GRU neural network-based shield wall back grouting condition monitoring method
Through the monitoring method based on GRU neural network, the acoustic signals during grouting are monitored by using the test device and waveguide rods, and the problem of difficult to evaluate the grouting effect in shield construction is solved, and the accurate monitoring of the grouting condition behind the shield wall is achieved and improve construction safety.
Patent Information
- Application Number
- CN202411130613.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-16
- Publication Date
- 2024-12-03
AI Technical Summary
During the shield construction process, the concealment of grouting construction makes it difficult to evaluate its effect. If the construction is poor, it may cause air leakage or instability of the excavation surface, and even lead to safety accidents.
The monitoring method based on GRU neural network is adopted to monitor the acoustic signals during grouting through the test device and waveguide rod, and the GRU model is combined for training and real-time monitoring, and the grouting effect dynamic diagram is derived in real time to determine the grouting pressure and quantity.
Accurate monitoring of the grouting condition behind the shield wall is achieved, construction safety and efficiency are improved, and the reliability of grouting effect is ensured.
Smart Images

Figure CN119066964A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of shield slag improvement, and in particular to a monitoring method for shield wall back grouting condition based on a GRU neural network. Background Art
[0002] The underground railway has the characteristics of large passenger capacity and fast transportation. It is an effective means of transportation to alleviate urban traffic pressure, so it has been vigorously developed in recent years. The shield method is often used as a construction method in the construction of underground railways. During the shield excavation process, it is inevitable to inspect and replace the shield cutter head, and close construction will cause disturbances to existing structures. The grouting material of the mud effect can well solve the problems of shield pressure cutter change and reduce construction disturbance. However, due to its concealment, it is difficult to evaluate the effectiveness of the mud effect grouting in actual construction. Grouting construction is affected by many aspects. If the construction effect is poor, it will cause air leakage during the pressure cutter change process, which may cause the excavation face to become unstable, and even cause major safety accidents. In order to solve such problems, a monitoring system that can monitor the grouting status behind the shield wall is designed to accurately grasp the grouting effect, which is of great significance to ensure the safety of shield tunnel construction. Based on this, the present invention proposes a monitoring method for the grouting status behind the shield wall based on a GRU neural network. Summary of the invention
[0003] In order to solve the above-mentioned deficiencies in the prior art, the present invention provides a method for monitoring the grouting condition of the shield wall based on the GRU neural network during the shield construction process.
[0004] In order to achieve the above technical purpose, the technical solution adopted by the present invention is:
[0005] A method for monitoring the back grouting condition of a shield wall based on a GRU neural network comprises the following steps:
[0006] S1: Connection of the test device to conduct air tightness test of pipeline and grouting model tube; the test device includes a pressure system, a grouting model tube and a data collection and processing system, the pressure system is connected to the grouting model tube through an air pipeline, the data collection and processing system includes an acoustic sensor, a waveguide rod and a processor, the acoustic sensor is connected to the waveguide rod, and the acoustic sensor is connected to the processor for communication;
[0007] S2: Prepare the mud-reducing slurry and ethylene glycol aqueous solution. After the preparation is completed, fill the grouting model cylinder, then fill the ethylene glycol aqueous solution, sand, and mud-reducing slurry in sequence, and bury the waveguide rod of the data collection and processing system in the sand;
[0008] S3: Pressurize the inside of the grouting model cylinder through the pressure system and observe the seepage process until the mud-reducing slurry fills the sand layer;
[0009] S4: The signal generated by the waveguide rod is monitored by an acoustic sensor, and the monitoring signal is transmitted to the processor for extraction and processing, and then substituted into the GRU model for training;
[0010] S5: During on-site construction, waveguide rods are installed in the soil layer for real-time monitoring. After real-time monitoring data is processed, it is input into the trained GRU model to export the dynamic graph of grouting effect in real time;
[0011] S6: According to the dynamic diagram of grouting effect, comprehensively consider other engineering factors and determine the grouting pressure and grouting volume.
[0012] Preferably, the pressure system includes a servo-capable air compressor, and a pressure regulating valve is provided on the air pipeline.
[0013] Furthermore, the grouting model cylinder comprises a cylinder body made of transparent organic glass, the upper side of the cylinder body is connected to an upper flange, the lower side of the cylinder body is connected to a lower flange, the connection gaps between the upper flange, the lower flange and the cylinder body are filled with rubber sealing strips, and a fixed water-permeable baffle is radially arranged inside the cylinder body; an inlet valve is arranged at the upper part of the cylinder body, and an outlet valve is arranged below the fixed water-permeable baffle at the lower part of the cylinder body;
[0014] The acoustic sensor collects the bending waveguide signals generated by the friction and collision between sand particles and the waveguide rod during the grouting process in the grouting model tube and transmits them to the processor, which processes, converts, analyzes and saves the collected bending waveguide signals.
[0015] Preferably, the ethylene glycol concentration in the ethylene glycol aqueous solution is 50%.
[0016] Further, in step S2, the method of filling the grouting mold cylinder is:
[0017] S21, open the upper flange, and pour ethylene glycol aqueous solution into the grouting model cylinder to fix the permeable baffle position;
[0018] S22, filling the sand and soil collected from the site layer by layer, compacting each layer of sand and soil and scraping the surface;
[0019] S23. Pour a sufficient amount of grouting slurry above the sand layer, close the upper flange, and complete the filling of the grouting model tube.
[0020] Furthermore, in step S3, the step of pressurizing the grouting model tube is as follows: closing the inlet valve and outlet valve of the grouting model tube, adjusting the pressure regulating valve to the target test pressure, and then opening the inlet valve and outlet valve to charge the grouting model tube with pressure from the pressure system.
[0021] Furthermore, in step S4, the extraction and processing method of the monitoring signal is as follows: the waveguide signal captured by the acoustic sensor is passed through a 60dB amplifier to improve the signal-to-noise ratio, filter the background noise, and then convert the electrical signal into a digital signal, after which the data is processed by a processor and signal analysis software.
[0022] Furthermore, in step S5, the method for installing waveguide rods in the soil layer is as follows: waveguide rod installation holes are reserved on the shield segments, and after the shield machine is excavated, the waveguide rods in each ring are installed immediately after each ring of segments is installed, and 4 waveguide rods are installed in each ring in a centrally symmetrical distribution. Grouting begins after 4 waveguide rods are installed in each ring of segments.
[0023] Furthermore, the other engineering factors include the permeability of the stratum, the water and soil pressure of the surrounding strata, the excavation route, the slurry type, the segment strength and the shield type.
[0024] Furthermore, it also includes step S7: after the grouting is completed, the grouting range is detected, and then the GRU model is imported for correction learning; the grouting range is detected by using the geophysical method, and the medium within the 1m thickness range behind the shield segment wall is detected by the ground penetrating radar non-destructive testing technology to identify the diffusion range information of the grouting body, and the detected data is then substituted into the GRU model training.
[0025] The beneficial effects of the present invention are:
[0026] The test device in the present invention is used to simulate the process of slurry penetrating the formation. The test results are used as the basis for judging the actual grouting effect. The GRU (Gated Circular Unit) model is applied to the monitoring of the grouting condition behind the shield construction wall. The waveguide monitoring technology is used to monitor the grouting condition. The waveguide signal generated by the friction and collision between sand particles and waveguide rods is cleverly linked to the grouting condition. Due to the characteristics of small waveguide attenuation and long propagation distance, it is convenient for data collection and processing and subsequent analysis to obtain more accurate monitoring results. The GRU model has good long-term memory performance and can discard information that may mislead the current output. In addition, it has the characteristics of few parameters and fast training speed. It is applied to the monitoring of grouting behind the shield wall, which can help output the grouting condition behind the shield wall more accurately and intelligently in real time. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0028] Figure 1 is a schematic diagram of a flow chart of an embodiment of the present invention;
[0029] Figure 2 It is a structural schematic diagram of the test device in the present invention.
[0030] Figure numerals: 1-air compressor, 11-pressure regulating valve, 21-barrel body, 22-upper flange, 23-lower flange, 24-fixed permeable baffle, 25-inlet valve, 26-outlet valve, 31-acoustic sensor, 32-waveguide rod, 33-processor. DETAILED DESCRIPTION
[0031] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians of the art without making creative work are within the scope of protection of the present application.
[0032] A monitoring method for shield wall back grouting condition based on GRU neural network, such as Figure 1 As shown, the following steps are included:
[0033] S1: Connect the test device and conduct air tightness test on the pipeline and the grouting model tube.
[0034] The test device includes a pressure system, a grouting model tube, and a data collection and processing system. The pressure system includes an air compressor 1 that can realize servo, and the servo pressure is 0.8Bar. The pressure system is connected to the grouting model tube through an air pipeline, which is used to pressurize the grouting model tube; preferably, a pressure regulating valve 11 is provided on the air pipeline; the grouting model tube includes a tube body 21 made of transparent organic glass, the upper side of the tube body 21 is connected to an upper flange 22, and the lower side of the tube body 21 is connected to a lower flange 23. The gaps between the upper flange 22, the lower flange 23 and the tube body 21 are filled with rubber sealing strips, and a fixed water-permeable partition 24 is provided radially inside the tube body 21; the tube body 2 An inlet valve 25 is provided at the top of the cylinder 21, and an outlet valve 26 is provided at the bottom of the cylinder 21 below the fixed permeable baffle 24; the data collection and processing system includes an acoustic sensor 31, a waveguide rod 32 and a processor 33, the acoustic sensor 31 is connected to the waveguide rod 32, and the acoustic sensor 31 is connected to the processor 33 for communication, and the acoustic sensor 31 collects the bending waveguide signal generated by the friction and collision between the soil particles and the waveguide rod 32 during the grouting process of the formation simulation system and transmits it to the processor 33, and the processor 33 processes, transforms, analyzes and saves the collected bending waveguide signal.
[0035] S2: Prepare the mud-reducing slurry and ethylene glycol aqueous solution. After the preparation is completed, start filling the grouting model cylinder, and then fill the ethylene glycol aqueous solution, sand, and mud-reducing slurry in turn, and bury the waveguide rod of the data collection and processing system in the sand.
[0036] The ethylene glycol concentration in the ethylene glycol aqueous solution is 50%, and the ethylene glycol aqueous solution is selected because it has low volatility and good stability; the grouting material and proportion of the mud-killing slurry should be determined according to the actual project.
[0037] The method of filling the grouting model cylinder is:
[0038] S21, open the upper flange 22, and pour ethylene glycol aqueous solution into the grouting model tube to the position of the fixed permeable partition 24;
[0039] S22, filling the sand collected from the site layer by layer, burying the waveguide rod 32 of the data collection and processing system therein, and compacting each layer of sand and soil and scraping the surface;
[0040] S23, pour a sufficient amount of grouting slurry above the sand layer, close the upper flange 22, and complete the filling of the grouting model tube.
[0041] S3: Pressurize the inside of the grouting model cylinder through the pressure system, observe the seepage process, and close the outlet valve 26 when the mud-killing slurry fills the entire sand layer.
[0042] The method for pressurizing the grouting model tube is to first close the inlet valve 25 and the outlet valve 26, adjust the pressure regulating valve 11 to the target test pressure, and then open the inlet valve 25 and the outlet valve 26 to fill the grouting model tube with pressure.
[0043] S4: The signal generated by the waveguide rod 32 is monitored by an acoustic sensor, the monitoring signal is extracted and processed, and then substituted into the GRU model for training.
[0044] In step S4, the monitoring signal extraction and processing method is: the waveguide signal captured by the acoustic sensor 31 is passed through a 60dB amplifier to improve the signal-to-noise ratio, filter the background noise, and then convert the electrical signal into a digital signal, which is then processed by a processor and signal analysis software.
[0045] S5: During on-site construction, waveguide rods are installed in the soil layer for real-time monitoring. After real-time monitoring data is processed, it is input into the trained GRU model to export the dynamic graph of the grouting effect in real time.
[0046] Specifically, to install the waveguide rod in the soil layer, firstly, it is necessary to reserve the waveguide rod installation hole on the shield segment, and then, after the shield machine is excavated, the waveguide rod of each ring of segments is installed immediately after the installation is completed, and 4 waveguide rods are installed in each ring in a centrally symmetrical distribution. Finally, grouting is started after 4 waveguide rods are installed in each ring of segments. The data processing is the same as the monitoring signal processing method in step S4. The processed data is imported into the GRU model trained in step S4 in real time. The GRU model realizes real-time monitoring of the grouting condition behind the shield wall through learning analysis, and derives a dynamic graph of the grouting effect.
[0047] S6: According to the dynamic diagram of grouting effect, comprehensively consider other engineering factors and determine the grouting pressure and grouting volume.
[0048] The other engineering factors include the permeability of the stratum, the water and soil pressure of the surrounding strata, the excavation route, the slurry type, the segment strength and the shield type, etc.
[0049] Furthermore, step S7 is included: after the grouting is completed, the grouting range is detected, and then the GRU model is imported for correction learning. Preferably, the grouting range is detected by geophysical exploration. The ground penetrating radar non-destructive testing technology can detect the medium within a thickness range of about 1m behind the shield segment wall, and the diffusion range of the grouting body and other information can be identified. The data obtained by the detection is then substituted into the GRU model training to improve the wear monitoring accuracy of the model.
[0050] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, technicians familiar with the field may make various corresponding changes and deformations based on the present invention, but these corresponding changes and deformations should all fall within the scope of protection of the claims attached to the present invention.
Claims
1. A method for monitoring the back grouting condition of a shield wall based on a GRU neural network, characterized in that: The steps include: S1: Connection of the test device to conduct air tightness test of pipeline and grouting model tube; the test device includes a pressure system, a grouting model tube and a data collection and processing system, the pressure system is connected to the grouting model tube through an air pipeline, the data collection and processing system includes an acoustic sensor, a waveguide rod and a processor, the acoustic sensor is connected to the waveguide rod, and the acoustic sensor is connected to the processor for communication; S2: Prepare the mud-reducing slurry and ethylene glycol aqueous solution. After the preparation is completed, fill the grouting model cylinder, then fill the ethylene glycol aqueous solution, sand, and mud-reducing slurry in sequence, and bury the waveguide rod of the data collection and processing system in the sand; S3: Pressurize the inside of the grouting model cylinder through the pressure system and observe the seepage process until the mud-reducing slurry fills the sand layer; S4: The signal generated by the waveguide rod is monitored by an acoustic sensor, and the monitoring signal is transmitted to the processor for extraction and processing, and then substituted into the GRU model for training; S5: During on-site construction, waveguide rods are installed in the soil layer for real-time monitoring. After real-time monitoring data is processed, it is input into the trained GRU model to export the dynamic graph of grouting effect in real time; S6: According to the dynamic diagram of grouting effect, comprehensively consider other engineering factors and determine the grouting pressure and grouting volume.
2. The method for monitoring the back grouting condition of the shield wall based on the GRU neural network according to claim 1 is characterized in that: The pressure system includes an air compressor capable of realizing servo, and a pressure regulating valve is arranged on the air pipeline.
3. The method for monitoring the back grouting condition of the shield wall based on the GRU neural network according to claim 2 is characterized in that: The grouting model cylinder comprises a cylinder body made of transparent organic glass, the upper side of the cylinder body is connected to an upper flange, the lower side of the cylinder body is connected to a lower flange, the connection gaps between the upper flange, the lower flange and the cylinder body are filled with rubber sealing strips, and a fixed water-permeable baffle is radially arranged inside the cylinder body; an inlet valve is arranged on the upper part of the cylinder body, and an outlet valve is arranged below the fixed water-permeable baffle at the lower part of the cylinder body; The acoustic sensor collects the bending waveguide signals generated by the friction and collision between sand particles and the waveguide rod during the grouting process in the grouting model tube and transmits them to the processor, which processes, converts, analyzes and saves the collected bending waveguide signals.
4. The method for monitoring the back grouting condition of the shield wall based on the GRU neural network according to claim 1 is characterized in that: The ethylene glycol concentration in the ethylene glycol aqueous solution is 50%.
5. The method for monitoring the back grouting condition of the shield wall based on the GRU neural network according to claim 3 is characterized in that: In step S2, the method of filling the grouting mold cylinder is: S21, open the upper flange, and pour ethylene glycol aqueous solution into the grouting model cylinder to fix the permeable baffle position; S22, filling the sand and soil collected from the site layer by layer, compacting each layer of sand and soil and scraping the surface; S23. Pour a sufficient amount of grouting slurry above the sand layer, close the upper flange, and complete the filling of the grouting model tube.
6. The method for monitoring the back grouting condition of the shield wall based on the GRU neural network according to claim 3 is characterized in that: In step S3, the step of pressurizing the grouting model tube is as follows: closing the inlet valve and outlet valve of the grouting model tube, adjusting the pressure regulating valve to the target test pressure, and then opening the inlet valve and outlet valve to charge the grouting model tube with pressure from the pressure system.
7. The method for monitoring the back grouting condition of the shield wall based on the GRU neural network according to claim 1 is characterized in that: In step S4, the monitoring signal extraction and processing method is as follows: the waveguide signal captured by the acoustic sensor is passed through a 60dB amplifier to improve the signal-to-noise ratio, filter the background noise, and then convert the electrical signal into a digital signal, after which the data is processed by a processor and signal analysis software.
8. The method for monitoring the back grouting condition of the shield wall based on the GRU neural network according to claim 1 is characterized in that: In step S5, the method for installing the waveguide rod in the soil layer is as follows: waveguide rod installation holes are reserved on the shield segments, and after the shield machine is excavated, the waveguide rods of each ring of segments are immediately installed after the installation is completed, and 4 waveguide rods are installed in each ring in a centrally symmetrical distribution. Grouting begins after 4 waveguide rods are installed in each ring of segments.
9. The method for monitoring the back grouting condition of the shield wall based on the GRU neural network according to claim 1 is characterized in that: The other engineering factors include the permeability of the formation, the water and soil pressure of the surrounding formations, the excavation route, the slurry type, the segment strength and the shield type.
10. The method for monitoring the back grouting condition of the shield wall based on the GRU neural network according to claim 1, characterized in that: It also includes step S7: after the grouting is completed, the grouting range is detected, and then the GRU model is imported for correction learning; the grouting range is detected by using the geophysical method, and the medium within the 1m thickness range behind the shield segment wall is detected by the ground penetrating radar non-destructive testing technology to identify the diffusion range information of the grouting body, and the detected data is then substituted into the GRU model training.
Citation Information
Cited By
Sandy gravel stratum catastrophe grouting prevention and control virtual-real blending collaborative test method and system
CN119992947A
