Over-excavation monitoring and early warning method and system for tunneling working face based on artificial intelligence
By using artificial intelligence-based methods in the boring machine, using image matching analysis and Kalman filtering algorithm to calculate the accurate initial value of prediction errors, the early warning of the boring machine's super-excavation situation is achieved, and the prediction model in the existing technology is solved The problems of slow fitting speed and low accuracy improve the accuracy and speed of early warning.
Patent Information
- Application Number
- CN202411308054.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-19
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-09-19
AI Technical Summary
The prior art cannot warning the overexhaust in advance in the boring machine, resulting in the unavoidable loss of overexhaust. The traditional Kalman filtering algorithm initializes the prediction error setting method, resulting in slow fitting speed and low accuracy of the prediction model.
Using an artificial intelligence-based method, the prediction speed is obtained by acquiring tunnel images, using the state transition equation of the Kalman filtering algorithm, and the confidence and prediction error are calculated through image matching analysis, and the accurate prediction error initial value is set to assist in super-drawing monitoring. Early warning.
It improves the accuracy of image measurement speed, provides fast and accurate early warning analysis, and avoids the occurrence of super-excavation phenomenon.
Smart Images

Figure CN118815498B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of monitoring and early warning of tunnel boring machines, and in particular to an artificial intelligence-based tunnel boring face over-excavation monitoring and early warning method and system. Background Art
[0002] When tunnel boring machines are digging tunnels, they often exceed the preset digging range. When the digging range is exceeded, it will not only affect the progress of the project, but also affect the quality of the project.
[0003] The existing patent application document CN116241265A shows an over-excavation warning device, method, system, equipment and medium for a cantilever roadheader. The patent document determines whether the cutting head has over-excavated by the position of the cutting head relative to the tunnel. The over-excavation warning method in the patent document can only issue a warning when over-excavation has occurred. When over-excavation has occurred, it means that over-excavation losses have already occurred. Although issuing a warning at this time can prevent further over-excavation, the over-excavation losses before the warning cannot be compensated.
[0004] In order to solve the problem of over-excavation warning after the fact, it is necessary to give early warning when over-excavation is likely to occur in the future, so as to help the tunnel boring machine operator to make timely adjustments to avoid the occurrence of over-excavation. In order to know in advance whether the tunnel boring machine is likely to over-excavate, it is necessary to predict the excavation speed at a future moment, and then obtain the excavation position at a future moment.
[0005] As a commonly used prediction algorithm, the Kalman filter algorithm can be used to predict the mining speed at future moments. Before using the Kalman filter algorithm for prediction, it is necessary to initialize an initial prediction error and fit the prediction model based on the initial prediction error. Traditionally, a non-zero value is randomly initialized as the initial prediction error. This method of setting the initial prediction error will result in a slower prediction model fitting speed and lower prediction model fitting accuracy. Summary of the invention
[0006] In order to solve the problems of slow prediction model fitting speed and low prediction model fitting accuracy, the present invention provides an artificial intelligence-based tunneling face over-excavation monitoring and early warning method and system.
[0007] In a first aspect, the present invention provides an artificial intelligence-based over-excavation monitoring and early warning method for a tunneling working face, comprising:
[0008] Collecting the first tunnel image and the second tunnel image at each moment;
[0009] The state transfer equation of the Kalman filter algorithm is used to obtain the first predicted speed at each moment;
[0010] Acquire key points in the first tunnel image and the second tunnel image at each moment, combine any three non-collinear key points in the first tunnel image to obtain a plurality of key point combinations, perform image matching analysis based on the key point combinations, and calculate matching reliability;
[0011] Screening points from key points in the first tunnel image; obtaining the measured distance from the ranging point to the tunnel boring machine based on the matching results of each key point combination; dividing the difference between the measured distance obtained from the ranging point in the first tunnel image at one moment and the measured distance from the matching pixel point of the ranging point in the second tunnel image at the previous moment to the tunnel boring machine by the time interval to obtain the image measurement speed at the moment, taking the difference between the image measurement speed and the first predicted speed as the prediction error, and calculating the confidence, wherein the confidence is positively correlated with the matching reliability and negatively correlated with the matching deviation of the ranging point;
[0012] Taking confidence as weight, all prediction errors at each moment are weighted and calculated to obtain the final prediction error at each moment. The final prediction error is used to set the initial value of the prediction error of the Kalman filter algorithm to assist in over-digging monitoring and early warning.
[0013] The present invention determines the influence of image matching error on the accuracy of image measurement speed by analyzing the image matching accuracy in the process of obtaining image measurement speed, thereby obtaining a relatively accurate image measurement speed. The relatively accurate image measurement speed can be used as a comparison object of the first prediction speed to obtain a relatively accurate prediction error, thereby providing an accurate initial value of the prediction error to the prediction algorithm, providing a basis for accurate and rapid early warning analysis.
[0014] Preferably, the image matching analysis based on key point combination includes:
[0015] According to the positional relationship between the three key points in the key point combination and the corresponding most similar pixel points in the second tunnel image, the second tunnel image is mapped and transformed to obtain the matching relationship between the first tunnel image under the key point combination and each pixel point in the second tunnel image; the matching relationship between each pixel point in the first tunnel image at adjacent moments under the key point combination is obtained;
[0016] The corresponding matching pixel points of the three key points in the key point combination in the first tunnel image at the previous moment constitute a matching pixel point combination of the key point combination, and the matching relationship between each pixel point in the first tunnel image at the previous moment and the second tunnel image under the matching pixel point combination is obtained.
[0017] The present invention obtains a variety of matching results by performing matching analysis based on each key point combination, thereby providing more data basis for subsequent image measurement speed analysis.
[0018] Preferably, the method for obtaining the matching reliability is:
[0019] A key point in the first tunnel image at the first moment is used as the analysis key point, and the matching reliability is positively correlated with the information entropy of the window where each key point in a key point combination is located, and is negatively correlated with the distance between the matching pixel point of an analysis key point in the second tunnel image under the matching relationship based on the key point combination and the most similar pixel point, and is negatively correlated with the distance between the matching pixel point of the analysis key point in the first tunnel image at the previous moment and the most similar pixel point, and is also negatively correlated with the distance between the matching pixel point of the analysis key point in the first tunnel image at the previous moment and the most similar pixel point in the second tunnel image under the matching relationship based on the matching pixel point combination.
[0020] The present invention reflects the matching reliability by matching the distance between the pixel point and the most similar pixel point and the information entropy, etc., which can accurately describe the accuracy of each matching result and provide a basis for the subsequent accurate analysis and measurement of the accuracy of the distance.
[0021] Preferably, screening the ranging points from the key points in the first tunnel image includes:
[0022] Calculate the probability of ranging points;
[0023] The key point corresponding to the maximum possibility of the ranging point in the first tunnel image at a moment is used as the ranging point in the first tunnel image at that moment.
[0024] The present invention selects key points with higher matching accuracy from multiple key points as distance measurement points. The distance measurement points obtained in this way can more accurately obtain the image measurement speed, providing a basis for the subsequent accurate acquisition of the initial value of the prediction error.
[0025] Preferably, obtaining the measured distance from the distance measuring point to the roadheader based on the matching result of each key point combination includes:
[0026] Based on the matching relationship between each pixel point in the first tunnel image and the second tunnel image at a moment under a key point combination, the binocular ranging algorithm is used to obtain the measured distance from the ranging point in the first tunnel image at that moment under the key point combination to the tunnel boring machine.
[0027] Preferably, the method for obtaining the matching deviation of the ranging point includes:
[0028] The matching deviation of the ranging point is positively correlated with the distance between the matching pixel point of the ranging point in the second tunnel image and the most similar pixel point, and is positively correlated with the distance between the matching pixel point of the ranging point in the first tunnel image at the previous moment and the most similar pixel point, and is also positively correlated with the distance between the matching pixel point of the ranging point in the first tunnel image at the previous moment and the most similar pixel point in the second tunnel image.
[0029] The present invention considers the distance relationship between the matching pixel point of the distance measuring point and the most similar pixel point when analyzing the matching deviation, so as to accurately describe the matching error of the distance measuring point, and provide a basis for the subsequent judgment of the accuracy of the image measurement speed.
[0030] Preferably, the step of setting the initial value of the prediction error of the Kalman filter algorithm by using the final prediction error includes:
[0031] The average of the final prediction errors at all times is taken as the comprehensive prediction error, and the comprehensive prediction error is taken as the initial value of the prediction error of the Kalman filter algorithm.
[0032] Preferably, the auxiliary over-excavation monitoring and early warning includes:
[0033] Obtain the GPS positioning data of the tunnel boring machine at each moment, and subtract the GPS positioning data at each moment from the GPS positioning data at the previous moment to obtain the positioning speed at each moment. Based on the positioning speeds at all moments, use the Kalman filter algorithm to perform prediction analysis to obtain the second predicted speed at the future moment. Multiply the time difference between the future moment and the current moment by the second predicted speed, and add the GPS positioning data at the current moment to obtain the tunnel boring position at the future moment.
[0034] Determine whether the excavation position at a future moment is within the design range. If it is not within the design range, an early warning alarm will be issued.
[0035] Preferably, the state transfer equation using the Kalman filter algorithm to obtain the first predicted speed at each moment includes:
[0036] The Kalman filter algorithm is used to construct a state transfer equation according to the output power of the roadheader, and the output power at each moment is input into the state transfer equation to obtain the first predicted speed at each moment.
[0037] In the second aspect, the present invention provides an artificial intelligence-based tunneling face over-excavation monitoring and early warning system, comprising: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned artificial intelligence-based tunneling face over-excavation monitoring and early warning method is implemented.
[0038] By adopting the above technical solution, the above-mentioned artificial intelligence-based excavation working face over-excavation monitoring and early warning method is generated into a computer program and stored in a memory to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.
[0039] The present invention has the following technical effects:
[0040] The present invention analyzes the image matching accuracy in the process of obtaining image measurement speed to determine the influence of image matching error on the accuracy of image measurement speed, thereby obtaining a relatively accurate image measurement speed, so that the relatively accurate image measurement speed can be used as a comparison object of the first prediction speed to obtain a relatively accurate prediction error, thereby providing an accurate initial value of the prediction error to the prediction algorithm, and further providing a basis for the prediction algorithm to quickly and accurately fit the prediction model, so as to assist in rapid and accurate early warning analysis. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts.
[0042] Figure 1 It is a flow chart of a method in an artificial intelligence-based tunneling face over-excavation monitoring and early warning method according to an embodiment of the present invention. DETAILED DESCRIPTION
[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0044] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.
[0045] The embodiment of the present invention discloses an artificial intelligence-based over-excavation monitoring and early warning method for a tunneling working face, referring to Figure 1 , comprising steps S1 to S6:
[0046] S1: Acquire the first tunnel image and the second tunnel image at each moment.
[0047] Specifically, cameras are installed on the left and right sides of the rear end of the tunnel boring machine, and the cameras are used to capture images of the excavated tunnel every 30 seconds. The images captured by the left camera at each moment are recorded as first tunnel images, and the images captured by the right camera at each moment are recorded as second tunnel images.
[0048] S2: Obtain a first predicted speed.
[0049] It should be noted that in the Kalman filter algorithm, in order to achieve a faster fitting of the prediction model, a relatively accurate initial value of the prediction error needs to be set. In order to obtain a relatively accurate initial value of the prediction error, the prediction error value at each moment needs to be obtained; and the prediction error value at each moment is the difference between the prediction value at each moment and the true value (accurate value). Therefore, in order to obtain the prediction error value at each moment, the prediction value at each moment needs to be obtained first.
[0050] Optionally, as an example, a method for obtaining the first predicted speed includes:
[0051] The Kalman filter algorithm is used to construct a state transfer equation based on the output power of the roadheader. The output power at each moment is input into the state transfer equation to obtain the first predicted speed at each moment.
[0052] It should be noted that constructing a state transfer equation using a Kalman filter algorithm is a prior art and will not be described in detail here.
[0053] S3: Acquire key points in the first tunnel image and the second tunnel image at each moment, combine any three non-collinear key points in the first tunnel image to obtain a plurality of key point combinations, perform image matching analysis based on the key point combinations, and calculate matching reliability.
[0054] It should be noted that in order to obtain a relatively accurate prediction error at each moment, it is also necessary to obtain a relatively accurate true value at each moment. In this embodiment, the true value is the real excavation speed of the tunnel boring machine. Since the real excavation speed cannot be directly obtained, the speed can only be measured to reflect the real excavation speed. The means of obtaining and measuring the excavation speed are: one is to measure the excavation speed through GPS positioning data, and the other is to measure the excavation speed through positioning data obtained by image analysis; both means of measuring the excavation speed will have errors. For example, in underground tunnels, the accuracy of GPS positioning is low, so the relatively accurate excavation speed cannot be obtained based on GPS positioning. Since there are many factors causing GPS positioning errors, the GPS positioning error cannot be analyzed relatively simply using the influencing factors, and thus accurate positioning data cannot be obtained; the other is to obtain positioning data through image analysis. Since the tunnel is relatively dim, the quality index of the collected image is relatively low. The low image quality index will reduce the image matching quality, and image matching is an important step in the positioning process. Therefore, the accuracy of traditional image positioning analysis is poor; but it is now known that the image matching quality is the main factor affecting the positioning analysis error, and other factors have little effect. Therefore, relatively accurate positioning data can be obtained by analyzing the image matching quality. The position can be reflected by the relative distance between objects. Therefore, by analyzing the image matching quality, the relatively accurate relative distance between objects can be obtained, and then relatively accurate positioning data can be obtained.
[0055] S30: Acquire key points in the first tunnel image and the second tunnel image at each moment.
[0056] It should be noted that in order to analyze the quality of image matching, image matching analysis needs to be performed, and image matching analysis requires obtaining key points first.
[0057] Optionally, as an example, acquiring key points in the first tunnel image and the second tunnel image at each moment includes:
[0058] The method of obtaining key points in the scale invariant feature transform matching algorithm (Scale Invariant Feature Transform, referred to as Sift algorithm) is used to process the first tunnel image and the second tunnel image respectively to obtain several key points in the first tunnel image and several key points in the second tunnel image.
[0059] It should be noted that using the Sift algorithm to obtain key points is an existing technology and will not be described in detail here.
[0060] S31: Combining any three non-collinear key points in the first tunnel image to obtain a plurality of key point combinations.
[0061] It should be noted that according to the prior art, as long as the positions of the three non-collinear points in the plane are determined, the position of the plane can be determined. Therefore, as long as the matching relationship between the three non-collinear points between the two images is determined, the matching relationship between the two images is also determined. Therefore, every three non-collinear points between the images will determine a matching result. In order to analyze the accuracy of the matching, it is necessary to analyze the accuracy of each matching result.
[0062] Any three non-collinear key points in the first tunnel image are combined to obtain a plurality of key point combinations.
[0063] S32: Perform image matching analysis based on key point combination.
[0064] Optionally, as an example, image matching analysis is performed based on a key point combination, including:
[0065] The method of calculating the similarity between key points in the Sift algorithm is used to calculate the similarity between a key point in the key point combination and each key point in the second tunnel image, and the key point corresponding to the maximum similarity in the second tunnel image is taken as the best matching pixel point of the key point in the key point combination in the second tunnel image.
[0066] According to the positional relationship between the three key points in the key point combination and the corresponding best matching pixel points in the second tunnel image, the second tunnel image is mapped and transformed according to the existing mapping transformation method to obtain the matching relationship between the first tunnel image under the key point combination and each pixel point in the second tunnel image.
[0067] Similarly, the matching relationship between the pixels in the first tunnel image at adjacent moments under the key point combination is obtained.
[0068] Based on the matching relationship between each pixel point in the first tunnel image at adjacent moments under the key point combination, the matching pixel point of a key point in the key point combination in the first tunnel image at the previous moment is obtained, and the corresponding matching pixel points of the three key points in the key point combination in the first tunnel image at the previous moment form a matching pixel point combination. The matching relationship between each pixel point in the first tunnel image at the previous moment and the second tunnel image under the matching pixel point combination is obtained.
[0069] S33: Calculate matching reliability.
[0070] It should be noted that in order to evaluate the matching quality of the matching results under each key point combination, it is necessary to analyze the image matching deviation. When the matching effect of the key point combination is good, each key point should match the most similar pixel point. Therefore, the matching effect of each key point combination can be reflected by analyzing the position deviation between the matching pixel point of the key point under the key point combination and the most similar pixel point. At the same time, accurate matching can only be performed when the information around each key point and the corresponding matching pixel point in the key point combination is relatively rich; when the information around each key point and the corresponding matching pixel point in the key point combination is lost, it is very likely that the matching error will be caused by the lack of information. Therefore, the matching reliability of each key point combination can also be described by analyzing the richness of the surrounding information of each key point and the corresponding matching pixel point in the key point combination.
[0071] Preferably, as an example, the matching reliability satisfies the relationship:
[0072]
[0073] Among them, the first tunnel image at a moment Key points are used as analysis key points, and the matching results based on any key point combination are represents the position coordinates of the matching pixel points of the analysis key point in the second tunnel image, represents the position coordinates of the best matching pixel point of the analysis key point in the second tunnel image, The position coordinates are The pixel point and position coordinates are The vector of pixels, Representation vector The larger the value, the smaller the position coordinate is. The pixel point and position coordinates are The larger the distance between the pixel points, the greater the difference between the matching result of the first tunnel image and the second tunnel image based on the key point combination and the optimal matching result, and the worse the matching accuracy of the first tunnel image and the second tunnel image based on the key point combination. represents the position coordinates of the matching pixel points of the key analysis point in the first tunnel image at the previous moment, represents the position coordinates of the best matching pixel point of the key analysis point in the first tunnel image at the previous moment, The position coordinates are The pixel point and position coordinates are The vector of pixels, Representation vector The larger the value, the smaller the position coordinate is. The pixel point and position coordinates are The greater the distance between the pixel points, the greater the difference between the matching result and the optimal matching result of the matching between the first tunnel images at adjacent moments based on the key point combination, and thus the worse the matching accuracy of the matching between the first tunnel images at adjacent moments based on the key point combination; the matching result of the matching pixel point combination corresponding to the key point combination, represents the position coordinates of the matching pixel point in the second tunnel image of the matching pixel point in the first tunnel image at the previous moment of the analysis key point, represents the position coordinates of the best matching pixel point in the second tunnel image of the matching pixel point of the first tunnel image at the previous moment of the analysis key point, The position coordinates are The pixel point and position coordinates are The vector of pixels, Representation vector The larger the value, the smaller the position coordinate is. The pixel point and position coordinates are The greater the distance between the pixel points, the greater the difference between the matching result of the first tunnel image at the previous moment and the second tunnel image based on the matching pixel point combination and the optimal matching result, and thus the worse the matching accuracy of the first tunnel image at the previous moment and the second tunnel image based on the matching pixel point combination. Indicates the preset anti-zero parameter. In this embodiment, 0.001 is taken as an example for description. Other embodiments may take other values, and this embodiment does not specifically limit it. represents the number of key points in the first tunnel image; Indicates the key point combination The information entropy of the window where the key point is located is larger. The richer the information around a key point, the more accurate the key point matching result based on the richer information. Indicates the key point combination The information entropy of the key point in the window where the matching pixel point of the second tunnel image is located is larger, indicating that the key point combination is The richer the information around the matching pixel of a key point in the second tunnel image, the more accurate the key point matching result based on the richer information. Indicates the key point combination The information entropy of the matching pixel point in the first tunnel image at the previous moment in the window where the key point is located is larger. The richer the information around the matching pixel of the key point in the first tunnel image at the previous moment, the more accurate the matching result based on the richer information. Indicates the key point combination The information entropy of the window where the matching pixel of the key point in the first tunnel image at the previous moment is located in the second tunnel image is located. The larger the value, the higher the information entropy of the key point in the first tunnel image at the previous moment in the window where the matching pixel of the second tunnel image is located. The richer the information around the matching pixel point in the second tunnel image of the matching pixel point in the first tunnel image at the previous moment, the more accurate the matching result obtained based on the richer information. Indicates the matching reliability of the key point combination.
[0074] Among them, as an example, the method of obtaining the window is: taking a pixel point as the center area as a window. Indicates the preset window side length. In this embodiment, The value 11 is taken as an example for description. Other embodiments may take other values, and this embodiment does not make any specific limitation.
[0075] S4: selecting distance measuring points from the key points in the first tunnel image; and obtaining the measured distance from the distance measuring point to the tunnel boring machine based on the matching result of each key point combination.
[0076] S40: Screening distance measurement points from key points in the first tunnel image.
[0077] It should be noted that, in order to determine the relative distance between the tunnel boring machine and other positions at each moment, one of the multiple other positions needs to be selected as the positioning reference point. In this embodiment, the positioning reference point is represented by a distance measurement point.
[0078] Optionally, as an example, screening the ranging points from the key points in the first tunnel image includes: taking any key point in the first tunnel image as the ranging point.
[0079] It should be noted that when using image analysis to perform positioning analysis, in order to obtain accurate positioning results, it is necessary to ensure that the matching of the selected ranging points is relatively accurate, otherwise the relative position obtained based on a ranging point with an incorrect matching is also inaccurate. For example, the actual position of the ranging point in the first tunnel image at a certain moment is A, and the ranging point A accurately matches the pixel point B in the first tunnel image at the previous moment, but due to the matching error, A is matched with the pixel point C in the first tunnel image at the previous moment. The accurate moving distance should be the difference between the ranging value based on A and the ranging value based on B, while the moving distance obtained in the actual process is the difference between the ranging value based on A and the ranging value based on C. This leads to errors in the acquisition of the moving distance. Therefore, a key point with higher matching accuracy should be selected as the ranging point to ensure the accuracy of positioning.
[0080] Preferably, as an example, screening the ranging points among the key points in the first tunnel image includes steps S400-S401.
[0081] S400: Calculate the distance measurement point possibility.
[0082] Preferably, as an example, the ranging point possibility satisfies the relationship:
[0083]
[0084] in, Indicates The first tunnel image at a moment under the combination of key points The position coordinates of the matching pixels of the key points in the second tunnel image, Indicates In the first tunnel image at this moment under the combination of key points The position coordinates of the best matching pixel point of the key point in the second tunnel image, The position coordinates are The pixel point and position coordinates are The vector of pixels, Representation vector The larger the value, the smaller the position coordinate is. The pixel point and position coordinates are The larger the distance between the pixels of the first tunnel image, the larger the distance between the pixels of the first tunnel image. The key point is based on The matching accuracy of the key point combination is poor. When the matching result of a key point is poor, the measured distance obtained by using it as a distance measurement point is also inaccurate, so it should not be selected as a distance measurement point; Indicates In the first tunnel image at this moment under the combination of key points The position coordinates of the matching pixel points of the key points in the first tunnel image at the previous moment, Indicates In the first tunnel image at this moment under the combination of key points The position coordinates of the best matching pixel point of the key point in the first tunnel image at the previous moment, The position coordinates are The pixel point and position coordinates are The vector of pixels, Representation vector The larger the value, the smaller the position coordinate is. The pixel point and position coordinates are The larger the distance between the pixels, the larger the distance between the pixels. The key point is based on The matching accuracy of the key point combination is poor. When the matching result of a key point is poor, the measured distance obtained by using it as a distance measurement point is also inaccurate, so it should not be selected as a distance measurement point; Indicates In the first tunnel image at this moment under the matching pixel point combination corresponding to the key point combination The position coordinates of the matching pixel points in the second tunnel image of the matching pixel points of the key points in the first tunnel image at the previous moment, Indicates In the first tunnel image at this moment under the combination of key points The best matching pixel point in the second tunnel image of the matching pixel point of the key point in the first tunnel image at the previous moment, The position coordinates are The pixel point and position coordinates are The vector of pixels, Representation vector The larger the value, the smaller the position coordinate is. The pixel point and position coordinates are The larger the distance between the pixels of the first tunnel image, the larger the distance between the pixels of the first tunnel image. The matching pixel of the key point at the previous moment is based on the The matching accuracy of the matching pixel point combination corresponding to the key point combination is poor. When the matching result of the key point at the previous moment is poor, The measured distance obtained by using a key point as a distance measurement point is also inaccurate, so it should not be selected as a distance measurement point. Indicates The matching reliability of the key point combination. Indicates the number of keypoint combinations. Indicates the first tunnel image The possibility of ranging points for key points.
[0085] S401: Screening distance measurement points from key points of the first tunnel image according to the distance measurement point possibilities.
[0086] Optionally, as an example, a key point corresponding to the maximum possibility of the ranging point in the first tunnel image is used as the ranging point.
[0087] S41: Obtain the measured distance from the distance measuring point to the tunnel boring machine based on the matching result of each key point combination.
[0088] Optionally, as an example, based on the matching relationship between each pixel point in the first tunnel image and the second tunnel image at a moment under a key point combination, a binocular ranging algorithm is used to obtain the measured distance from the ranging point in the first tunnel image at that moment under the key point combination to the tunnel boring machine.
[0089] S5: The difference between the measured distance obtained by the ranging point in the first tunnel image at one moment and the measured distance from the matching pixel point of the ranging point in the second tunnel image at the previous moment to the tunnel boring machine is divided by the time interval to obtain the image measurement speed at the moment, and the difference between the image measurement speed and the first predicted speed is taken as the prediction error, and the confidence is calculated.
[0090] S50: The image measurement speed at a moment is obtained by dividing the difference between the measured distance obtained by the distance measuring point in the first tunnel image at a moment and the measured distance from the matching pixel point of the distance measuring point in the second tunnel image at the previous moment to the tunnel boring machine by the time interval.
[0091] Optionally, as an example, the image measurement speed at a moment is obtained by dividing the difference between the measured distance at a moment and the measured distance from the matching pixel point of the distance measurement point to the roadheader at the previous moment by the time interval, including:
[0092] According to the matching relationship between the pixels in the first tunnel image at adjacent moments under a key point combination, the measured distance from the matching pixel point of the distance measuring point in the first tunnel image at one moment to the tunnel boring machine at the previous moment is obtained, and recorded as the reference measured distance at that moment based on the key point combination.
[0093] The image measurement speed at this moment based on the key point combination is obtained by dividing the difference between the measured distance from the distance measuring point to the tunnel boring machine in the first tunnel image at this moment under the key point combination and the reference measured distance by the time interval.
[0094] Optionally, as an example, the time interval is the time difference between the current moment and the previous moment.
[0095] S51: Calculate confidence.
[0096] Preferably, as an example, the confidence satisfies the relationship:
[0097]
[0098] in, Indicates The matching reliability of the key point combination. The larger the value, the higher the matching accuracy based on the z-th key point combination. Therefore, the image measurement speed obtained based on the matching result of the key point combination is more accurate. Indicates The position coordinates of the matching pixel points of the ranging point in the first tunnel image at a moment under the combination of key points in the second tunnel image, represents the position coordinates of the best matching pixel point in the second tunnel image of the ranging point in the first tunnel image at a moment, The position coordinates are The pixel point and position coordinates are The vector of pixels, Representation vector The larger the value, the smaller the position coordinate is. The pixel point and position coordinates are The larger the distance between pixels, the larger the The matching accuracy of the distance measurement points in the first tunnel image and the second tunnel image under the combination of key points is poor. When the matching accuracy is poor, the distance measurement result obtained is less accurate. Indicates The position coordinates of the matching pixel points in the first tunnel image at the previous moment under the combination of key points in the first tunnel image at one moment, represents the position coordinates of the best matching pixel point of the ranging point in the first tunnel image at one moment in the first tunnel image at the previous moment, The position coordinates are The pixel point and position coordinates are The vector of pixels, Representation vector The larger the value, the smaller the position coordinate is. The pixel point and position coordinates are The larger the distance between pixels, the larger the The matching accuracy between the first tunnel image under the combination of key points and the ranging points in the first tunnel image at the previous moment is poor. When the matching accuracy is poor, the ranging result obtained is less accurate. Indicates The position coordinates of the matching pixel points in the second tunnel image of the ranging point in the first tunnel image at the moment and the matching pixel points in the first tunnel image at the previous moment under the matching pixel point combination corresponding to the key point combination, The position coordinates of the best matching pixel point in the second tunnel image representing the distance measurement point in the first tunnel image at this moment and the matching point in the first tunnel image at the previous moment, The position coordinates are The pixel point and position coordinates are The vector of pixels, Representation vector The larger the value, the smaller the position coordinate is. The pixel point and position coordinates are The larger the distance between pixels, the larger the The matching accuracy of the ranging points under the matching pixel point combination of the key point combination in the first tunnel image at the previous moment and the corresponding matching pixel points in the second tunnel image is poor. When the matching accuracy is poor, the ranging result obtained is less accurate. Indicates the distance measurement point possibility of the distance measurement point. The larger the value, the higher the overall matching accuracy of the distance measurement point, and thus the higher the accuracy of the speed measurement based on the image of the distance measurement point; Indicates The confidence level of the prediction error at that moment under the combination of key points.
[0099] S6: Using confidence as weight, all prediction errors at each moment are weighted to obtain the final prediction error at each moment, and the final prediction error is used to set the initial value of the prediction error of the prediction algorithm to assist in over-digging monitoring and early warning.
[0100] S60: Taking the confidence as the weight, all prediction errors at each moment are weighted and calculated to obtain the final prediction error at each moment.
[0101] Optionally, as an example, all prediction errors at each moment are weighted and calculated to obtain the final prediction error at each moment, including:
[0102] The confidence level is used as the weight of the prediction error at a moment under each key point combination, and the prediction errors at that moment under all key point combinations are weighted summed to obtain the final prediction error at that moment.
[0103] S61: Using the final prediction error, an initial value of the prediction error of the prediction algorithm is set to assist in over-digging monitoring and early warning.
[0104] Optionally, as an example, the final prediction error is used to set the initial value of the prediction error of the prediction algorithm to assist in over-digging monitoring and early warning, including:
[0105] The GPS positioning data of the tunnel boring machine at each moment is obtained, and the positioning speed at each moment is obtained by subtracting the GPS positioning data at each moment from the GPS positioning data at the previous moment.
[0106] The average of the final prediction errors at all times is used as the comprehensive prediction error, and the comprehensive prediction error is used as the initial value of the prediction error of the Kalman filter algorithm; based on the positioning speed at all times, all Kalman filter algorithms are used for prediction analysis to obtain the second predicted speed at the future time. The time difference between the future time and the current time is multiplied by the second predicted speed at the future time, and then the GPS positioning data at the current time is added to obtain the excavation position at the future time, and it is determined whether the excavation position at the future time is within the design range. If it is not within the design range, an early warning alarm is issued.
[0107] An embodiment of the present invention also discloses an artificial intelligence-based excavation working face over-excavation monitoring and early warning system, including a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, the artificial intelligence-based excavation working face over-excavation monitoring and early warning method according to the present invention is implemented.
[0108] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.
[0109] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high bandwidth memory, a hybrid storage cube, etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device.
[0110] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.
[0111] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. An artificial intelligence-based monitoring and early warning method for over-excavation in tunneling working faces. It is characterized in that Includes steps: Collecting the first tunnel image and the second tunnel image at each moment; The state transfer equation of the Kalman filter algorithm is used to obtain the first predicted speed at each moment; Acquire key points in the first tunnel image and the second tunnel image at each moment, combine any three non-collinear key points in the first tunnel image to obtain a plurality of key point combinations, perform image matching analysis based on the key point combinations, and calculate matching reliability; screening ranging points from key points in the first tunnel image; The measured distance from the ranging point to the tunnel boring machine is obtained based on the matching result of each key point combination; the image measurement speed at the moment is obtained by dividing the difference between the measured distance obtained from the ranging point in the first tunnel image at one moment and the measured distance from the matching pixel point of the ranging point in the second tunnel image at the previous moment to the tunnel boring machine by the time interval, the difference between the image measurement speed and the first predicted speed is taken as the prediction error, and the confidence is calculated, and the confidence is positively correlated with the matching reliability and negatively correlated with the matching deviation of the ranging point; Using confidence as weight, all prediction errors at each moment are weighted to obtain the final prediction error at each moment, and the final prediction error is used to set the initial value of the prediction error of the Kalman filter algorithm; Obtain the GPS positioning data of the tunnel boring machine at each moment, and subtract the GPS positioning data at each moment from the GPS positioning data at the previous moment to obtain the positioning speed at each moment. Based on the positioning speeds at all moments, use the Kalman filter algorithm to perform prediction analysis to obtain the second predicted speed at the future moment. Multiply the time difference between the future moment and the current moment by the second predicted speed, and add the GPS positioning data at the current moment to obtain the tunnel boring position at the future moment. Determine whether the excavation position at a future moment is within the design range. If it is not within the design range, an early warning alarm will be issued.
2. The artificial intelligence-based over-excavation monitoring and early warning method for tunneling working face according to claim 1, It is characterized in that The image matching analysis based on key point combination includes: According to the positional relationship between the three key points in the key point combination and the corresponding most similar pixel points in the second tunnel image, the second tunnel image is mapped and transformed to obtain the matching relationship between the first tunnel image under the key point combination and each pixel point in the second tunnel image; the matching relationship between each pixel point in the first tunnel image at adjacent moments under the key point combination is obtained; The corresponding matching pixel points of the three key points in the key point combination in the first tunnel image at the previous moment constitute a matching pixel point combination of the key point combination, and the matching relationship between each pixel point in the first tunnel image at the previous moment and the second tunnel image under the matching pixel point combination is obtained.
3. The artificial intelligence-based over-excavation monitoring and early warning method for tunneling working face according to claim 2, It is characterized in that The method for obtaining the matching reliability is: A key point in the first tunnel image at the first moment is used as the analysis key point, and the matching reliability is positively correlated with the information entropy of the window where each key point in a key point combination is located, and is negatively correlated with the distance between the matching pixel point of an analysis key point in the second tunnel image under the matching relationship based on the key point combination and the most similar pixel point, and is negatively correlated with the distance between the matching pixel point of the analysis key point in the first tunnel image at the previous moment and the most similar pixel point, and is also negatively correlated with the distance between the matching pixel point of the analysis key point in the first tunnel image at the previous moment and the most similar pixel point in the second tunnel image under the matching relationship based on the matching pixel point combination.
4. The artificial intelligence-based over-excavation monitoring and early warning method for tunneling working face according to claim 2, It is characterized in that The step of screening the ranging points from the key points in the first tunnel image comprises: Calculate the probability of ranging points; The key point corresponding to the maximum possibility of the ranging point in the first tunnel image at a moment is used as the ranging point in the first tunnel image at that moment.
5. The artificial intelligence-based over-excavation monitoring and early warning method for tunneling working face according to claim 1, It is characterized in that The method of obtaining the measured distance from the distance measuring point to the roadheader based on the matching result of each key point combination includes: Based on the matching relationship between each pixel point in the first tunnel image and the second tunnel image at a moment under a key point combination, the binocular ranging algorithm is used to obtain the measured distance from the ranging point in the first tunnel image at that moment under the key point combination to the tunnel boring machine.
6. The artificial intelligence-based over-excavation monitoring and early warning method for tunneling working face according to claim 1, It is characterized in that The method for obtaining the matching deviation of the ranging point includes: The matching deviation of the ranging point is positively correlated with the distance between the matching pixel point of the ranging point in the second tunnel image and the most similar pixel point, and is positively correlated with the distance between the matching pixel point of the ranging point in the first tunnel image at the previous moment and the most similar pixel point, and is also positively correlated with the distance between the matching pixel point of the ranging point in the first tunnel image at the previous moment and the most similar pixel point in the second tunnel image.
7. The artificial intelligence-based over-excavation monitoring and early warning method for tunneling working face according to claim 1, It is characterized in that The method of setting the initial value of the prediction error of the Kalman filter algorithm by using the final prediction error includes: The average of the final prediction errors at all times is taken as the comprehensive prediction error, and the comprehensive prediction error is taken as the initial value of the prediction error of the Kalman filter algorithm.
8. The artificial intelligence-based over-excavation monitoring and early warning method for tunneling working face according to claim 1, It is characterized in that The state transfer equation using the Kalman filter algorithm to obtain the first predicted speed at each moment includes: The Kalman filter algorithm is used to construct a state transfer equation according to the output power of the roadheader, and the output power at each moment is input into the state transfer equation to obtain the first predicted speed at each moment.
9. Over-excavation monitoring and early warning system for tunneling working face based on artificial intelligence, It is characterized in that include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an artificial intelligence-based excavation working face over-excavation monitoring and early warning method according to any one of claims 1 to 8 is implemented.
Citation Information
Patent Citations
Over-excavation early warning device, method, system and equipment for cantilever type heading machine and medium
CN116241265A
Payload estimation system and method
AU2007242056A1
Real-time positioning method for underground roadway tunneling equipment based on UWB
CN113075650A
Three-dimensional synchronous modeling and ground pressure response processing method and system based on goaf scanning
CN114565712A
Tunnel face front survey method
JP2017156106A
