Ultra-wide thin-wall solid high pier temperature field real-time monitoring and crack early warning system and method

Abstract

The invention relates to the technical field of temperature monitoring, in particular to an ultra-wide thin-wall solid high pier temperature field real-time monitoring and crack early warning system and method. According to the method, the temperature data of the target temperature field and the image data of the target temperature field are preprocessed to obtain the preprocessed temperature data and the preprocessed temperature image, then the target temperature field model is constructed based on the preprocessed temperature data and the preprocessed temperature image, and the crack thermal stress is analyzed to obtain the crack thermal stress. And establishing a crack model by combining the target temperature field model and crack thermal stress analysis, finally obtaining a prediction result output by the trained crack model, judging whether a crack generation risk exists according to the prediction result output by the crack model, and if the crack generation risk exists, giving out an early warning. The real-time image data is compared with the normal image through the crack model, the risk of crack generation is predicted in combination with the predicted temperature of the target temperature field model, and early warning is given out in time.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature monitoring, and in particular to a system and method for real-time monitoring of temperature fields and early warning of cracks in ultra-wide thin-walled solid high piers. Background Art

[0002] High pier structures are widely used in bridges, tunnels and other major projects. Due to their large size, thin walls and wide spans, they often produce complex temperature field distributions in temperature difference environments. The temperature field refers to the collection of temperatures at each point in a material system. It is a function of time and space coordinates, reflecting the distribution of temperature in space and time. The temperature field can be divided into steady-state temperature field and unsteady-state temperature field. A steady-state temperature field refers to a temperature field in which the temperature of each point does not change with time, while an unsteady-state temperature field refers to a temperature field in which the temperature of each point changes with time.

[0003] A Chinese patent with publication number CN113466291A discloses an analysis method for detecting cracks in large-volume concrete based on temperature field changes. The crack conditions of the large-volume concrete are analyzed based on the temperature field, stress field and infrared thermal image, and the crack size is calculated using a customized calibration gauge and a perspective transformation formula to complete the crack analysis. However, the prior art only analyzes existing cracks, but it is difficult to predict the occurrence of cracks based on temperature data, resulting in the inability to prevent them in a timely manner, posing a major safety hazard. Summary of the invention

[0004] The purpose of the present invention is to address the problems existing in the background technology and to propose a real-time monitoring system and method for the temperature field of ultra-wide thin-walled solid high piers and a crack early warning system and method.

[0005] The technical solution of the present invention:

[0006] On the one hand, the present application provides a method for real-time monitoring of the temperature field and early warning of cracks in ultra-wide thin-walled solid high piers, including:

[0007] Setting acquisition parameters, acquiring temperature data of the target temperature field and image data of the target temperature field according to the acquisition parameters, and preprocessing the temperature data of the target temperature field and the image of the target temperature field respectively to obtain preprocessed temperature data and a preprocessed temperature image;

[0008] Constructing a target temperature field model based on preprocessed temperature data and preprocessed temperature images;

[0009] Conduct crack thermal stress analysis and create a crack model by combining the target temperature field model with the crack thermal stress analysis;

[0010] Collect real-time temperature data, input the real-time temperature data into the crack model in sequence, obtain the prediction results output by the trained crack model, and judge whether there is a risk of cracks based on the prediction results output by the crack model. If there is a risk of cracks, issue an early warning.

[0011] Preferably, setting acquisition parameters, acquiring temperature data of the target temperature field and image data of the target temperature field according to the acquisition parameters, and preprocessing the temperature data of the target temperature field and the image of the target temperature field respectively to obtain preprocessed temperature data and preprocessed temperature image, including:

[0012] Create a temperature data table;

[0013] Set acquisition parameters for each target temperature field respectively; the acquisition parameters include temperature acquisition position, image acquisition position, acquisition cycle and acquisition frequency. Generally speaking, temperature data and image data are one-to-one corresponding, and image data and temperature data at the same acquisition frequency are recorded as a data pair;

[0014] Collect multiple temperature data of each target temperature field according to the collection parameters, and add all the collected temperature data to the temperature data table;

[0015] According to the acquisition parameters, multiple temperature images of each target temperature field are respectively acquired, and all the acquired temperature images are added to the temperature data table; the acquired temperature images include crack images and normal images.

[0016] Preferably, setting acquisition parameters, acquiring temperature data of the target temperature field and image data of the target temperature field according to the acquisition parameters, and preprocessing the temperature data of the target temperature field and the image of the target temperature field respectively to obtain preprocessed temperature data and preprocessed temperature image, further comprising:

[0017] Select a target temperature field;

[0018] Select all temperature data of each acquisition time of the target temperature field from the temperature data table;

[0019] Calculate the average temperature of all temperature data at each acquisition time;

[0020] Set temperature deviation threshold;

[0021] Select a temperature data and determine whether the difference between the temperature data and the average temperature is greater than or equal to the temperature deviation threshold;

[0022] If the difference between the temperature data and the average temperature is greater than or equal to the temperature deviation threshold, the temperature data is deleted;

[0023] Return to select a temperature data and determine whether the difference between the temperature data and the average temperature is greater than or equal to the temperature deviation threshold, until all temperature data of the target temperature field are selected;

[0024] De-noising all temperature images in the temperature data table to obtain a de-noised image;

[0025] Binarize the denoised image to obtain a binary image;

[0026] Set a crack evaluation standard, and divide the binary image into a crack image and a normal image according to the crack evaluation standard; record both the crack image and the normal image as preprocessed images;

[0027] Return to select a target temperature field until all target temperature fields have been selected.

[0028] Preferably, constructing a target temperature field model based on the preprocessed temperature data and the preprocessed temperature image includes:

[0029] Set the boundary conditions of the target temperature field;

[0030] The divergence theorem is introduced into the boundary condition through formula 1 to obtain the divergence boundary condition;

[0031]

[0032] Where F is the divergence, Fx, Fy and Fz are the direction vectors of the divergence on the x, y and z axes respectively, and l x , l y and l z are the components of the unit external normal vector of the boundary condition on the x, y and z axes, respectively, V is the temperature field space, and S is the temperature field plane;

[0033] The functional integral equation is discretized with respect to the divergence boundary condition by formula 2, and the functional minimum is solved to obtain the finite element equation of the temperature field;

[0034]

[0035] Where T is the temperature at a temperature collection location in the temperature field, [T] is the node temperature vector, F is the divergence, k is the load generated by the internal heat source, and I is the variational function.

[0036] Preferably, a crack thermal stress analysis is performed, and a correlation model is created by combining the target temperature field model and the crack thermal stress analysis; the correlation model of the target temperature field model and the crack thermal stress is recorded as a crack model, including:

[0037] Thermal stress analysis of the temperature field is performed using formula 3;

[0038] ε=α(TL-Tre) Formula 3;

[0039] Among them, ε is the thermal stress, TL is the load temperature, that is, the temperature data at the temperature collection location, and Tre is the reference temperature.

[0040] Preferably, a crack thermal stress analysis is performed, and a correlation model is created by combining the target temperature field model and the crack thermal stress analysis; the correlation model of the target temperature field model and the crack thermal stress is recorded as a crack model, and further includes:

[0041] Create a crack model;

[0042] Acquire a plurality of pre-processed image data and pre-processed temperature data;

[0043] Selecting preprocessed image data and preprocessed temperature data at the same acquisition time in the same acquisition cycle, establishing a coupling relationship between the preprocessed image data and the preprocessed temperature data at the same acquisition time; recording the coupling relationship between the preprocessed image data and the preprocessed temperature data at the same acquisition time as a training pair;

[0044] The preprocessed image data and the preprocessed temperature data are input into the crack model as a training set to train the crack model, so that the crack model continuously learns the relationship between the image data and the temperature data, and a trained crack model is obtained.

[0045] Preferably, the preprocessed image data and the preprocessed temperature data are input as training sets into the crack model to train the crack model, so that the crack model continuously learns the relationship between the image data and the temperature data to obtain a trained crack model, including:

[0046] Select a target temperature field;

[0047] Selecting an acquisition cycle of the target temperature field and acquiring a crack image and a normal image of the acquisition cycle;

[0048] All the crack images are sorted according to the order of acquisition time, and the first crack image and the temperature data corresponding to the first crack image are obtained; the first crack image is recorded as the first image, and the temperature data corresponding to the first crack image is recorded as the first temperature;

[0049] Obtain the first N normal images of the first image and the temperature data corresponding to the first N normal images according to the order of acquisition time;

[0050] The coupling relationship between the normal image and the crack image is established according to the order of acquisition time, and the normal image, the temperature data corresponding to the normal image, the first image, the first temperature, and the coupling relationship between the normal image and the crack image are used as a training sample;

[0051] Return and select an acquisition cycle of the target temperature field and obtain the crack image and the normal image of the acquisition cycle, so as to obtain multiple training samples of the target temperature field;

[0052] Divide all training samples into test sets and training sets according to random proportions;

[0053] Return to select a target temperature field until all target temperature fields are selected, thereby creating a test set and training set corresponding to each target temperature field;

[0054] Input the training samples in the training set into the crack model to train the crack model, and obtain a trained crack model; the trained crack model has the ability to predict the time node of crack generation according to the input image data and temperature data;

[0055] The test set is input into the trained crack model to determine whether the trained crack model is trained.

[0056] Preferably, real-time temperature data is collected, and the real-time temperature data is sequentially input into the crack model to obtain the prediction results output by the trained crack model, including:

[0057] Obtain real-time temperature data and real-time image data;

[0058] Input the real-time temperature data into the target temperature field model to obtain the predicted temperature output by the target temperature field;

[0059] The predicted temperature and the implemented image data are input into the crack model to obtain the predicted results output by the crack model.

[0060] Preferably, the predicted temperature and the implemented image data are input into the crack model to obtain the predicted result output by the crack model, including:

[0061] Set similarity threshold and temperature threshold;

[0062] Determine whether there is a normal image whose similarity with the real-time image data is greater than or equal to a similarity threshold;

[0063] If the similarity between the real-time image data and the normal image is greater than or equal to the similarity threshold, the first image and the first temperature corresponding to the normal image are obtained;

[0064] Determine whether the difference between the first temperature and the predicted temperature is less than the temperature threshold;

[0065] If the difference between the first temperature and the predicted temperature is less than the temperature threshold, an alarm is issued.

[0066] On the other hand, the present application also provides a real-time monitoring system for the temperature field of ultra-wide thin-walled solid high piers and a crack early warning system, including:

[0067] An acquisition component, the acquisition component comprising a temperature acquisition unit and an image acquisition unit;

[0068] A processing component, the processing component includes a processor, a data processing unit and an image processing unit, the temperature data is processed by the data unit, the temperature image is processed by the image processing unit, and the processor executes the real-time monitoring of the temperature field and crack warning method of the ultra-wide thin-walled solid high pier as described above.

[0069] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0070] By preprocessing the temperature data of the target temperature field and the image data of the target temperature field, preprocessed temperature data and a preprocessed temperature image are obtained, and then a target temperature field model is constructed based on the preprocessed temperature data and the preprocessed temperature image. A crack thermal stress analysis is performed, and a crack model is created by combining the target temperature field model with the crack thermal stress analysis. Finally, the prediction result output by the trained crack model is obtained, and it is determined whether there is a risk of cracks based on the prediction result output by the crack model. If there is a risk of cracks, an early warning is issued. The present application compares the real-time image data with the normal image through the crack model, and predicts the risk of cracks in combination with the predicted temperature of the target temperature field model and issues an early warning in time. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 It is a schematic flow chart of the real-time monitoring method of the temperature field and crack early warning method of the ultra-wide thin-wall solid high pier proposed by the present invention;

[0072] Figure 2 This is a structural schematic diagram of the real-time monitoring system for temperature field and crack warning of ultra-wide thin-walled solid high piers proposed by the present invention.

[0073] Reference numerals

[0074] 100, acquisition component; 101, temperature acquisition unit; 102, image acquisition unit;

[0075] 200, processing component; 201, processor; 202, data processing unit; 203, image processing unit. DETAILED DESCRIPTION

[0076] Embodiment 1, as Figure 1 As shown, the real-time monitoring method of the temperature field and crack early warning of the ultra-wide thin-walled solid high pier proposed by the present invention includes:

[0077] S100, setting acquisition parameters, acquiring temperature data of a target temperature field and image data of a target temperature field according to the acquisition parameters, and preprocessing the temperature data of the target temperature field and the image of the target temperature field respectively to obtain preprocessed temperature data and a preprocessed temperature image;

[0078] S200, constructing a target temperature field model based on the preprocessed temperature data and the preprocessed temperature image;

[0079] S300, performing crack thermal stress analysis, and creating a crack model by combining the target temperature field model with the crack thermal stress analysis;

[0080] S400, collecting real-time temperature data, inputting the real-time temperature data into the crack model in sequence, obtaining the prediction results output by the trained crack model, and judging whether there is a risk of cracks according to the prediction results output by the crack model, and issuing an early warning if there is a risk of cracks.

[0081] In the present invention, the temperature data of the target temperature field and the image data of the target temperature field are preprocessed to obtain preprocessed temperature data and a preprocessed temperature image, and then a target temperature field model is constructed based on the preprocessed temperature data and the preprocessed temperature image. A crack thermal stress analysis is performed, and a crack model is created by combining the target temperature field model and the crack thermal stress analysis. Finally, the prediction result output by the trained crack model is obtained, and it is determined whether there is a risk of cracks based on the prediction result output by the crack model. If there is a risk of cracks, an early warning is issued. The present application compares the real-time image data with the normal image through the crack model, and predicts the risk of cracks in combination with the predicted temperature of the target temperature field model and issues an early warning in time.

[0082] In an optional embodiment, the S100 includes:

[0083] S110, creating a temperature data table;

[0084] S120, respectively setting acquisition parameters for each target temperature field; the acquisition parameters include a temperature acquisition position, an image acquisition position, an acquisition cycle, and an acquisition frequency. Generally speaking, temperature data and image data are one-to-one corresponding, and image data and temperature data at the same acquisition frequency are recorded as a data pair;

[0085] Specifically, multiple temperature collection positions should be set for each target temperature field to ensure that complete temperature data of the target temperature field can be collected;

[0086] For a target temperature field, since the frequency and period of collecting temperature data and collecting image data are the same, the temperature data and image data of a target temperature field are one-to-one corresponding;

[0087] S130, collecting multiple temperature data of each target temperature field according to the collection parameters, and adding all the collected temperature data to the temperature data table;

[0088] S140, collecting multiple temperature images of each target temperature field according to the collection parameters, and adding all the collected temperature images to a temperature data table; the collected temperature images include crack images and normal images.

[0089] It should be noted that by setting corresponding acquisition parameters for each target temperature field and collecting temperature data and image data of the target temperature field according to the acquisition parameters, the collected data is used as the data source for creating the target temperature field model. In order to improve the reliability of the target temperature field model, the number of data collected should be increased when collecting data to ensure the comprehensiveness of the data.

[0090] For example, there are target temperature fields A, B and C. The acquisition period of target temperature field A is set to 7 days, 10 days and 15 days respectively, and the acquisition frequency of target temperature field A is set to 1 time / hour. The acquisition period of target temperature field B is set to 5 days, 10 days and 15 days respectively, and the acquisition frequency of target temperature field B is set to 1 time / hour. The acquisition period of target temperature field C is set to 5 days, 10 days and 15 days respectively, and the acquisition frequency of target temperature field C is set to 1 time / hour. After the acquisition parameters are set, the temperature data and image data collected are shown in Table-1 Temperature Data Table.

[0091] Table-1

[0092]

[0093]

[0094] In an optional embodiment, the S100 further includes:

[0095] S150, selecting a target temperature field;

[0096] S151, selecting all temperature data of each acquisition time of the target temperature field from the temperature data table;

[0097] S152, calculating the average temperature of all temperature data at each acquisition time;

[0098] S153, setting a temperature deviation threshold;

[0099] Specifically, due to the different acquisition cycles, the temperature deviation threshold can be set independently for each acquisition cycle;

[0100] S154, selecting a temperature data and determining whether the difference between the temperature data and the average temperature is greater than or equal to a temperature deviation threshold;

[0101] S155, if the difference between the temperature data and the average temperature is greater than or equal to the temperature deviation threshold, the temperature data is deleted;

[0102] Specifically, if the difference between the temperature data and the average temperature is less than the temperature deviation threshold, the temperature data is retained;

[0103] When deleting temperature data, image data in the same acquisition cycle and at the same acquisition time as the temperature data should also be deleted, thereby reducing the processing load of subsequent image data preprocessing;

[0104] S156, returning to step S154, until all temperature data of the target temperature field are selected;

[0105] S157, performing denoising processing on all temperature images in the temperature data table to obtain denoised images;

[0106] S158, performing binarization processing on the denoised image to obtain a binarized image;

[0107] Specifically, binarization is to set the grayscale value of the image pixel to 0 or 255, so that the entire image presents an obvious black and white effect. In the binarization process, the grayscale value range of the image is compressed from 0 to 255 to 0 or 255, the RGB value of the black pixel is (0,0,0), and the RGB value of the white pixel is (255,255,255);

[0108] S159, setting a crack evaluation standard, and dividing the binary image into a crack image and a normal image according to the crack evaluation standard;

[0109] Specifically, the image data of the target temperature field including cracks is the crack image;

[0110] After the crack image is determined, the temperature data in the same acquisition cycle and the same acquisition time as the crack image is recorded as the crack temperature, and the temperature data in the same acquisition cycle and the same acquisition time as the normal image is recorded as the normal temperature;

[0111] Since the target temperature field contains cracks in its image data in the later stage of crack generation, the color at the crack is generally displayed as black, so it is possible to determine whether the image data contains cracks by determining the number of black pixels contained in the binary image, so the step S159 includes:

[0112] K100 sets the crack threshold;

[0113] K101 selects a binary image and counts the number of black pixels in the binary image;

[0114] K102 determines whether the number of black pixels in the binary image is greater than or equal to a crack threshold;

[0115] K103 if the number of black pixels in the binary image is greater than or equal to the crack threshold, the binary image is recorded as a crack image;

[0116] K104: if the number of black pixels in the binary image is less than the crack threshold, the binary image is recorded as a normal image;

[0117] K105, return to step K100, until all the binary images are selected;

[0118] S160, returning to step S150, until all target temperature fields are selected;

[0119] Optionally, in order to reduce the amount of data in the temperature data table, all image data and temperature data in an acquisition period that only includes normal images may be deleted.

[0120] It should be noted that for each target temperature field, since the sensor may be damaged during the data collection process, resulting in abnormal collected temperature data, the temperature deviation threshold is set to remove abnormal points contained in the temperature data, thereby ensuring the reliability of the temperature data for subsequent user target temperature field model training.

[0121] After the preprocessing of the temperature data is completed, the image data is preprocessed so as to be divided into normal images and crack images, so as to subsequently create a crack model as a comparison group.

[0122] In an optional embodiment, the S200 includes:

[0123] S210, setting the boundary conditions of the target temperature field;

[0124] Specifically, the boundary conditions include first-type boundary conditions, second-type boundary conditions and third-type boundary conditions, the first boundary conditions are used to determine the wall temperature boundary, the second boundary conditions are used to determine the heat flux density boundary, and the third boundary conditions are used to determine the convection heat transfer boundary;

[0125] S220, introducing the divergence theorem into the boundary condition through formula 1 to obtain the divergence boundary condition;

[0126]

[0127] Where F is the divergence, Fx, Fy and Fz are the direction vectors of the divergence on the x, y and z axes respectively, and l x , l y and l z are the components of the unit external normal vector of the boundary condition on the x, y and z axes, respectively, V is the temperature field space, and S is the temperature field plane;

[0128] Specifically, the divergence theorem is one of the important theorems of vector analysis, which is directly derived from Coulomb's law. It completely relies on the inverse square law of the force between charges. Applying the divergence theorem to metal conductors in electrostatic equilibrium conditions leads to the conclusion that there is no net charge inside the conductor. Therefore, determining whether there is a net charge inside the conductor is an important method to verify Coulomb's law.

[0129] S230, discretizing the divergence boundary condition of the functional integral equation by formula 2, and solving the functional minimum to obtain the finite element equation of the temperature field;

[0130]

[0131] Where T is the temperature at a temperature collection location in the temperature field, [T] is the node temperature vector, F is the divergence, k is the load generated by the internal heat source, and I is the variation function;

[0132] Specifically,

[0133] Among them, B is the derivative matrix of the shape function matrix, N is the shape function matrix, D is the elasticity matrix, h is the convective heat transfer coefficient, V is the temperature field space, S is the temperature field plane, [F e ] is the load vector, is the thermal conductivity term, is the convective heat transfer term.

[0134] It should be noted that the heat conduction term is the load generated by the internal heat source, the convection heat transfer term is the load generated by the heat flux density, and the load vector refers to the load generated in the process of convection heat transfer;

[0135] Finite element analysis is a numerical calculation method that discretizes a continuous physical system into a finite number of units, analyzes each unit, and then combines the results to obtain an approximate solution for the overall system. This method is particularly important in the study of temperature field and thermal deformation problems, and can help reveal their internal laws. The temperature field finite element analysis uses digital simulation methods to study the changing laws of the temperature field, thereby providing a basis for the subsequent creation of a crack model.

[0136] In an optional embodiment, the S300 includes:

[0137] S310, performing thermal stress analysis on the temperature field using Formula 3;

[0138] ε=α(TL-Tre) Formula 3;

[0139] Among them, ε is the thermal stress, TL is the load temperature, that is, the temperature data at the temperature collection location, and Tre is the reference temperature;

[0140] Specifically, thermal stress refers to the stress generated when an object cannot expand or contract completely freely due to external constraints and mutual constraints between its internal parts when the temperature changes. It is precisely because of the existence of thermal stress that cracks will appear in the high pier corresponding to the target temperature field during the continuous temperature change.

[0141] It should be noted that, in general, rising temperature will cause objects to expand, while lowering temperature will cause objects to contract. If this process is repeated continuously, there is a risk of cracks in the object. Therefore, by creating and training a crack model to predict the occurrence of cracks, we can prevent them in advance.

[0142] In an optional embodiment, the step 300 further includes:

[0143] S320, creating a crack model;

[0144] S330, acquiring a plurality of pre-processed image data and pre-processed temperature data;

[0145] Specifically, in order to improve the applicability of the crack model, the preprocessed image data and preprocessed temperature data of multiple target temperature fields should be selected respectively;

[0146] S340, selecting preprocessed image data and preprocessed temperature data at the same acquisition time in the same acquisition cycle, establishing a coupling relationship between the preprocessed image data and the preprocessed temperature data at the same acquisition time; recording the coupling relationship between the preprocessed image data and the preprocessed temperature data at the same acquisition time as a training pair;

[0147] S350, inputting the preprocessed image data and the preprocessed temperature data as a training set into the crack model to train the crack model, so that the crack model continuously learns the relationship between the image data and the temperature data, and obtains a trained crack model.

[0148] It should be noted that after the crack model is created, the preprocessed temperature data and preprocessed image data are used as training samples of the crack model to train the crack model so that the crack model can predict whether cracks will occur and the predicted time node of crack occurrence based on the temperature data and image data.

[0149] In an optional embodiment, the S350 includes:

[0150] S351, selecting a target temperature field;

[0151] S352, selecting an acquisition period of the target temperature field and acquiring a crack image and a normal image of the acquisition period;

[0152] S353, sorting all the crack images according to the order of acquisition time, and obtaining the first crack image and the temperature data corresponding to the first crack image; recording the first crack image as the first image, and recording the temperature data corresponding to the first crack image as the first temperature;

[0153] S354, acquiring the first N normal images of the first image and the temperature data corresponding to the first N normal images according to the order of acquisition time;

[0154] S355, establishing a coupling relationship between a normal image and a crack image according to the sequence of acquisition time, and taking the normal image, the temperature data corresponding to the normal image, the first image, the first temperature, and the coupling relationship between the normal image and the crack image as a training sample;

[0155] S356, returning to step S352, thereby obtaining a plurality of training samples of the target temperature field;

[0156] S357, divide all training samples into test set and training set according to random proportion;

[0157] S358, returning to step S351, until all target temperature fields are selected, thereby creating a test set and a training set corresponding to each target temperature field;

[0158] S359, inputting the training samples in the training set into the crack model to train the crack model, and obtaining a trained crack model; the trained crack model has the ability to predict the time node of crack generation according to the input image data and temperature data;

[0159] S360, input the test set into the trained crack model to determine whether the trained crack model is fully trained.

[0160] It should be noted that all image data are divided into crack images and normal images through preprocessing. Therefore, the acquisition time and the first temperature of the high pier cracks corresponding to the target temperature field are found by determining the first image, and then multiple normal images before the first image and multiple normal temperatures before the first temperature are found according to the order of acquisition time, so that the crack model can learn the transformation process of normal image-first image and normal temperature-first temperature. With a sufficient number of training samples, the crack model can predict the possible cracks and the time of crack occurrence based on the input normal images and normal temperatures.

[0161] In an optional embodiment, the S400 includes:

[0162] S410, acquiring real-time temperature data and real-time image data;

[0163] S420, inputting the real-time temperature data into the target temperature field model to obtain the predicted temperature output by the target temperature field;

[0164] S430, inputting the predicted temperature and real-time image data into the crack model to obtain the predicted result output by the crack model.

[0165] It should be noted that the predicted temperature of the target temperature field is obtained by combining the target temperature field with the real-time temperature data, and then the real-time image data of the target temperature field and the predicted temperature are simultaneously input into the crack model to determine whether there is a risk of cracks.

[0166] In an optional embodiment, the S430 includes:

[0167] S431, setting a similarity threshold and a temperature threshold;

[0168] S432, determining whether there is a normal image whose similarity with the real-time image data is greater than or equal to a similarity threshold;

[0169] Specifically, when executing step S432, the normal image with the same acquisition time should be compared with the real-time image data. For example, the real-time temperature images are image A, image B and image C, and the corresponding acquisition times are 6:00 on the 1st, 7:00 on the 1st and 8:00 on the 1st, respectively. Then, when comparing, image A should be compared with the normal image at 6:00 on the 1st, image B should be compared with the normal image at 7:00 on the 1st and image C should be compared with the normal image at 8:00 on the 1st.

[0170] Since the image data has been preprocessed in the above embodiment, when executing step S432, it is only necessary to preprocess the real-time image data and compare the real-time image data with the normal image using formula 4;

[0171]

[0172] Among them, μ x is the average gray value of the normal image, μ y is the average gray value of real-time image data, σ x is the standard deviation of the normal image, σ y is the standard deviation of the real-time image data, σ xy is the covariance of the normal image and the real-time image data, C1 and C2 are constants;

[0173] S433, if the similarity between the real-time image data and the normal image is greater than or equal to the similarity threshold, obtaining the first image and the first temperature corresponding to the normal image;

[0174] Specifically, since multiple real-time image data are compared with multiple normal images one by one, the subsequent steps are performed only when the similarity between each real-time image data and the corresponding normal image is greater than or equal to the similarity threshold;

[0175] S434, determining whether the difference between the first temperature and the predicted temperature is less than a temperature threshold;

[0176] S435, if the difference between the first temperature and the predicted temperature is less than the temperature threshold, an alarm is issued;

[0177] Specifically, when an alarm is issued, the collection time corresponding to the first temperature is used as the warning time.

[0178] It should be noted that by setting a similarity threshold, the similarity between the real-time image data and the normal image in the crack model is first verified. When the similarity between the real-time image data and the normal image in the crack model is greater than or equal to the similarity threshold, then based on the transformation process from normal image to crack image, it can be concluded that there is a certain risk of cracks.

[0179] Since the generation of cracks is closely related to temperature, the possibility of cracks is judged by re-verifying the temperature threshold between the predicted temperature and the first temperature. When the difference between the first temperature and the predicted temperature is less than the temperature threshold, it can be proved that the predicted temperature is very close to the first temperature. Then, under the same temperature data, the possibility of cracks in the target temperature field is extremely high, and an early warning needs to be issued in time.

[0180] The present application sets up two stages of verification, namely the image verification stage and the temperature verification stage. Only through the two-stage verification can it be proved that there is a possibility of cracks in the target temperature field, thereby improving the accuracy of the early warning while ensuring the efficiency of the early warning.

[0181] The present application also provides a real-time monitoring system for the temperature field of an ultra-wide thin-wall solid high pier and a crack early warning system, including a collection component 100 and a processing component 200.

[0182] The acquisition component 100 includes a temperature acquisition unit 101 and an image acquisition unit 102, and the processing component 200 includes a processor 201, a data processing unit 202 and an image processing unit 203. The temperature data is processed by the data unit, and the temperature image is processed by the image processing unit 203. The processor 201 executes the real-time monitoring of the temperature field and crack warning method of the ultra-wide thin-walled solid high pier described in Example 1.

[0183] It should be noted that the temperature data of the target temperature field is collected by the temperature acquisition unit 101, and the image data of the target temperature field is collected by the image acquisition unit 102. The collected temperature data and image data are transmitted to the processing component 200, and the data is processed by each unit in the processing component 200, and a crack warning is issued in time.

[0184] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. Real-time monitoring of temperature field and crack early warning method for ultra-wide thin-walled solid high pier, characterized in that: include: Setting acquisition parameters, acquiring temperature data of the target temperature field and image data of the target temperature field according to the acquisition parameters, and preprocessing the temperature data of the target temperature field and the image of the target temperature field respectively to obtain preprocessed temperature data and a preprocessed temperature image; Constructing a target temperature field model based on preprocessed temperature data and preprocessed temperature images; Conduct crack thermal stress analysis and create a crack model by combining the target temperature field model with the crack thermal stress analysis; Collect real-time temperature data, input the real-time temperature data into the crack model in sequence, obtain the prediction results output by the trained crack model, and judge whether there is a risk of cracks based on the prediction results output by the crack model. If there is a risk of cracks, issue an early warning.

2. The method for real-time monitoring of temperature field and early warning of cracks of ultra-wide thin-walled solid high piers according to claim 1 is characterized in that: Set acquisition parameters, acquire temperature data of the target temperature field and image data of the target temperature field according to the acquisition parameters, and preprocess the temperature data of the target temperature field and the image of the target temperature field respectively to obtain preprocessed temperature data and preprocessed temperature image, including: Create a temperature data table; Set acquisition parameters for each target temperature field respectively; the acquisition parameters include temperature acquisition position, image acquisition position, acquisition cycle and acquisition frequency. Generally speaking, temperature data and image data are one-to-one corresponding, and image data and temperature data at the same acquisition frequency are recorded as a data pair; Collect multiple temperature data of each target temperature field according to the collection parameters, and add all the collected temperature data to the temperature data table; According to the acquisition parameters, multiple temperature images of each target temperature field are respectively acquired, and all the acquired temperature images are added to the temperature data table; the acquired temperature images include crack images and normal images.

3. The method for real-time monitoring of temperature field and early warning of cracks of ultra-wide thin-walled solid high piers according to claim 2 is characterized in that: The acquisition parameters are set, and the temperature data of the target temperature field and the image data of the target temperature field are acquired according to the acquisition parameters, and the temperature data of the target temperature field and the image of the target temperature field are preprocessed respectively to obtain preprocessed temperature data and preprocessed temperature image, and further comprising: Select a target temperature field; Select all temperature data of each acquisition time of the target temperature field from the temperature data table; Calculate the average temperature of all temperature data at each acquisition time; Set temperature deviation threshold; Select a temperature data and determine whether the difference between the temperature data and the average temperature is greater than or equal to the temperature deviation threshold; If the difference between the temperature data and the average temperature is greater than or equal to the temperature deviation threshold, the temperature data is deleted; Return to select a temperature data and determine whether the difference between the temperature data and the average temperature is greater than or equal to the temperature deviation threshold, until all temperature data of the target temperature field are selected; De-noising all temperature images in the temperature data table to obtain a de-noised image; Binarize the denoised image to obtain a binary image; Set a crack evaluation standard, and divide the binary image into a crack image and a normal image according to the crack evaluation standard; record both the crack image and the normal image as preprocessed images; Return to select a target temperature field until all target temperature fields have been selected.

4. The method for real-time monitoring of temperature field and early warning of cracks of ultra-wide thin-walled solid high piers according to claim 3 is characterized in that: The target temperature field model is constructed based on the preprocessed temperature data and the preprocessed temperature image, including: Set the boundary conditions of the target temperature field; The divergence theorem is introduced into the boundary condition through formula 1 to obtain the divergence boundary condition; Where F is the divergence, Fx, Fy and Fz are the direction vectors of the divergence on the x, y and z axes respectively, and l x , l y and l z are the components of the unit external normal vector of the boundary condition on the x, y and z axes, respectively, V is the temperature field space, and S is the temperature field plane; The functional integral equation is discretized with respect to the divergence boundary condition by formula 2, and the functional minimum is solved to obtain the finite element equation of the temperature field; Where T is the temperature at a temperature collection location in the temperature field, [T] is the node temperature vector, F is the divergence, k is the load generated by the internal heat source, and I is the variational function.

5. The method for real-time monitoring of temperature field and early warning of cracks of ultra-wide thin-walled solid high piers according to claim 4 is characterized in that: Conduct crack thermal stress analysis and create a correlation model by combining the target temperature field model with the crack thermal stress analysis; The correlation model between the target temperature field model and the crack thermal stress is recorded as the crack model, including: Thermal stress analysis of the temperature field is performed using Formula 3; ε=α(TL-Tre) Formula 3; Among them, ε is the thermal stress, TL is the load temperature, that is, the temperature data at the temperature collection location, and Tre is the reference temperature.

6. The method for real-time monitoring of temperature field and early warning of cracks of ultra-wide thin-walled solid high piers according to claim 5 is characterized in that: Conduct crack thermal stress analysis and create a correlation model by combining the target temperature field model with the crack thermal stress analysis; The correlation model between the target temperature field model and the crack thermal stress is recorded as the crack model, which also includes: Create a crack model; Acquire a plurality of pre-processed image data and pre-processed temperature data; Selecting preprocessed image data and preprocessed temperature data at the same acquisition time in the same acquisition cycle, establishing a coupling relationship between the preprocessed image data and the preprocessed temperature data at the same acquisition time; recording the coupling relationship between the preprocessed image data and the preprocessed temperature data at the same acquisition time as a training pair; The preprocessed image data and the preprocessed temperature data are input into the crack model as a training set to train the crack model, so that the crack model continuously learns the relationship between the image data and the temperature data, and a trained crack model is obtained.

7. The method for real-time monitoring of temperature field and early warning of cracks of ultra-wide thin-walled solid high piers according to claim 6 is characterized in that: The preprocessed image data and preprocessed temperature data are input as training sets into the crack model to train the crack model, so that the crack model continuously learns the relationship between the image data and the temperature data, and obtains the trained crack model, including: Select a target temperature field; Selecting an acquisition cycle of the target temperature field and acquiring a crack image and a normal image of the acquisition cycle; All the crack images are sorted according to the order of acquisition time, and the first crack image and the temperature data corresponding to the first crack image are obtained; the first crack image is recorded as the first image, and the temperature data corresponding to the first crack image is recorded as the first temperature; Obtain the first N normal images of the first image and the temperature data corresponding to the first N normal images according to the order of acquisition time; The coupling relationship between the normal image and the crack image is established according to the order of acquisition time, and the normal image, the temperature data corresponding to the normal image, the first image, the first temperature, and the coupling relationship between the normal image and the crack image are used as a training sample; Return and select an acquisition cycle of the target temperature field and obtain the crack image and the normal image of the acquisition cycle, so as to obtain multiple training samples of the target temperature field; Divide all training samples into test sets and training sets according to random proportions; Return to select a target temperature field until all target temperature fields are selected, thereby creating a test set and training set corresponding to each target temperature field; Input the training samples in the training set into the crack model to train the crack model, and obtain a trained crack model; the trained crack model has the ability to predict the time node of crack generation according to the input image data and temperature data; The test set is input into the trained crack model to determine whether the trained crack model is trained.

8. The method for real-time monitoring of temperature field and early warning of cracks of ultra-wide thin-walled solid high piers according to claim 7 is characterized in that: Collect real-time temperature data, input the real-time temperature data into the crack model in sequence, obtain the prediction results output by the trained crack model, and judge whether there is a risk of cracks based on the prediction results output by the crack model. If there is a risk of cracks, issue an early warning, including: Obtain real-time temperature data and real-time image data; Input the real-time temperature data into the target temperature field model to obtain the predicted temperature output by the target temperature field; The predicted temperature and the implemented image data are input into the crack model to obtain the predicted results output by the crack model.

9. The method for real-time monitoring of temperature field and early warning of cracks of ultra-wide thin-walled solid high piers according to claim 8 is characterized in that: Input the predicted temperature and the implemented image data into the crack model to obtain the predicted results output by the crack model, including: Set similarity threshold and temperature threshold; Determine whether there is a normal image whose similarity with the real-time image data is greater than or equal to a similarity threshold; If the similarity between the real-time image data and the normal image is greater than or equal to the similarity threshold, the first image and the first temperature corresponding to the normal image are obtained; Determine whether the difference between the first temperature and the predicted temperature is less than the temperature threshold; If the difference between the first temperature and the predicted temperature is less than the temperature threshold, an alarm is issued.

10. Real-time monitoring and crack warning system for ultra-wide thin-walled solid high pier temperature field, characterized by: include: An acquisition component, the acquisition component comprising a temperature acquisition unit and an image acquisition unit; A processing component, wherein the processing component includes a processor, a data processing unit and an image processing unit, wherein the temperature data is processed by the data unit, the temperature image is processed by the image processing unit, and the processor executes the real-time monitoring and crack warning method for the temperature field of an ultra-wide thin-walled solid high pier as described in any one of claims 1 to 9.

Images