, the displacement-microseismic simultaneous and spatial integrated tunnel collapse monitoring and early warning system includes a sensor integration device 100, a data processing device 200, and an acoustic and visual alarm device 300. The sensor integration device 100 is connected to the data processing device 200, and the data processing device 200 is connected to the acoustic and visual alarm device 300.

[0042] The displacement sensor and microseismic sensor integrated device 100 includes a high-precision three-dimensional displacement meter 101 and a microseismic acceleration sensor 102. The displacement data is obtained by the high-precision displacement meter 101, and the microseismic waveform data is obtained by the microseismic acceleration sensor 102. The displacement sensor and microseismic sensor integrated device 100 sends the displacement data and the microseismic waveform data to the data processing device 200.

[0043] The data processing device 200 performs noise filtering on the collected microseismic waveform data to obtain a relatively pure rock fracture signal, and then extracts microseismic characteristic parameters, including microseismic energy release rate (MSER), fractal dimension (D t ) and main frequency characteristic parameters, and calculate the displacement change rate (x) based on the displacement meter data at the moment when the microseismic waveform signal is generated due to rock fracture. x 、x y 、x z 、x 总 ); and calculate the collapse possibility through the soft rock early warning model of displacement-microseismic simultaneous space and time integration of the above seven quantitative characteristic parameters, and store the above data in the data storage unit.

[0044] If the above calculation result is landslide, the data processing device 200 sends the instruction to the sound and light alarm device 300 through the data transmission cable, otherwise the warning calculation ends immediately.

[0045] After receiving the collapse instruction from the data processing device 200, the sound and light alarm device 300 immediately responds, the alarm light flashes continuously, and the buzzer keeps beeping.

[0046] like Figure 2 As shown, the sensor integration device 100 includes a high-precision three-way displacement sensor 101, an acceleration-type microseismic sensor 102, a bolt connector 103, a microseismic sensor fixture 104, a data transmission cable steel pipe sleeve 105, a data transmission cable 106 and a waveguide 107. Among them, the high-precision three-way displacement sensor 101 is used to collect displacement data, and the acceleration-type microseismic sensor 102 is used to collect microseismic waveform data. In order to improve the effectiveness of displacement data for landslide warning, the above sensors are generally installed 20 cm away from the surrounding rock borehole. The bolt connector 103 is used to connect the displacement sensor and the microseismic sensor into one, and has a built-in transmission cable, which is used to transmit the data monitored by the high-precision three-way displacement sensor to the data transmission cable 106. The microseismic sensor fixture 104 is used to make the microseismic sensor in close contact with the surrounding rock, which is conducive to the microseismic sensor receiving the microseismic signal generated by the surrounding rock fracture. The data transmission cable steel pipe sleeve 105 is used to install the microseismic sensor fixture 104, and is also used to place the data transmission cable 106 to prevent damage to the data transmission cable 106 during the grouting process. The data transmission cable 106 is used to transmit the collected data to the data processing device 200 outside the borehole. The waveguide 107 is used to transmit the microseismic signal generated by the deep rock mass fracture to the microseismic sensor, thereby reducing the attenuation of the microseismic signal during the transmission process of the broken rock mass. Generally, the waveguide 107 needs to be implanted in the surrounding rock for more than 200 cm.

[0047] like Figure 3 As shown, a method for installing a permanent sensor integrated device 100 includes the following steps: the screw 103-2 of the bolt connector 103 is connected to the high-precision three-way displacement sensor 101, and the screw 103-1 is connected to the acceleration-type microseismic sensor nut 102-1. It is worth noting that a transmission cable is built into the outer side of the screw, and after the bolt connection is tight, it can be used to transmit the displacement data collected by the high-precision three-way displacement sensor 101 to the data transmission cable 105. After the sensor integrated device 100 is pushed to the bottom of the surrounding rock borehole and is in close contact with the waveguide 107, since the data transmission cable steel pipe sleeve 105 and the acceleration-type microseismic sensor 102 are nested through the inner hexagon, the data transmission cable steel pipe sleeve 105 is rotated counterclockwise, so that the screw 103-1 and the nut 102-1 are more tightly connected. At the same time, due to the thrust, the microseismic sensor fixing device 104 opens to the surrounding of the borehole, so that the sensor integrated device 100 is tightly fixed in the surrounding rock borehole. After the installation is completed, the holes are sealed by grouting, and after the slurry solidifies, the high-precision three-dimensional displacement sensor 101 is reset and zeroed.

[0048] like Figure 4As shown, the data processing device 200 includes a power supply module 201, a filter module 202, a characteristic parameter calculation module 203, a data storage module 204, an early warning model calculation module 205, an input interface 206 and an output interface 207. The power supply module uses a large-capacity battery to supply power to the sensor integration device 100 and the data processing device 200; the filter module 202 is used to remove the environmental noise signal of non-rock fracture in the microseismic signal and obtain a relatively pure rock fracture signal; the characteristic parameter calculation module 203 is used to extract microseismic characteristic parameters from the microseismic waveform, including microseismic energy release rate (MSER), fractal dimension (D t ), the main frequency characteristic parameters, and calculate the displacement change rate (x) at the time when the rock fracture signal is generated based on the displacement data x 、x y 、x z 、x 总 ); the data storage module 204 is used to store all data in the early warning calculation process. It uses large-capacity mobile storage devices such as mobile hard disks to facilitate monitoring personnel to retrieve data at any time for further research; the early warning model calculation module 205 embeds the trained GPC early warning model, inputs displacement and microseismic characteristic parameter data, and predicts the landslide risk; the input interface 206 is used to receive real-time data collected by the sensor integration device 100; the output interface 207 is used to send the landslide command to the sound and light alarm device 300.

[0049] like Figure 5 As shown, the sound and light alarm device 300 includes a buzzer 301, an alarm light 302, a data input interface 303, a fixing screw 304, a battery 305, a control circuit board 306, a recovery button 307 and an alarm button 308. Among them, the buzzer 301 is used to emit a continuous buzzing sound when receiving a landslide warning instruction; the alarm light 302 is used to emit a continuous red flashing light when receiving a landslide warning instruction; the data input interface 303 is used to receive a landslide warning instruction sent by the data processing device 200; the fixing screw 304 is used to fix the alarm device on the surrounding rock of the tunnel; the battery 305 is used to power the sound and light alarm device 300; the control circuit board 306 is used to control the buzzer 301 and the alarm light 302 to work after receiving the "landslide" instruction; the recovery button 307 is used to manually turn off the alarm light and the buzzer after the warning ends; and the alarm button 308 is used to manually turn on the alarm system when other dangerous situations occur.

[0050] like Figure 6 As shown, a real-time monitoring and early warning method for tunnel collapse integrating displacement and microseismic simultaneous space and time includes the following steps:

[0051] Step S1: Drill a hole in the tunnel surrounding rock, fix the sensor integrated device 100 to the bottom of the hole, the integrated device is connected to the data processing device 200 outside the hole through a data transmission cable, and grouting and sealing are performed after installation.

[0052] Step S2: The sensor integration device 100 monitors the tunnel surrounding rock in real time, and transmits the displacement data and microseismic waveform data to the data processing device 200 via the data transmission cable.

[0053] Step S3: The data processing device 200 performs noise filtering on the received microseismic waveform data to remove environmental noise such as blasting and drilling, and extract a relatively pure rock fracture signal.

[0054] Step S4: Extract microseismic characteristic parameters based on the microseismic waveform signal generated by rock fracture, including microseismic energy release rate (MSER), fractal dimension (D t ) and main frequency, and obtain the displacement data when the rock fracture generates the microseismic waveform signal, and calculate the displacement change rate (x x 、x y 、x z 、x 总 ).

[0055] Step S5: embedding the trained displacement-microseismic simultaneous space-time landslide warning model in the data processing device 200).

[0056] Step S6: The microseismic energy release rate (MSER) ​​and fractal dimension (D t ), main frequency and displacement change rate (x x 、x y 、x z 、x 总 ) and other seven quantitative characteristic parameters are automatically saved in the data storage unit of the data processing device 200.

[0057] Step S7: The seven quantitative characteristic parameters extracted in step S4 are introduced into the trained landslide warning model to determine the landslide risk. If the determination result is landslide, the prediction result is sent to the sound and light alarm device 300, otherwise the calculation process ends immediately.

[0058] Step S8: If the sound and light alarm device 300 receives the landslide instruction sent by the data processing device 200, the alarm device responds immediately, the buzzer keeps beeping and the alarm light keeps flashing.

[0059] For example, in step S6, the microseismic energy release rate (MSER), the fractal dimension (D t ), main frequency and displacement change rate (x x 、xy 、x z 、x 总 The calculation method of seven quantitative characteristic parameters such as ) is as follows:

[0060] (1) Calculation method of microseismic energy release rate (MSER)

[0061] The microseismic energy release rate (MSER) ​​characterizes the rate at which relative energy is released to the outside world when the rock breaks, and can be used to describe the severity of rock failure.

[0062]

[0063] Where: A(t) represents the microseismic amplitude at time t.

[0064] (2) Fractal dimension of microseismic amplitude (D t )

[0065] According to the theory of fractal geometry, the calculation formula of the correlation integral C(t) of the time distribution of the microseismic amplitude signal generated by the rock during the fracture process is:

[0066]

[0067] Where: T is the total duration of the calculation period; t is the time interval between microseismic amplitude signals during the T time process; M(t) is the logarithm of the sum of the amplitude numbers within the t time period; M is the sum of the amplitude numbers within the T time range.

[0068] Temporal fractal dimension D of microseismic amplitude t The calculation formula is:

[0069]

[0070] (3) Calculation method of microseismic dominant frequency

[0071] The microseismic main frequency is transformed from the time domain signal to the frequency domain using the Fast Fourier Transform (FFT) method. The calculation is relatively complicated. The FFT toolbox in MATLAB can be used to implement the fast Fourier transform of the microseismic signal. The specific calculation method is not introduced in detail in this patent.

[0072] (4) Displacement change rate (x x 、x y 、x z 、x 总 )

[0073]

[0074]

[0075] Where xx (t), x y (t) and x z (t) is the displacement value of the high-precision three-dimensional displacement meter in the x, y and z directions at time t.

[0076] The real-time monitoring and early warning system for tunnel collapse with simultaneous and spatial integrated monitoring of displacement and microseismicity can provide early warning of surrounding rock collapse in the tunnel by analyzing the existing data at the project site using the support vector machine method, thus providing safety protection for on-site workers. In addition, by extracting the real-time monitoring data from the data storage device, the existing model can be further trained to improve the accuracy of the early warning model. Of course, the Gaussian process classification model trained in this project can also be applied to other similar surrounding rock collapse early warning projects.

[0077] Gaussian process (GP) has a strict statistical learning theory foundation, has good adaptability to complex problems such as high dimension, small sample, nonlinearity, etc., and has the advantage of adaptive parameter acquisition. There is no problem of establishing topological structure and difficulty in selecting parameters. At the same time, it can obtain output with probabilistic significance. It can well overcome the problems existing in methods such as neural networks and support vector machines. Gaussian binary classification method (GPC) is a classification method based on the GP machine learning principle.

[0078] In the Gaussian process binary classification problem, the training sample set is D = {(x i ,y i )|i=1,…,m} T , where x is the input vector and y is the output value, which is usually -1 or 1. For a certain x, p(y|x) is distributed as Bernoulli distribution, and the probability of y=1 is p(y=1|x)=Φ(f(x)), where f(x) is called the potential function and Φ(x) is called the response function. The commonly used response function is the probability density function, as shown in formula (8).

[0079]

[0080] In the Gaussian process learning process, it is assumed that the prior distribution of the target function p(f|X), also denoted as p(f), belongs to the Gaussian distribution:

[0081]

[0082] Among them, K is the n×n order covariance function matrix, K ij = k(x i ,x j ), k(x,y) is the covariance function between the data. The covariance function must satisfy the requirement of generating a non-negative positive definite covariance matrix for any set of points.

[0083] The method embeds the trained GPC displacement-microseismic simultaneous and spatial integrated real-time landslide warning model into the data processing device 200. When the rock fracture generates a microseismic waveform signal, the data processing device 200 obtains the microseismic energy release rate (MSER), fractal dimension (D t ), main frequency and displacement change rate (x x 、x y 、x z 、x 总 ) and other characteristic parameters, and import the above indicators into the early warning model for calculation to predict whether the tunnel is in a safe state.

[0084] The training process of the displacement-microseismic simultaneous and spatial integrated monitoring tunnel collapse early warning model in step S5 is as follows: Figure 7 As shown, the following steps are included:

[0085] Step 5.1 Select Gaussian process classification as the displacement-microseismic simultaneous and spatial integrated tunnel collapse early warning model, and initialize the displacement-microseismic simultaneous and spatial integrated monitoring tunnel collapse early warning model. Input the existing monitoring data of a tunnel excavation section as sample data, as shown in Table 1, including microseismic energy release rate (MSER), fractal dimension (D t ), main frequency and displacement change rate (x x 、x y 、x z 、x 总 ).

[0086] Table 1 Existing monitoring data of a tunnel excavation section

[0087]

[0088] Step 5.2 The covariance function of the present invention selects the square exponential covariance function, and its expression is as follows:

[0089]

[0090] Step 5.3: Train the GPC model. Divide the sample data into training and test sets. Based on the 10-fold cross-validation method, use any 9 subsets of the sample as training sets and the remaining 1 subset as validation sets. Assign initial hyperparameters and learn the training samples through the squared exponential covariance function.

[0091] Step 5.4: Determine the posterior model. By maximizing the log-likelihood function of the training sample potential function f, the optimal hyperparameters of the covariance function are obtained, and then the posterior model is determined.

[0092] Step 5.5 trains the early warning model through the training set, obtains the predicted result data of the early warning model by inputting the verification set, compares the predicted result data with the actual result data in the verification set, and calculates the accuracy of the predicted result.

[0093] Step 5.6 compares the accuracy of the above prediction results. If the expected accuracy is met, a model is generated; if not, the training set and test set are randomly re-divided and step 5.3 is repeated.

[0094] The present invention combines displacement and microseismic signals to warn tunnel collapse in real time in space and time, which can not only use multiple parameters to warn of collapse disasters and improve the accuracy of early warning, but also use the Gaussian process binary classification method to establish an automatic early warning model, saving collapse reaction time and greatly reducing or avoiding equipment losses and personal injuries of on-site workers. In addition, the real-time monitoring and early warning model of tunnel collapse trained by displacement-microseismic integrated monitoring in space and time in the present invention can also provide reference for tunnel collapse disaster early warning under similar geological conditions.

[0095] The specific implementation methods described above are merely illustrative of the spirit of the present invention, but the protection scope of the present invention is not limited thereto. Technicians in the technical field to which the present invention belongs may make various modifications or supplements to the described specific embodiments or replace them in a similar manner, but they will not deviate from the spirit of the present invention or exceed the protection scope of the attached claims.

Claims

1. A tunnel collapse monitoring and early warning system integrating displacement and microseismic space and time, characterized by: It comprises a sensor integration device (100), a data processing device (200) and an audible and visual alarm device (300); A sensor integrated device (100) is used to collect displacement data and microseismic waveform data, and send the data to a data processing device (200); The data processing device (200) is used to process the displacement data and the microseismic waveform data to obtain the displacement change rate (x x 、x y 、x z 、x 总 ) and microseismic characteristic parameter data, including microseismic energy release rate MSER, fractal dimension D t and main frequency, and is also used to predict the microseismic characteristic parameter data and displacement data through the trained early warning model. If the prediction result is a landslide, all the data are saved in a data storage device and sent to an audible and visual alarm device (300). Otherwise, after the data is saved in the data storage device, the calculation process is immediately terminated and the calculation of the next set of data is started; The sound and light alarm device (300) is used to respond to the landslide prediction result output by the data processing device.

2. The displacement-microseismic simultaneous and spatial integrated tunnel collapse monitoring and early warning system according to claim 1 is characterized in that: The sensor integration device (100) comprises a high-precision three-dimensional displacement sensor (101), an acceleration-type microseismic sensor (102), a bolt connector (103), a microseismic sensor fixing device (104), a data transmission cable steel pipe sleeve (105), and a data transmission cable (106); The high-precision three-dimensional displacement sensor (101) is used to collect displacement data; the acceleration-type microseismic sensor (102) is used to collect microseismic waveform data; the bolt connector (103) is used to connect the displacement sensor and the microseismic sensor into one, and has a built-in transmission cable, which is used to transmit the data monitored by the high-precision three-dimensional displacement sensor to the data transmission cable (106); the microseismic sensor fixing device (104) is used to make the microseismic sensor closely contact with the surrounding rock, which is conducive to the microseismic sensor receiving the microseismic signal generated by the surrounding rock fracture; the data transmission cable steel pipe sleeve (105) is used to install the microseismic sensor fixing device (104) and is also used to place the data transmission cable (106) to prevent the data transmission cable (106) from being damaged during the grouting process; the data transmission cable (106) is used to transmit the collected data to the data processing device (200) outside the borehole; and the waveguide (107) is used to transmit the microseismic signal generated by the deep rock mass fracture.

3. The displacement-microseismic simultaneous and spatial integrated tunnel collapse monitoring and early warning system according to claim 1 is characterized in that: The sensor integrated device (100) installation device and method include the following steps: the screw rod (103-2) of the bolt connector (103) is connected to the high-precision three-dimensional displacement sensor (101), and the screw rod (103-1) is connected to the acceleration-type microseismic sensor nut (102-1); a transmission cable is built into the outer side of the screw rod (103-1) to transmit the displacement data collected by the high-precision three-dimensional displacement sensor (101) to the data transmission cable (105); the sensor integrated device (100) is pushed to the surrounding rock borehole and the waveguide (101). 7) After close contact, the data transmission cable steel pipe sleeve (105) and the acceleration type microseismic sensor (102) are nested through the inner hexagon, and the data transmission cable steel pipe sleeve (105) is rotated counterclockwise so that the screw (103-1) and the nut (102-1) are tightly connected, and the microseismic sensor fixing device (104) is opened to the surrounding of the borehole, so that the sensor integrated device (100) is tightly fixed in the surrounding rock borehole; after the installation is completed, the hole is grouting and sealing, and the high-precision three-dimensional displacement sensor (101) is reset and zeroed after the grout solidifies.

4. The displacement-microseismic simultaneous and spatial integrated tunnel collapse monitoring and early warning system according to claim 1 is characterized in that: The data processing device (200) comprises a power supply module (201), a filtering module (202), a characteristic parameter calculation module (203), a data storage module (204), an early warning model calculation module (205), an input interface (206) and an output interface (207); the power supply module (201) is used to supply power to the device; the filtering module (202) is used to remove environmental noise signals other than rock fractures in microseismic signals; the characteristic parameter calculation module (203) is used to calculate characteristic parameters of displacement and microseismic, including microseismic energy release rate MSER, fractal dimension D t , main frequency and displacement change rate (x x 、x y 、x z 、x 总 ); The data storage module (204) is used to store all data in the early warning calculation process, which uses a large-capacity storage device; The early warning model calculation module (205) predicts the landslide risk by embedding the trained GPC early warning model and inputting the three-dimensional displacement change rate and microseismic characteristic parameter data.

5. A tunnel collapse monitoring and early warning method integrating displacement and microseismic space and time, characterized in that: The tunnel collapse monitoring and early warning system with simultaneous and spatial integration of displacement and microseismicity as described in any one of claims 1 to 4 is adopted; the method described comprises the following steps: Step S1: drilling a hole in the surrounding rock of the tunnel, fixing the sensor integrated device (100) to the bottom of the hole, connecting the sensor integrated device (100) to the data processing device (200) outside the hole via a data transmission cable, and performing grouting and sealing after installation; Step S2: the sensor integration device (100) monitors the tunnel surrounding rock in real time, and transmits the displacement data and microseismic waveform data to the data processing device (200) via the data transmission cable; Step S3: the data processing device (200) performs noise filtering on the received microseismic waveform data, removes blasting and drilling environmental noise, and extracts a relatively pure rock fracture signal; Step S4: Extract microseismic characteristic parameters based on the microseismic waveform signal generated by rock fracture, including microseismic energy release rate MSER, fractal dimension D t and main frequency characteristic parameters, and simultaneously obtain the displacement data when the rock fracture generates the microseismic waveform signal, and calculate the displacement change rate (x x 、x y 、x z 、x 总 ); Step S5: embedding the trained displacement-microseismic simultaneous space-time landslide warning model in the data processing device (200); Step S6: The microseismic energy release rate MSER and fractal dimension D calculated in step S4 are t , main frequency and displacement change rate (x x 、x y 、x z 、x 总 ) seven quantitative characteristic parameters are automatically stored in a data storage unit of a data processing device (200); Step S7: the seven quantitative characteristic parameters extracted in step S4 are imported into the trained landslide warning model to determine the landslide risk; if the determination result is landslide, the prediction result is sent to the sound and light alarm device (300); otherwise, the calculation process ends immediately; Step S8: If the sound and light alarm device (300) receives the landslide instruction sent by the data processing device (200), the alarm device responds immediately, the buzzer continues to sound and the alarm light continues to flash.

6. The tunnel collapse monitoring and early warning method with simultaneous and spatial integration of displacement and microseismicity according to claim 5 is characterized in that: Select the microseismic energy release rate MSER and fractal dimension D t , main frequency and displacement change rate x x 、x y 、x z 、x 总 ) Seven quantitative indicators are used as evaluation indicators for determining tunnel collapse risk warning.

7. The displacement-microseismic simultaneous and spatial integrated tunnel collapse monitoring and early warning method according to claim 5 is characterized in that: In step S5, the early warning model is trained by using the Gaussian process classification (GPC) method, which includes the following steps: Step 5.1 Select Gaussian process classification as the displacement-microseismic simultaneous and spatial integrated tunnel collapse early warning model, initialize the displacement-microseismic simultaneous and spatial integrated monitoring tunnel collapse early warning model; input the existing monitoring data of a tunnel excavation section as sample data, including microseismic energy release rate MSER, fractal dimension D t , main frequency and displacement change rate (x x 、x y 、x z 、x 总 ); Step 5.2 The covariance function of the present invention selects a square exponential covariance function; Step 5.3: Train the GPC model; divide the sample data into training set and test set, and based on the 10-fold cross-validation method, use any 9 subsets of the sample as training sets and the remaining 1 subset as validation set; assign initial hyperparameters and learn the training samples through the square exponential covariance function; Step 5.4: Determine the posterior model; by maximizing the log-likelihood function of the potential function f of the training sample, the optimal hyperparameters of the covariance function are obtained, and then the posterior model is determined; Step 5.5: Train the early warning model through the training set, obtain the prediction result data of the early warning model by inputting the verification set, compare the prediction result data with the actual result data in the verification set, and calculate the accuracy of the prediction result; Step 5.6 compares the accuracy of the above prediction results. If the expected accuracy is met, a model is generated; if not, the training set and test set are randomly re-divided and step 5.3 is repeated.

Citation Information

Cited By

  • Intelligent early warning method, system and equipment for karst tunnel collapse based on micro-seismic multi-precursor characteristics

    CN119811050A

  • Tunnel surrounding rock stability automatic monitoring and early warning method and system

    CN120403781A