Method and system for detecting defects of frame rail type tunnel structure

Abstract

The invention discloses a method and system for detecting defects of a frame rail type tunnel structure, and relates to the technical field of tunnel defect detection.The method is characterized in that temperature information in a tunnel is obtained in real time, and the temperature difference within a preset time interval is calculated based on the temperature information; obtaining vibration information in the tunnel in real time, wherein the vibration information comprises vibration amplitude and vibration frequency; if the temperature difference or the vibration amplitude or the vibration frequency meets a preset condition, image information in the tunnel is collected, and based on the image information in the tunnel, whether the tunnel structure has defects or not is judged through a pre-trained defect recognition model; if yes, the position of the defect is determined, and defect characteristic parameters are measured; and determining a repair period based on the defect characteristic parameters, and repairing the defect to the position of the defect within the repair period. By adopting the scheme provided by the invention, the detection efficiency and the detection precision of tunnel defects can be improved, and the working intensity of workers and the resource consumption of system equipment are reduced.

Description

Technical Field

[0001] The present application relates to the technical field of tunnel defect detection, and in particular to a method and system for detecting structural defects of a track-mounted tunnel. Background Art

[0002] Track-mounted tunnels include subway tunnels and railway tunnels. Subway and railway transportation are important components of the transportation system. Therefore, safety is a prerequisite for subway and railway operations and is the top priority of operation and maintenance. For the detection of tunnel defects, traditional manual measurement uses visual inspection and total station, which has the disadvantages of high detection intensity, low efficiency, and insufficient precision. In the prior art, the detection efficiency of tunnel defects can be greatly improved by analyzing the real-time collected images with the help of intelligent image analysis models. However, subway tunnels and railway tunnels are usually dark in light and are affected by weather (such as fog), which may lead to insufficient clarity of images collected in the tunnel, making it impossible for the image analysis model to make accurate judgments. Therefore, there is still the disadvantage of insufficient precision. In addition, in this method, the image acquisition equipment in the tunnel needs to be kept online at all times, but there are no defects in the images collected during most of the time in this process. Real-time processing of a large amount of image information will bring a lot of additional burden to the server, and the resource consumption is large. Summary of the invention

[0003] In order to overcome the above-mentioned problems or partially solve the above-mentioned problems, the purpose of this application is to provide a method and system for detecting structural defects of a track-type tunnel, so as to improve the detection efficiency and detection accuracy of tunnel defects, ensure the timeliness and effectiveness of defect detection, and reduce the workload of staff and the resource consumption of system equipment.

[0004] This application is implemented through the following technical solutions: In a first aspect, the present embodiment provides a method for detecting structural defects of a track-type tunnel, the method comprising: acquiring temperature information in the tunnel in real time, and calculating a temperature difference within a predetermined time interval based on the temperature information; acquiring vibration information in the tunnel in real time, the vibration information comprising a vibration amplitude and a vibration frequency; if the temperature difference is greater than a temperature difference threshold or the vibration amplitude is greater than an amplitude threshold or the vibration frequency is greater than a frequency threshold, collecting image information in the tunnel, and judging whether the tunnel structure has defects based on the image information in the tunnel and using a pre-trained defect recognition model; if so, determining the location of the defect and driving a defect detection robot to the location of the defect to measure defect characteristic parameters; determining a repair period based on the defect characteristic parameters, and repairing the defect within the repair period; recording historical data, and preparing training samples for training a defect prediction model based on the historical data; the historical data comprising the time when the defect was discovered, the location of the defect, the material in which the defect occurred, and temperature information and vibration information within N days before the defect was discovered; predicting the time and location of the defect using a defect prediction model, and driving the defect detection robot to measure abnormal characteristic parameters of the predicted location before the predicted time; judging whether the predicted location has an abnormality based on the abnormal characteristic parameters of the predicted location; and if so, repairing the abnormality of the predicted location before the predicted time.

[0005] Based on the first aspect, in an embodiment of the present application, the detection method further includes: iteratively training the GAN network in the initial state to obtain the defect recognition model, the GAN network includes a generator and a discriminator, which includes the following iterative training process: Step 1: Initialize the parameters of the generator G and the parameters of the discriminator D ; Step 2: Sample a real samples and a noise samples from the real tunnel image samples, use the noise samples as the input of the generator, and obtain a generated samples through the generator; fix the parameters of the generator , optimize the parameters of the training discriminator , to improve the accuracy of the discriminator in distinguishing real samples from generated samples; where the images in the real samples are crack images, the images in the noise samples are crack-like non-crack images, and a is a positive integer; Step 3: Cycle k times to optimize and update parameters After that, fix the parameters of the discriminator , optimize the parameters of the training generator , to reduce the gap between the generated samples and the real samples; where k is a positive integer; Step 4: Repeat steps 2-3 until the sample discrimination probability of the discriminator meets the expected value.

[0006] Based on the first aspect, in an embodiment of the present application, the defect is a crack, the defect characteristic parameters include the depth of the crack, and determining the repair period based on the defect characteristic parameters includes: judging whether the depth of the crack is greater than a threshold depth, and if so, determining the repair period to be T1, otherwise determining the repair period to be T2; wherein T1<T2.

[0007] Based on the first aspect, in an embodiment of the present application, the method for measuring the depth H of the crack includes: selecting n detection points equidistantly along the surface of the crack curve; wherein n is a positive integer; The laser ranging device is used to traverse and scan all the detection points to measure the depth of all the detection points. If the detection surface is regarded as a plane, at the i-th detection point, the laser ranging device is at the position point, and the laser is shot into the crack from the incident point at an incident angle. The laser shot can irradiate the deepest point, and the depth of the i-th detection point can be expressed as: ,in, Indicates the deepest point With location point The distance between Indicates the incident point With location point The distance between represents the angle of incidence; The depth H of the crack has a value range of: ,in, Indicates the depth of the nth detection point.

[0008] Based on the first aspect, in the embodiment of the present application, the anomaly includes a lining void, the anomaly characteristic parameter includes the depth of the void, and the depth of the void The measurement methods include: Determine the dielectric constant of the lining medium and the propagation speed of electromagnetic waves in the lining medium; Obtain a ground penetrating radar waveform at the predicted position, and read the two-way propagation time of the electromagnetic wave from the inner wall surface of the tunnel to the upper interface of the cavity and the two-way propagation time of the electromagnetic wave from the inner wall surface of the tunnel to the upper interface of the cavity based on the ground penetrating radar waveform; The calculation formula of the depth of the cavity is: ;in, represents the dielectric constant of the lining medium, represents the depth of the upper interface of the cavity, and its calculation formula is: ; It represents the depth of the lower interface of the cavity, and its calculation formula is: ;in, represents the propagation speed of electromagnetic waves in the lining medium, represents the two-way propagation time of electromagnetic waves from the inner wall surface of the tunnel to the upper interface of the cavity, It represents the two-way propagation time of electromagnetic waves from the inner wall surface of the tunnel to the lower interface of the cavity.

[0009] Based on the first aspect, in an embodiment of the present application, determining the dielectric constant of the lining medium and the propagation speed of the electromagnetic wave in the lining medium includes: Randomly select m test points on the inner wall of the tunnel and measure the lining thickness at the jth test point among the m test points; where j and m are both positive integers; Record electromagnetic waves in the lining thickness of The two-way propagation time in the lining medium, where the electromagnetic wave propagates along the thickness direction of the lining medium; The calculation expression of the dielectric constant of the lining medium is: , where C represents the propagation speed of electromagnetic waves in the air, represents the lining thickness at the jth test point, It means that the electromagnetic wave is Two-way travel time in the lining medium; The calculation expression of the propagation speed of electromagnetic waves in lining media is: .

[0010] Based on the first aspect, in an embodiment of the present application, the abnormalities include lining voids and loose lining.

[0011] In a second aspect, the present application provides a system for detecting structural defects of a track-type tunnel, the system comprising: The temperature detection module is used to obtain the temperature information in the tunnel in real time, and calculate the temperature difference within a predetermined time interval based on the temperature information; the vibration information detection module is used to obtain the vibration information in the tunnel in real time, and the vibration information includes the vibration amplitude and the vibration frequency; the defect detection robot is used to measure the characteristic parameters of defects or anomalies; the first execution module is used to collect image information in the tunnel when the temperature difference is greater than the temperature difference threshold or the vibration amplitude is greater than the amplitude threshold or the vibration frequency is greater than the frequency threshold, and based on the image information in the tunnel, call a pre-trained defect recognition model to determine whether there is a defect in the tunnel structure; the second execution module is used to determine the location of the defect and drive the defect detection robot to the location of the defect to measure the defect when there is a defect in the tunnel structure. defect characteristic parameters; a deadline prediction module, used to determine the repair deadline based on the defect characteristic parameters, and repair the defects within the repair deadline; a model training module, used to record historical data, and to produce training samples for training the defect prediction model based on the historical data; the historical data includes the time when the defect was discovered, the location of the defect, the material in which the defect occurred, and the temperature information and vibration information N days before the defect was discovered; a defect prediction model, used to predict the time and location of the defect; a third execution module, used to drive the defect detection robot to measure the abnormal characteristic parameters of the predicted location before the predicted time; an abnormality detection module, used to determine whether there is an abnormality at the predicted location based on the abnormal characteristic parameters of the predicted location; if so, repair the abnormality at the predicted location before the predicted time.

[0012] Based on the second aspect, in an embodiment of the present application, the temperature detection module includes a plurality of temperature detection modules, and the plurality of temperature detection modules are equidistantly arranged along the tunnel and close to the inner wall of the tunnel.

[0013] Based on the second aspect, in an embodiment of the present application, the vibration information detection module includes: an amplitude detection module, which is arranged on the side wall or top wall of the tunnel, and is used to measure the vibration amplitude of the inner wall of the tunnel in real time; a frequency detection module, which is arranged on the side wall or top wall of the tunnel, and is used to measure the vibration frequency of the inner wall of the tunnel in real time.

[0014] Based on the second aspect, in an embodiment of the present application, a defect feature detection device and an abnormal feature detection device are installed on the detection robot; the detection robot is used to drive the defect feature detection device and the abnormal feature detection device to move in the target tunnel so that the detection range of the defect feature detection device covers all areas in the target tunnel; the defect feature detection device includes a laser ranging device, and the abnormal feature detection equipment includes a geological radar.

[0015] Compared with the prior art, this application has at least the following advantages and beneficial effects: In a method for detecting structural defects of a track-type tunnel provided by the present invention, by real-time monitoring of the changes in parameters such as temperature and vibration amplitude that may cause tunnel defects, the activation timing of the image acquisition device and the defect recognition model is controlled, thereby reducing the working time of the image acquisition device and the operating burden of the cloud server (reducing resource consumption) while ensuring the timeliness of defect detection. In addition, when the defect recognition model determines that there is a "defect" in the tunnel structure, the accuracy of the judgment of the defect recognition model can be further confirmed by measuring the characteristic parameters of the "defect", that is, the characteristic parameters can reflect whether the "defect" is real, thereby avoiding unnecessary work for maintenance personnel due to misjudgment of the defect recognition model; wherein, whether the "defect" is real can be reflected by the length of the given repair period. The proposal of the repair period helps maintenance personnel to arrange their work more reasonably or provide a reference for the optimal deployment of automatic repair equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present application and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 A schematic flow chart of a method for detecting structural defects of a track-type tunnel provided in an embodiment of the present invention; Figure 2 A cross-sectional schematic diagram at the i-th detection point on a crack provided by an embodiment of the present invention; Figure 3 A schematic block diagram of the structure of a system for detecting structural defects of a track-mounted tunnel provided in an embodiment of the present invention.

[0017] Marks and corresponding parts names in the attached drawings: 1. Temperature detection module; 11. Temperature detection module; 2. Vibration information detection module; 21. Amplitude detection module; 22. Frequency detection module; 3. First execution module; 4. Second execution module; 5. Defect detection robot; 6. Deadline prediction module; 7. Model training module; 8. Defect prediction model; 9. Third execution module; 10. Anomaly detection module. DETAILED DESCRIPTION

[0018] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with examples and drawings. The illustrative implementation scheme of the present application and its description are only used to explain the present application and are not intended to limit the present application.

[0019] Please refer to Figure 1 , Figure 1 A schematic flow chart of a method for detecting structural defects of a track-type tunnel provided in an embodiment of the present invention is shown in FIG. Figure 1 As shown, the method includes: S1. Acquire temperature information in the tunnel in real time, and calculate the temperature difference within a predetermined time interval based on the temperature information; S2. Acquire vibration information in the tunnel in real time, wherein the vibration information includes vibration amplitude and vibration frequency; Specifically, common tunnel structural defects include cracks, collapse, water leakage, etc. Without considering sudden factors (such as mudslides, earthquakes, etc.), collapse and water leakage are usually manifestations of further deterioration of cracks. Therefore, this embodiment takes crack detection as an example for discussion. The reasons for the formation of cracks in tunnel structures include temperature, humidity, load bearing, vibration amplitude / frequency, and aging of its own materials, among which temperature and vibration amplitude / frequency changes are usually the direct causes of crack formation. Taking railway tunnels as an example, they are usually located in the mountains in the wild, and the temperature changes during the day and night are large. Thermal expansion and contraction may cause cracking of the tunnel lining, and there are trains passing through the tunnel almost every day. Each time a train passes, the tunnel structure will vibrate. Each vibration may cause deformation inside the tunnel structure, causing uneven stress and cracking. Therefore, this embodiment mainly considers the relationship between the two major factors of temperature and vibration and cracks.

[0020] Furthermore, since subway tunnels and some railway tunnels are usually long along the line, the temperature and vibration at different locations may not be consistent. Therefore, in this embodiment, it is necessary to arrange multiple temperature (sensor) detection modules equidistantly along the tunnel to measure the temperature at different locations. In addition, in order to avoid affecting the operation of the train and prevent the temperature detection module from being sucked into the bottom of the train and causing losses, it is also necessary to make the installation position of the temperature detection module away from the rails (that is, set at a position close to the inner wall of the tunnel); further, the same arrangement can be made for the vibration information detection device, the difference is that the vibration information detection device includes at least two functions, which are used to monitor the vibration amplitude and vibration frequency of the inner wall of the tunnel respectively. Based on the above settings, the position or range where the temperature difference or vibration amplitude / frequency exceeds the threshold can be accurately located, and then the range where the image needs to be collected can be determined.

[0021] S3. If the temperature difference is greater than the temperature difference threshold, or the vibration amplitude is greater than the amplitude threshold, or the vibration frequency is greater than the frequency threshold, image information in the tunnel is collected, and based on the image information in the tunnel, a pre-trained defect recognition model is used to determine whether there are defects in the tunnel structure; Specifically, the setting of the temperature difference threshold needs to be based on the properties of the tunnel structure material, and an exemplary setting of 15-20°C is acceptable. Usually, the vibration generated in the tunnel is mainly caused by the passing of trains, so the setting of the amplitude threshold and frequency threshold should be slightly smaller than the vibration amplitude and frequency generated by the passing of ordinary trains, that is, at least one defect (crack) detection is ensured every time a train passes.

[0022] Specifically, in this embodiment, the defect recognition model is obtained by iteratively training the GAN network in the initial state, wherein the GAN network includes a generator and a discriminator, and includes the following iterative training process: Step 1: Initialize the parameters of the generator G and the parameters of the discriminator D ; Step 2: Sample a real samples and a noise samples from the real tunnel image samples, use the noise samples as the input of the generator, and obtain a generated samples through the generator; fix the parameters of the generator , optimize the parameters of the training discriminator , to improve the accuracy of the discriminator in distinguishing real samples from generated samples; where the images in the real samples are crack images, the images in the noise samples are crack-like non-crack images, and a is a positive integer.

[0023] Specifically, noise samples refer to anomalies, outliers, or incorrectly labeled data samples that exist in the data set and do not conform to the true distribution (for example, in this implementation, if the image in the true sample is an image of a crack, then the noise sample is an image that does not contain a crack but has water marks or shadows that are very similar in shape to the crack).

[0024] Step 3: Cycle k times to optimize and update parameters After that, fix the parameters of the discriminator , optimize the parameters of the training generator , to reduce the gap between generated samples and real samples; where k is a positive integer.

[0025] Specifically, k can be set to a fixed constant value.

[0026] Step 4: Repeat steps 2-3 until the sample discrimination probability of the discriminator meets the expected value.

[0027] Specifically, since the ideal situation of the GAN network is that the sample discrimination probability is 0.5, that is, the probability that the discriminator D judges that the object sample is a real sample is 0.5, and the probability that the object sample is a generated sample is also 0.5, that is, the discriminator D cannot distinguish between real samples and generated samples, but in actual operation, you only need to set the expected value to a value close to 0.5.

[0028] After the training is completed, the generator is discarded, and the discriminator with high-precision discrimination ability is used as the defect recognition model to determine whether the tunnel structure has defects. Based on the above steps, a higher-precision discriminator can be trained with as few labeled samples (real samples) as possible, thereby reducing the classification task (distinguishing whether there are defects and the type of defects, etc.) from relying on a large amount of labeled data. In this way, the workload of collecting and making samples in the early stage of defect recognition model training can be greatly reduced.

[0029] S4, determine the location of the defect and drive the defect detection robot to the location of the defect to measure the defect characteristic parameters; determine the repair period based on the defect characteristic parameters, and repair the defect within the repair period (notify the maintenance personnel or command the automatic repair equipment to go to the location of the defect within the repair period).

[0030] Specifically, the location of the defect can be determined by reversely deducing the location of the image acquisition device. In addition, in this step, the defect characteristic parameters can be measured by using a detection robot. The detection robot is driven to the location of the defect to accurately measure the characteristic parameters of the "defect" identified by the defect recognition model. If the detection robot approaches the "defect" and finds that the defect recognition model is a misjudgment, it can also be reflected through the characteristic parameters (for example, the crack depth is detected to be zero). Furthermore, a defect characteristic detection device is also installed on the detection robot; the detection robot can drive the defect characteristic detection device to move in the target tunnel so that the detection range of the defect characteristic detection device covers all areas in the target tunnel; the defect characteristic detection device includes a laser ranging device.

[0031] Furthermore, determining the repair period based on the defect characteristic parameters includes: judging whether the depth of the crack is greater than a threshold depth; if so (indicating that the situation is more urgent at this time), determining the repair period to be T1 (which can be 3 to 7), otherwise determining the repair period to be T2 (which can be 15 to 60); wherein T1 < T2.

[0032] Exemplarily, if the defect is a crack, and the defect characteristic parameter includes the depth of the crack, then the method for measuring the depth H of the crack includes: Select n detection points equidistantly along the crack curve surface; Use a laser rangefinder to scan all the detection points and measure the depth of all the detection points; If the detection surface is considered as a plane, please refer to Figure 2 , at the i-th detection point, the laser ranging device is at the position point, and the laser is shot into the crack from the incident point at an incident angle. The injected laser can irradiate the deepest point, and the depth of the i-th detection point can be expressed as: ,in, Indicates the deepest point With location point The distance between Indicates the incident point With location point The distance between represents the angle of incidence; The depth H of the crack has a value range of: ,in, Indicates the depth of the nth detection point.

[0033] In the above method, based on the correlation between temperature difference and cracks, vibration and cracks, the location where cracks may occur is predicted to ensure the timeliness of defect detection. At the same time, based on the changes in temperature and vibration amplitude parameters, the activation timing of image acquisition equipment and defect recognition model is controlled, which can greatly reduce resource consumption.

[0034] Furthermore, the detection method also includes: S5. Record historical data and create training samples for training defect prediction models based on the historical data; the historical data includes the time when the defect was discovered, the location of the defect, the material in which the defect occurred, and the temperature information and vibration information within N days before the defect was discovered; S6. Predict the time and location of defect occurrence using a defect prediction model, and measure characteristic parameters of the predicted location before the predicted time; determine whether there is an abnormality at the predicted location based on the characteristic parameters of the predicted location; if so, notify maintenance personnel or command automatic repair equipment to go to the predicted location before the predicted time to repair the abnormality, and N can be 1 to 5 for example.

[0035] Specifically, the anomalies include lining voids and loose linings. The above two anomalies may further deteriorate and generate cracks or cause collapse. Therefore, in order to improve the safety of tunnel use, maintenance can be carried out in advance when anomalies are detected to avoid defects (cracks). For example, taking lining voids as an example, the abnormal characteristic parameters that need to be measured include the depth of the void. The depth of the void The measurement methods include: Determine the dielectric constant of the lining medium and the propagation speed of electromagnetic waves in the lining medium; Obtain a ground penetrating radar waveform at the predicted position, and read the two-way propagation time of the electromagnetic wave from the inner wall surface of the tunnel to the upper interface of the cavity and the two-way propagation time of the electromagnetic wave from the inner wall surface of the tunnel to the upper interface of the cavity based on the ground penetrating radar waveform; The calculation formula of the depth of the cavity is: ;in, represents the dielectric constant of the lining medium, represents the depth of the upper interface of the cavity, and its calculation formula is: ; It represents the depth of the lower interface of the cavity, and its calculation formula is: ;in, represents the propagation speed of electromagnetic waves in the lining medium, represents the two-way propagation time of electromagnetic waves from the inner wall surface of the tunnel to the upper interface of the cavity, It represents the two-way propagation time of electromagnetic waves from the inner wall surface of the tunnel to the lower interface of the cavity.

[0036] Specifically, the determining of the dielectric constant of the lining medium and the propagation speed of the electromagnetic wave in the lining medium includes: Randomly select m test points on the inner wall of the tunnel and measure the lining thickness at the jth test point among the m test points; where j and m are both positive integers; Record electromagnetic waves in the lining thickness of The two-way propagation time in the lining medium, where the electromagnetic wave propagates along the thickness direction of the lining medium; The calculation expression of the dielectric constant of the lining medium is: , where C represents the propagation speed of electromagnetic waves in the air, represents the lining thickness at the jth test point, It means that the electromagnetic wave is Two-way travel time in the lining medium; The calculation expression of the propagation speed of electromagnetic waves in lining media is: That is, in this embodiment, the propagation speed is taken as the average value of m test points to ensure the accuracy and universality of the data.

[0037] Combined with the detection method provided in the above embodiment, this method controls the activation time of the image acquisition device and the defect recognition model by real-time monitoring of the parameter changes that may cause the formation of tunnel defects, such as temperature and vibration amplitude, thereby reducing the working time of the image acquisition device and the operating burden of the cloud server (reducing resource consumption) and ensuring the timeliness of defect detection. In addition, when the defect recognition model determines that there is a "defect" in the tunnel structure, the accuracy of the judgment of the defect recognition model can be further confirmed by measuring the characteristic parameters of the "defect", that is, the characteristic parameters can reflect whether the "defect" is real, so as to avoid adding unnecessary work to the maintenance personnel due to the misjudgment of the defect recognition model; among which, whether the "defect" is real can be reflected by the length of the given repair period. The proposal of the repair period helps maintenance personnel to arrange their work more reasonably or provide a reference for the optimal deployment of automatic repair equipment.

[0038] The embodiment of the present invention also provides a system for detecting structural defects of a track-type tunnel, such as Figure 3As shown, the detection system includes: a temperature detection module 1, which is used to obtain the temperature information in the tunnel in real time, and calculate the temperature difference within a predetermined time interval based on the temperature information; a vibration information detection module 2, which is used to obtain the vibration information in the tunnel in real time, and the vibration information includes the vibration amplitude and the vibration frequency; a defect detection robot 5, which is used to measure the characteristic parameters of defects or anomalies; a first execution module 3, which is used to collect image information in the tunnel when the temperature difference is greater than the temperature difference threshold or the vibration amplitude is greater than the amplitude threshold or the vibration frequency is greater than the frequency threshold, and based on the image information in the tunnel, call a pre-trained defect recognition model to determine whether there is a defect in the tunnel structure; a second execution module 4, which is used to determine the location of the defect and drive the defect detection robot to the location of the defect when there is a defect in the tunnel structure. The defect characteristic parameters are measured at the position; a deadline prediction module 6 is used to determine the repair deadline based on the defect characteristic parameters and repair the defects within the repair deadline; a model training module 7 is used to record historical data and produce training samples for training the defect prediction model based on the historical data; the historical data includes the time when the defect is discovered, the location of the defect, the material in which the defect occurs, and the temperature information and vibration information within N days before the defect is discovered; a defect prediction model 8 is used to predict the time and location of the defect; a third execution module 9 is used to drive the defect detection robot to measure the abnormal characteristic parameters of the predicted position before the predicted time; an abnormality detection module 10 is used to determine whether there is an abnormality at the predicted position based on the abnormal characteristic parameters of the predicted position; if so, the abnormality at the predicted position is repaired before the predicted time.

[0039] It can be seen that the detection system provided in this embodiment controls the activation time of the image acquisition device and the defect recognition model by real-time monitoring of the parameter changes that may cause the formation of tunnel defects, such as temperature and vibration amplitude, thereby reducing the working time of the image acquisition device and the operating burden of the cloud server (reducing resource consumption) and ensuring the timeliness of defect detection. In addition, when the defect recognition model determines that there is a "defect" in the tunnel structure, the accuracy of the judgment of the defect recognition model can be further confirmed by measuring the characteristic parameters of the "defect", that is, the characteristic parameters can reflect whether the "defect" is real, so as to avoid adding unnecessary work to the maintenance personnel due to the misjudgment of the defect recognition model; among which, whether the "defect" is real can be reflected by the length of the given repair period. The proposal of the repair period helps maintenance personnel to arrange their work more reasonably or provide a reference for the optimal deployment of automatic repair equipment.

[0040] Furthermore, the temperature detection module includes a plurality of temperature detection modules 11, and the plurality of temperature detection modules 11 are equidistantly arranged along the tunnel and close to the inner wall of the tunnel.

[0041] Furthermore, the vibration information detection module 2 includes: an amplitude detection module 21, which is arranged on the side wall or top wall of the tunnel, and is used to measure the vibration amplitude of the inner wall of the tunnel in real time; a frequency detection module 22, which is arranged on the side wall or top wall of the tunnel, and is used to measure the vibration frequency of the inner wall of the tunnel in real time.

[0042] Furthermore, the defect detection robot 5 is installed with defect feature detection equipment and abnormal feature detection equipment; the defect detection robot 5 is used to drive the defect feature detection equipment to move in the target tunnel so that the detection range of the defect feature detection equipment and the abnormal feature detection equipment covers all areas in the target tunnel; the defect feature detection equipment includes a laser ranging device, and the abnormal feature detection equipment includes a geological radar.

[0043] The system provided in the embodiment of the present application can be used to execute the method in the above-mentioned embodiment 1. Specifically, see the method for detecting structural defects of a track-type tunnel described in the above embodiment, which will not be described in detail here.

[0044] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only the specific implementation method of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A method for detecting structural defects of a track-type tunnel, characterized in that: Methods include: Acquire temperature information in the tunnel in real time, and calculate the temperature difference within a predetermined time interval based on the temperature information; Acquire vibration information in the tunnel in real time, wherein the vibration information includes vibration amplitude and vibration frequency; If the temperature difference is greater than the temperature difference threshold, or the vibration amplitude is greater than the amplitude threshold, or the vibration frequency is greater than the frequency threshold, image information in the tunnel is collected, and based on the image information in the tunnel, a pre-trained defect recognition model is used to determine whether there are defects in the tunnel structure; If it exists, the location of the defect is determined and the defect detection robot is driven to the location of the defect to measure the defect characteristic parameters; Determine the repair period based on defect characteristic parameters and repair the defects within the repair period; Record historical data and create training samples for training defect prediction models based on the historical data; the historical data includes the time when the defect was discovered, the location of the defect, the material in which the defect occurred, and the temperature information and vibration information within N days before the defect was discovered; The defect prediction model is used to predict the time and location of defect occurrence, and the defect detection robot is driven to measure the abnormal characteristic parameters of the predicted location before the predicted time; based on the abnormal characteristic parameters of the predicted location, it is determined whether there is an abnormality at the predicted location; if so, the abnormality at the predicted location is repaired before the predicted time.

2. A method for detecting structural defects of a track-type tunnel according to claim 1, characterized in that: The detection method further includes: iteratively training the GAN network in the initial state to obtain the defect recognition model, the GAN network includes a generator and a discriminator, and includes the following iterative training process: Step 1: Initialize the parameters of the generator G and the parameters of the discriminator D ; Step 2: Sample a real samples and a noise samples from the real tunnel image samples, use the noise samples as the input of the generator, and obtain a generated samples through the generator; fix the parameters of the generator , optimize the parameters of the training discriminator , to improve the accuracy of the discriminator in distinguishing real samples from generated samples; where the images in the real samples are crack images, the images in the noise samples are crack-like non-crack images, and a is a positive integer; Step 3: Cycle k times to optimize and update parameters After that, fix the parameters of the discriminator , optimize the parameters of the training generator , to reduce the gap between the generated samples and the real samples; where k is a positive integer; Step 4: Repeat steps 2-3 until the sample discrimination probability of the discriminator meets the expected value.

3. A method for detecting structural defects of a track-mounted tunnel according to claim 1, characterized in that: The defect is a crack, the defect characteristic parameter includes the depth of the crack, and determining the repair period based on the defect characteristic parameter includes: It is determined whether the depth of the crack is greater than a threshold depth. If so, the repair period is determined to be T1, otherwise the repair period is determined to be T2; wherein T1<T2.

4. A method for detecting structural defects of a track-mounted tunnel according to claim 3, characterized in that: The method for measuring the depth H of the crack includes: Select n detection points at equal intervals along the crack curve surface; where n is a positive integer; The laser ranging device is used to traverse and scan all the detection points to measure the depth of all the detection points. If the detection surface is regarded as a plane, at the i-th detection point, the laser ranging device is at the position point, and the laser is shot into the crack from the incident point at an incident angle. The laser shot can irradiate the deepest point, and the depth of the i-th detection point can be expressed as: ,in, Indicates the deepest point With location point The distance between Indicates the incident point With location point The distance between represents the incident angle; the depth H of the crack has a value range of: ,in, Indicates the depth of the nth detection point.

5. The method for detecting structural defects of a track-mounted tunnel according to claim 1, characterized in that: The anomaly includes a lining void, and the anomaly characteristic parameters include the depth of the void. The measurement methods include: Determine the dielectric constant of the lining medium and the propagation speed of electromagnetic waves in the lining medium; Obtain a ground penetrating radar waveform at the predicted position, and read the two-way propagation time of the electromagnetic wave from the inner wall surface of the tunnel to the upper interface of the cavity and the two-way propagation time of the electromagnetic wave from the inner wall surface of the tunnel to the upper interface of the cavity based on the ground penetrating radar waveform; The calculation formula of the depth of the cavity is: ;in, represents the dielectric constant of the lining medium, represents the depth of the upper interface of the cavity, and its calculation formula is: ; It represents the depth of the lower interface of the cavity, and its calculation formula is: ;in, represents the propagation speed of electromagnetic waves in the lining medium, represents the two-way propagation time of electromagnetic waves from the inner wall surface of the tunnel to the upper interface of the cavity, It represents the two-way propagation time of electromagnetic waves from the inner wall surface of the tunnel to the lower interface of the cavity.

6. A method for detecting structural defects of a track-mounted tunnel according to claim 5, characterized in that: The determining of the dielectric constant of the lining medium and the propagation speed of the electromagnetic wave in the lining medium comprises: Randomly select m test points on the inner wall of the tunnel and measure the lining thickness at the jth test point among the m test points; where j and m are both positive integers; Record electromagnetic waves in the lining thickness of The two-way propagation time in the lining medium, where the electromagnetic wave propagates along the thickness direction of the lining medium; The calculation expression of the dielectric constant of the lining medium is: , where C represents the propagation speed of electromagnetic waves in the air, represents the lining thickness at the jth test point, It means that the electromagnetic wave is Two-way travel time in the lining medium; The calculation expression of the propagation speed of electromagnetic waves in lining media is: .

7. A system for detecting structural defects of a track-type tunnel, characterized in that: The system includes: A temperature detection module is used to obtain temperature information in the tunnel in real time and calculate the temperature difference within a predetermined time interval based on the temperature information; A vibration information detection module is used to obtain vibration information in the tunnel in real time, wherein the vibration information includes vibration amplitude and vibration frequency; Defect detection robot, used to measure characteristic parameters of defects or anomalies; A first execution module, when the temperature difference is greater than a temperature difference threshold, or the vibration amplitude is greater than an amplitude threshold, or the vibration frequency is greater than a frequency threshold, is used to collect image information in the tunnel, and based on the image information in the tunnel, call a pre-trained defect recognition model to determine whether there are defects in the tunnel structure; The second execution module is used to determine the location of the defect and drive the defect detection robot to the location of the defect to measure the defect characteristic parameters when there is a defect in the tunnel structure; A deadline prediction module is used to determine the repair deadline based on defect characteristic parameters and repair the defect within the repair deadline; A model training module is used to record historical data and generate training samples for training defect prediction models based on the historical data; the historical data includes the time when the defect was discovered, the location of the defect, the material in which the defect occurred, and the temperature information and vibration information within N days before the defect was discovered; Defect prediction models, used to predict when and where defects will occur; The third execution module is used to drive the defect detection robot to measure the abnormal characteristic parameters of the predicted position before the predicted time; The anomaly detection module is used to determine whether there is an anomaly at the predicted location based on the anomaly feature parameters of the predicted location; if there is, the anomaly at the predicted location is repaired before the predicted time.

8. A system for detecting structural defects of a track-type tunnel according to claim 7, characterized in that: The temperature detection module comprises a plurality of temperature detection modules, and the plurality of temperature detection modules are equidistantly arranged along the tunnel and close to the inner wall of the tunnel.

9. A system for detecting structural defects of a track-type tunnel according to claim 7, characterized in that: The vibration information detection module includes: The amplitude detection module is arranged on the side wall or the top wall of the tunnel and is used to measure the vibration amplitude of the inner wall of the tunnel in real time; The vibration frequency detection module is arranged on the side wall or top wall of the tunnel and is used to measure the vibration frequency of the inner wall of the tunnel in real time.

10. A system for detecting structural defects of a track-type tunnel according to claim 7, characterized in that: The detection robot is equipped with defect feature detection equipment and abnormal feature detection equipment; the detection robot is used to drive the defect feature detection equipment and the abnormal feature detection equipment to move in the target tunnel so that the detection range of the defect feature detection equipment covers all areas in the target tunnel; the defect feature detection equipment includes a laser ranging device, and the abnormal feature detection equipment includes a geological radar.

Images