A structural stress monitoring system based on pressure-sensitive paint color difference recognition

Abstract

The present application relates to pressure sensitive paint technical field, specifically to a kind of structural stress monitoring system based on pressure sensitive paint color difference identification, system includes: image analysis module, frequency identification module, rhythm extraction module, path comparison module, early warning determination module.In the present application, through the combination of image color difference gradient density analysis and physical response frequency information, the dynamic tracking of color difference abnormal area in time series and response feature extraction are realized, the sustained extraction and comparison of color evolution trend and barycenter trajectory direction are utilized, the identification precision of abnormal area path change and the determination ability of structural deviation trend are improved, the identification sensitivity to the accelerated evolution state of region is enhanced by using the synchronous analysis of frequency change and mutation point, the matching relationship between color difference response and stress evolution is optimized, the crack early warning has time continuity and spatial positioning ability by using trend sequence and direction consistency analysis, and the abnormal behavior identification effect in complex environment is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of pressure sensitive paint, in particular to a structure stress monitoring system based on pressure sensitive paint color difference identification. BACKGROUND

[0002] The technical field of pressure sensitive paint mainly involves the development and application of functional coatings with the ability to respond sensitively to changes in external pressure. Such coatings express the external physical stimuli they perceive through color changes, have a microstructure composition with physical or chemical response characteristics, and can produce visible color difference changes under specific stress conditions. The core matters of this technical field include the material composition mechanism of pressure sensitive paint, the color response mechanism, the external stress perception mechanism, and the integration method with optical imaging technology. It is widely researched and applied in engineering structure monitoring, mechanical failure warning, aerospace material testing, etc. Among them, the traditional pressure sensitive paint color difference identification structure stress monitoring system refers to a device that visually observes or statically analyzes the color difference changes of the structure surface pressure sensitive coating through manual or single image sensing device to judge the stress distribution state. The technical matters of this device are to realize the spatial identification of stress concentration or potential crack risk in the key area of the structure. It usually uses an image acquisition device to obtain coating images and judges the stress state by comparing the average values of the RGB channels in the region. Preliminary crack analysis is performed through the distribution changes of the pixel points, and the physical response data of the structure surface are measured by strain sensors to assist stress change judgment.

[0003] The traditional structure stress monitoring technology based on pressure sensitive paint relies on the static acquisition of single frame images and the average value changes of RGB channels for color difference judgment, which lacks the ability to recognize the continuous evolution trend in the time dimension. It cannot accurately capture the abnormal evolution path in the process of periodic or weak gradual stress response, and misjudgment occurs in the case of dense mutation points but insignificant average value fluctuation. It is difficult to realize the quantitative analysis of the spatial deviation between the direction trend of abnormal color difference area and the structure design model. If irregular stress concentration or early crack deviation evolution occurs in a certain area of the structure, it is difficult to effectively recognize the response through static data, causing stress recognition errors and crack trend lag response, which limits the adaptability and response sensitivity in multi-period dynamic monitoring scenarios. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art and to provide a structure stress monitoring system based on pressure sensitive paint color difference identification.

[0005] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows: a structure stress monitoring system based on pressure sensitive paint color difference identification comprises:

[0006] The image analysis module collects a paint area image, analyzes color difference gradient change conditions of each pixel and neighborhood pixels in the image, identifies pixel mutation points and establishes a multi-radius concentric ring area as a center, analyzes mutation point distribution density trends, detects color difference abnormal areas, and generates density aggregation trends;

[0007] The frequency identification module calls the density aggregation trends, acquires response data of the color difference abnormal areas by using stress and temperature sensors, records peak occurrence times as response frequency values, analyzes consistency between stress and temperature response frequencies, evaluates regional resonance response states, and establishes response frequency information;

[0008] The rhythm extraction module calls the response frequency information, identifies paint color differences in adjacent frames in a color difference abnormal area image sequence, identifies color change trends, evaluates consistency of regional color evolution trends in a time sequence, and generates evolution rhythm information;

[0009] The path comparison module calls the evolution rhythm information, constructs a barycenter point evolution path according to barycenter coordinates of mutation points in each period of the abnormal area, and compares the path with a stress path in a target structure design model, analyzes consistency and offset angles of direction change trends, and generates offset direction information.

[0010] As a further scheme of the present application, the density aggregation trends include pixel mutation point distribution density, concentric ring area scale level, and density change ratio sequence, the response frequency information includes stress response frequency, temperature response frequency, and frequency consistency state, the evolution rhythm information includes color difference change time sequence, image frame color change trend, and time rhythm consistency level, and the offset direction information specifically is color difference barycenter evolution path, theoretical stress path direction, and barycenter path offset angle.

[0011] As a further scheme of the present application, the image analysis module includes:

[0012] The pixel gradient extraction submodule collects a pressure-sensitive paint area image, acquires RGB channel values of each pixel in the image, analyzes color difference gradient change conditions by calculating RGB difference values between the pixel and neighborhood pixels, identifies pixel mutation points and records image coordinates of the mutation points, and acquires pixel mutation point distribution data;

[0013] The ring domain density calculation submodule calls the pixel mutation point distribution data, constructs a plurality of multi-radius concentric ring areas as a center of each mutation point, counts the number of mutation points in each ring area, calculates mutation point number and density of each ring area, constructs a density value sequence in ascending order of radius, and acquires a pixel mutation point ring domain density sequence;

[0014] The abnormality detection evaluation submodule analyzes the mutation point distribution density trend according to the pixel mutation point ring domain density sequence, calculates the abnormality degree score value of each region, detects the color difference abnormal region, and generates the density aggregation trend.

[0015] As a further scheme of the present application, the frequency identification module comprises:

[0016] The data acquisition submodule calls the density aggregation trend, acquires the stress response data and temperature response data in the color difference abnormal region, extracts the cycle peak point sequence recorded by the stress sensor and the cycle peak point sequence recorded by the temperature sensor in each cycle, obtains a sensor response sequence group;

[0017] The frequency extraction submodule counts the peak value number of the stress and temperature response in each cycle according to the sensor response sequence group, takes the peak value number as a response frequency value, and obtains a cycle frequency matching pair sequence;

[0018] The consistency analysis submodule calls the cycle frequency matching pair sequence, analyzes the numerical change difference of the stress frequency and the temperature frequency in each cycle, calculates a frequency synchronization convergence value, evaluates the regional resonance response state, and establishes response frequency information.

[0019] As a further scheme of the present application, the rhythm extraction module comprises:

[0020] The amplitude sequence extraction submodule calls the response frequency information, analyzes the image sequence of the color difference abnormal region, identifies the color difference values of the corresponding pixels between the continuous image frames, calculates the color difference change amplitudes between the adjacent frames, and obtains a change amplitude sequence;

[0021] The change direction analysis submodule analyzes the change direction of the color difference values between the adjacent frames based on the change amplitude sequence, analyzes the direction change situation of each pixel in the multiple frame sequences, identifies the direction state of the color change in the continuous cycles, and generates a change trend sequence;

[0022] The trend consistency evaluation submodule compares the trend directions of multiple pixels in each cycle according to the change trend sequence, evaluates the trend consistency of the regional color evolution in the time sequence, and obtains evolution rhythm information.

[0023] As a further scheme of the present application, the path comparison module comprises:

[0024] The gravity extraction submodule calls the evolution rhythm information, locates the color difference abnormal region in the continuous cycles in the pressure-sensitive paint region, extracts a mutation point set in the abnormal region in each cycle, calculates the average value of the pixel point coordinates in the set as a gravity coordinate, and generates a mutation point gravity sequence;

[0025] The path construction submodule connects the gravity center coordinates of each period in chronological order based on the gravity center sequence of the mutation points, forms a continuous gravity center point trajectory, constructs an evolution path of the gravity center points, analyzes the actual stress path in the target structure, and establishes a path fitting sequence.

[0026] The direction offset calculation submodule calls the path fitting sequence, compares with the stress path in the target structure design model, calculates the path consistency degree, analyzes the consistency and offset angle of the direction change trend, and generates offset direction information.

[0027] As a further scheme of the present application, the system further comprises:

[0028] The warning judgment module calls the offset direction information, analyzes the color difference change frequency of the color difference abnormal area in the pressure-sensitive coating area in multiple continuous periods, records and analyzes the change trend of the frequency, identifies the mutation time point of the color difference gradient in each period, identifies the color difference response accelerated evolution state of the area by analyzing the synchronism and consistency degree of the frequency change trend and the gradient mutation time point, and generates crack trend synchronization information;

[0029] The crack trend synchronization information specifically refers to the frequency increasing trend, the gradient mutation time point sequence, and the response synchronism level.

[0030] As a further scheme of the present application, the warning judgment module comprises:

[0031] The frequency analysis submodule calls the offset direction information, analyzes the color difference change frequency of the color difference abnormal area in the pressure-sensitive coating area in multiple continuous periods, extracts the frequency value according to the continuous period sequence, and obtains a color difference frequency change sequence;

[0032] The mutation extraction submodule identifies the color difference gradient of the corresponding area in each period according to the color difference frequency change sequence, detects the change inflection point and locates the occurrence period based on the curve shape of the gradient value change with the period, obtains a gradient mutation time point sequence;

[0033] The synchronization evaluation submodule calls the gradient mutation time point sequence, analyzes the synchronism and consistency degree of the frequency change trend and the gradient mutation time point by extracting the frequency amplitude rate and the mutation density in each time period, identifies the color difference response accelerated evolution state of the area, and generates crack trend synchronization information.

[0034] Compared with the prior art, the present application has the following advantages and positive effects:

[0035] In the present application, by combining image color difference gradient density analysis with physical response frequency information, dynamic tracking of color difference abnormal area in time sequence and response feature extraction are realized, the recognition accuracy of abnormal area path change and the determination ability of structure deviation trend are improved by using continuous extraction and comparison of color evolution trend and gravity trajectory direction, the recognition sensitivity to the accelerated evolution state of the region is enhanced by using synchronous analysis of frequency change and mutation point, the matching relationship between color difference response and stress evolution is optimized, the crack warning has time continuity and spatial positioning ability by using trend sequence and direction consistency analysis, the original static determination mode is expanded to a multi-cycle dynamic monitoring mechanism, and the abnormal behavior recognition effect under the multi-source information fusion in complex stress environment is improved. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 is a system flowchart of the present application;

[0037] Figure 2 is an image analysis module flowchart of the present application;

[0038] Figure 3 is a frequency identification module flowchart of the present application;

[0039] Figure 4 is a rhythm extraction module flowchart of the present application;

[0040] Figure 5 is a path comparison module flowchart of the present application;

[0041] Figure 6 is a warning determination module flowchart of the present application. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical scheme and advantages of the present application clearer and more apparent, the present application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0043] In the description of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0044] Please refer to Figure 1A structural stress monitoring system based on color difference identification of pressure-sensitive paint includes:

[0045] The image analysis module collects paint area images, analyzes the color difference gradient change of each pixel and neighborhood pixels in the image, identifies pixel mutation points and establishes a multi-radius concentric ring area as the center, analyzes the distribution density trend of the mutation points, detects the color difference abnormal area, and generates the density aggregation trend;

[0046] The frequency identification module calls the density aggregation trend, uses stress and temperature sensors to obtain response data of the color difference abnormal area, records the number of peak values as the response frequency value, analyzes the consistency between stress and temperature response frequencies, evaluates the resonance response state of the area, and establishes response frequency information;

[0047] The rhythm extraction module calls the response frequency information, identifies the color difference of the paint in adjacent frames in the color difference abnormal area image sequence, identifies the trend of color change, evaluates the trend consistency of the color evolution of the area in the time sequence, and generates evolution rhythm information;

[0048] The path comparison module calls the evolution rhythm information, constructs the evolution path of the center of gravity of the mutation points in the abnormal area according to the center of gravity coordinates in each period, and compares it with the stress path in the target structure design model, analyzes the consistency and offset angle of the trend of direction change, and generates offset direction information;

[0049] The warning judgment module calls the offset direction information, analyzes the color difference change frequency of the color difference abnormal area of the pressure-sensitive paint area in multiple consecutive periods, records and analyzes the change trend of the frequency, identifies the mutation time point of the color difference gradient in each period, and through the analysis of the synchronization and consistency degree of the frequency change trend and the gradient mutation time point, identifies the color difference response accelerated evolution state of the area, and generates crack trend synchronization information.

[0050] The density aggregation trend includes the pixel mutation point distribution density, the concentric ring area scale level, and the density change ratio sequence. The response frequency information includes the stress response frequency, the temperature response frequency, and the frequency consistency state. The evolution rhythm information includes the color difference change time sequence, the color change trend between image frames, and the time rhythm consistency level. The offset direction information specifically refers to the color difference center of gravity evolution path, the theoretical stress path direction, and the center of gravity path offset angle. The crack trend synchronization information specifically refers to the frequency increasing trend, the gradient mutation time point sequence, and the response synchronization level.

[0051] Please refer to Figure 2 , the image analysis module includes:

[0052] The pixel gradient extraction sub-module collects the pressure-sensitive paint area image, obtains the RGB channel values of each pixel in the image, analyzes the color difference gradient change by calculating the RGB difference between the pixel and the neighborhood pixels, identifies the pixel mutation point and records the image coordinates of the mutation point, and obtains the pixel mutation point distribution data.

[0053] The pressure-sensitive paint area image is collected, that is, a high-resolution digital camera is used to continuously shoot the pressure-sensitive paint area on the surface of the structure to be measured, and an RGB channel image is obtained. The image resolution is 1920x1080 pixels, and the image is stored as a PNG format file frame by frame and marked with a time stamp. A specific example is that a coating area of about 100 cm2 is arranged on a steel beam bridge structure, a high-definition camera is installed at a distance of 0.5 meters from the structure surface, and an image is taken every 1 second. A total of 1800 images are obtained by continuously shooting for 30 minutes; then the image is traversed pixel by pixel, the R, G, and B channel values (value range 0-255) of each pixel point are called, the center pixel channel value is difference operated with the corresponding RGB channel value of the neighborhood 3x3 pixels (a total of 8 neighborhood pixels) centered on the pixel, for example, for a pixel point (600, 450), the RGB channel values of the center pixel point are recorded as (125, 120, 130), the RGB channel values of the neighborhood pixels are (128, 118, 133), (127, 119, 132), (126, 117, 129), (123, 121, 131), (124, 122, 130), (122, 123, 128), (125, 119, 127), and (124, 120, 129), respectively. The RGB values of each neighborhood pixel are subtracted from the corresponding RGB values of the center pixel to obtain the differences (3, -2, 3), (2, -1, 2), (1, -3, -1), (-2, 1, 1), (-1, 2, 0), (-3, 3, -2), (0, -1, -3), and (-1, 0, -1), respectively. Further, the absolute values of the differences are summed, for example, the absolute value sum of the first neighborhood pixel difference is |3|+|-2|+|3|=8, and the absolute value sums of the other neighborhood pixels are obtained in the same way. If the absolute value sum of any neighborhood pixel is greater than a set threshold value (for example, the threshold value is set to 6, which is obtained by statistical analysis of 200 images, see Table 1), the center pixel is determined to be a mutation point and the coordinates are recorded. For example, in the above example, the absolute value sum of the first neighborhood pixel is 8, which is greater than the threshold value 6, so the center pixel point (600, 450) is recorded as a mutation point. After the traversal is completed, the coordinates of all the mutation points are obtained.

[0054] Table 1 Pixel Mutation Threshold Setting Table:

[0055] ;

[0056] As shown in Table 1, the actual test shows that the threshold value is set to 6, which can effectively distinguish the mutation points and non-mutation points, and generate the mutation point distribution data.

[0057] The ring domain density calculation submodule calls the pixel mutation point distribution data, constructs multiple sets of multi-radius concentric ring regions centered on each mutation point, counts the number of mutation points in each ring region, calculates the number and density of mutation points in each ring region, constructs a density value sequence in ascending order of radius, and obtains a pixel mutation point ring domain density sequence;

[0058] The pixel mutation point coordinate data is called to establish multiple radius concentric ring regions centered on each mutation point, for example, the radii of the concentric rings are set to 2, 4, 6, and 8 pixels, and the number of mutation points in each concentric ring region is counted, for example, the number of mutation points counted in the 2-pixel radius ring centered on the point (600, 450) is 5, and the area calculation formula is , that is, the area is about 12.57 pixels², and the density is 0.199 per pixel², and similarly, the number of 4-pixel radius ring is 10, the area is about 50.27 pixels², and the density is 0.199 per pixel², and similarly, the number of 6-pixel radius ring is 15, the area is about 113.10 pixels², and the density is 0.133 per pixel², and similarly, the number of 8-pixel radius ring is 20, the area is about 201.06 pixels², and the density is 0.099 per pixel²; the density values of each ring are calculated in turn and a density value sequence is constructed, for example, the density value sequence is 0.398, 0.199, 0.133, and 0.099, and all the density values are obtained by actually calculating the number of mutation points divided by the corresponding ring area, and the pixel mutation point ring domain density sequence is obtained in this way.

[0059] The anomaly detection evaluation submodule analyzes the trend of the distribution density of the mutation points according to the pixel mutation point ring domain density sequence, and adopts the formula:

[0060] ;

[0061] The anomaly degree score value of each region is calculated, the color difference abnormal region is detected, and the density aggregation trend is generated.

[0062] wherein, is the anomaly degree score value, and is dimensionless, is the normalized value of the mutation point density in the th concentric ring region, which is obtained by normalizing the number of mutation points in each ring region divided by the corresponding ring area, is the normalized mutation point density value of the th concentric ring region, is the standard deviation of the normalized mutation point density value, which represents the dispersion degree of the distribution of the mutation points in different ring regions, the total number of concentric ring regions, the number of the current concentric ring region, the number of the ring region;

[0063] According to the ring domain density sequence, for example, the density value sequence is 0.398, 0.199, 0.133, 0.099, and the normalized density value sequence is obtained by normalizing the sequence, i.e. dividing each value by the maximum value of the sequence (0.398), which is 1.000, 0.500, 0.334, 0.249, respectively; the standard deviation of the normalized density value is calculated , for example, the sequence is calculated as follows:

[0064] ;

[0065] Subsequently, the formula is substituted to calculate the anomaly score, and the specific formula is:

[0066] ;

[0067] The above parameters are substituted into the formula to calculate:

[0068] ;

[0069] wherein the anomaly degree score value represents the abnormal aggregation intensity of the color difference mutation distribution of a certain concentric ring structure region in the image in the multi-scale space, and the unit is a dimensionless value. The larger the value, the stronger the instability of the region in terms of color difference mutation frequency, change rate and density consistency. Reflects the evolution trend characteristics of local structure response, which is used to screen out potential micro-crack starting points or stress concentration points from the image as the core decision factor in the structure stress anomaly screening mechanism, and is used for subsequent regional priority marking and tracking update. The formula objectively quantifies the mutation point density change trend and the dispersion degree of the distribution by using the normalized density difference and the standard deviation, effectively judging the abnormal degree of the region; the results show that the abnormal degree score of the region is 0.555, which is verified by the previous experiment (see Table 2), and the anomaly score value between 0.5 and 0.7 is determined as a mild anomaly, which can be further processed as density aggregation trend data.

[0070] Table 2: Abnormal score result corresponding state table:

[0071] ;

[0072] As shown in Table 2, the region anomaly score is 0.555, indicating that the region is in a mild abnormal state, and this numerical result can be directly arranged as density aggregation trend information for subsequent frequency identification module calling.

[0073] Referring to Figure 3 , the frequency identification module comprises:

[0074] The data acquisition sub-module calls the density aggregation trend, collects stress response data and temperature response data in the color difference abnormal area, extracts the cycle peak point sequence recorded by the stress sensor and the cycle peak point sequence recorded by the temperature sensor in each cycle, and obtains a sensor response sequence group;

[0075] The density aggregation trend is called, that is, the abnormal score (such as 0.555 mild abnormal area) identified in the foregoing is used to collect data in the color difference abnormal area. In actual operation, high-sensitivity strain gauges and high-precision thermocouple sensors are arranged on the surface of the area to record the stress response data and temperature response data of the structure in real time. For example, BX120-3AA strain gauges and K-type thermocouples are arranged in the specified abnormal area of the bridge structure. In each cycle (for example, 5 minutes per cycle, and the data is collected for 30 minutes, a total of 6 cycles), the voltage signals output by the sensors are sampled and calibrated, the sampled data is converted into stress and temperature data sequences, and the numerical peak value is used as the criterion for screening. For the stress data sequence of the strain gauge, the data points are compared with each other from the starting point of the sequence, and the peak point is recorded when a data point is greater than the adjacent data points. For example, the stress response data sequence of the first cycle contains data points [8.2MPa, 8.5MPa, 8.9MPa, 8.4MPa, 8.1MPa, 8.3MPa, 8.7MPa, 8.2MPa], and the peak values screened are 8.9MPa and 8.7MPa. The temperature data sequence is screened in the same way. For example, the temperature response data sequence in the first cycle is [25.1℃, 25.3℃, 25.5℃, 25.2℃, 25.0℃, 25.4℃, 25.7℃, 25.1℃], and the peak values recorded are 25.5℃ and 25.7℃. The sensor response sequence group is stored and generated according to the cycle.

[0076] The frequency extraction sub-module extracts the frequency according to the sensor response sequence group, counts the number of stress and temperature response peaks in each cycle, and takes the number as the response frequency value to obtain a cycle frequency matching pair sequence.

[0077] The frequency is extracted according to the sensor response sequence group, and the number of peaks recorded by the sensor is counted cycle by cycle. The specific process is to count the stress peak point sequence and the temperature peak point sequence of each cycle one by one. For example, the number of stress peak points in the first cycle is 2, and the number of temperature peak points is also 2. The number of peak points of the two sensors in each cycle is matched to obtain a cycle frequency matching pair sequence. For example, the data of 6 consecutive cycles is counted to obtain Table 3:

[0078] Table 3: Cycle peak value number statistics table:

[0079] ;

[0080] As shown in Table 3, the statistical calculation of each period is completed to obtain the period frequency matching pair sequence.

[0081] The consistency analysis submodule calls the periodic frequency matching pair sequence to analyze the numerical change differences of stress frequency and temperature frequency in each cycle using the formula:

[0082] ;

[0083] Calculate the frequency synchronization convergence value, evaluate the regional resonance response state, and establish the response frequency information;

[0084] in, For the The normalized frequency value of the stress sensor in the cycle is calculated by counting the The number of response peaks of the stress sensor within a cycle is normalized by combining the maximum number of peaks in all cycles. For the The normalized frequency value of the temperature sensor in a cycle is calculated by counting the The number of temperature sensor response peaks within a cycle is normalized by combining the maximum number of peaks within all cycles. For the The difference in stress frequency within a cycle is calculated as , For the The difference in temperature frequency within a cycle is calculated as , is a sign function that outputs 1 if the input value is positive and -1 if it is negative. is the collection cycle number, is the total number of current collection cycles, is the frequency synchronization convergence value;

[0085] Call the cycle frequency matching sequence data and perform normalization based on the number of peaks in each cycle. The normalization method is based on the maximum number of peaks in 6 cycles (the maximum number of stress peaks is 4, and the maximum number of temperature peaks is 4). Taking cycle 1 as an example, the stress frequency value is , temperature frequency value , other periods are calculated in sequence, such as period 2: , , process each cycle one by one; calculate the frequency difference between adjacent cycles, such as the stress frequency difference in cycle 2 , temperature-frequency difference ; then the frequency synchronization convergence value is calculated by substituting into the consistency analysis formula:

[0086] ;

[0087] The parameters are explained as follows: represents the first cycle stress frequency normalized value, represents the first cycle temperature frequency normalized value; the difference and are adjacent cycle frequency change values, respectively representing the frequency change of stress and temperature between the current cycle and the previous cycle, and the sign function is used to determine the positive and negative of the difference product, outputting 1 when the difference product is positive, and otherwise outputting -1; the formula is calculated by substituting the data of the last 6 cycles:

[0088] ;

[0089] The frequency synchronization convergence value is a comprehensive evaluation index for measuring the degree of similarity between the stress and temperature responses in the color difference abnormal area in multiple cycles. It reflects the closeness of the two frequency values in numerical value and whether their periodic change directions are consistent, and has the ability to sensitively identify the dynamic resonance trend. The closer the value is to 1, the higher the coordination between the two responses in multiple cycles, and there may be a potential resonance or structural concentrated stress area; the closer the value is to -1, the more opposite the response directions, the weaker the regional coupling, and the more separated the structure state; a value close to 0 indicates no significant trend correlation. This parameter, as an important part of the response frequency information, can be used for subsequent key analysis steps such as hazard trend evaluation, crack induction mechanism analysis, and path deviation deduction in structural monitoring. The result shows that the frequency synchronization convergence value is -0.388, which is lower than 0, indicating that the stress and temperature response frequencies have opposite change trends in the cycle change process. This calculation result further serves as the basis for determining the response state of the region resonance, and is included in the response frequency information sequence for subsequent call by the rhythm extraction module.

[0090] Referring to Figure 4 , the rhythm extraction module includes:

[0091] The amplitude sequence extraction submodule calls the response frequency information, analyzes the image sequence of the color difference abnormal area, identifies the color difference values of corresponding pixels between consecutive image frames, calculates the color difference change amplitude between adjacent frames, and obtains the change amplitude sequence.

[0092] Call response frequency information, according to the preceding abnormal area (such as abnormal score 0.555 corresponding to mild abnormal area) to select the corresponding area image sequence of the structure surface (for example, take a frame of image per second, continuously take 10 seconds, a total of 11 frames of image), call the RGB channel value of each pixel point of the image one by one to perform comparison operation on the continuous frames, for example, taking the specific pixel point coordinate (600, 450) as an example, the RGB channel value of the first frame pixel is (120, 125, 130), and the second frame is (123, 122, 128). The absolute value of the RGB difference value is calculated and summed, that is, the calculation process is: |123-120|+|122-125|+|128-130|=3+3+2=8. The continuous frame pixel difference value operation is performed in this way and recorded, for example, the subsequent third frame (126, 123, 131) of the pixel point (600, 450) is calculated as |126-123|+|123-122|+|131-128|=3+1+3=7. The color difference change data of all continuous frames is processed in turn to construct a change amplitude sequence. For example, the color difference amplitude of the pixel (600, 450) between 10 frames is [8, 7, 9, 6, 5, 8, 7, 4, 6, 5] in turn. Finally, the change amplitude sequence of all concerned pixels in the region is obtained.

[0093] The change direction analysis submodule analyzes the change direction of the color difference value between adjacent frames based on the change amplitude sequence, analyzes the direction change of each pixel in the frame sequence, identifies the direction state of the color change in the continuous period, and generates a change trend sequence.

[0094] The change amplitude sequence is called to analyze and judge each pixel in the change direction of the color difference change amplitude value between continuous frames, for example, for the pixel at coordinate (600, 450), the color difference amplitude changes from 8 to 7 from the first frame to the second frame, and the change direction is decreasing. From the second frame to the third frame, it changes from 7 to 9, and the direction is increasing. The direction change state is judged and recorded one by one. The change direction judgment method is to compare the amplitude values of the adjacent two frames. If the amplitude of the latter frame is greater than that of the former frame, the direction is defined as “increasing”. If the amplitude of the latter frame is less than that of the former frame, the direction is defined as “decreasing”. If they are equal, it is defined as “unchanged”. For example, the specific example sequence [8→7→9→6→5→8→7→4→6→5] corresponds to the change direction [decrease→increase→decrease→decrease→increase→decrease→decrease→increase→decrease]. The direction state of all concerned pixels in multiple continuous frames is recorded in turn to generate a change trend sequence.

[0095] The trend consistency evaluation submodule compares the trend direction of multiple pixels in each period according to the change trend sequence, evaluates the trend consistency of the color evolution of the region in the time sequence, and obtains the evolution rhythm information.

[0096] The change trend sequence is called to perform consistency evaluation according to the change trend of the pixels in multiple periods (for example, 6 consecutive periods), and the analysis manner is to statistically analyze the consistency proportion of the trend direction of the pixels in each period. The specific calculation manner is to calculate the proportion of the number of times that the change directions of the selected specific number of pixels (for example, 5 typical pixels) in the region are the same in each period. For example, the change trend directions of the 5 selected typical pixels in period 1 are shown in Table 4:

[0097] Table 4: Consistency table of pixel change trend direction

[0098] ;

[0099] As shown in Table 4, in period 1, the main direction is “decrease”, and the proportion is 4 / 5=80% (4 pixels have at least 3 directions of decrease in 4 change trends, accounting for 4 / 5 of the total pixels); then, the trend direction consistency of the consecutive 6 periods is calculated in the above manner, and the consistency proportion of each period is recorded. If the consistency proportion of a period exceeds 70%, the trend is defined as relatively consistent, if 50%-70%, it is defined as moderate consistency, and if less than 50%, it is defined as trend dispersion, to obtain the evolution rhythm information of the region in the time sequence.

[0100] Please refer to Figure 5 , the path comparison module includes:

[0101] The barycenter extraction submodule calls the evolution rhythm information, locates the color difference abnormal area in the continuous period of the pressure-sensitive paint area, extracts the mutation point set in the abnormal area in each period, calculates the average value of the pixel point coordinates in the set as the barycenter coordinates, and generates the mutation point barycenter sequence;

[0102] The evolution rhythm information is called to determine the color difference abnormal area in the continuous period in the pressure-sensitive coating area, for example, taking the abnormal area (area with abnormal score 0.555) of the surface of the bridge structure as an example, the abnormal area pixel coordinates are called and recorded in the continuous 5 periods (each period lasts for 5 minutes, a total of 25 minutes). For the mutation point set in the color difference abnormal area identified in each period, the center of gravity coordinates of each period are determined by traversing all the pixel point coordinate data in the set, adding the horizontal and vertical coordinates of each pixel point respectively, and then dividing by the total number of pixel points. For example, in period 1, a total of 10 pixel mutation points are detected, with coordinates (600, 450), (602, 448), (598, 452), (601, 451), (603, 449), (599, 453), (600, 449), (602, 450), (597, 452), and (601, 448). The sum of the horizontal coordinates is 600+602+598+601+603+599+600+602+597+601=6003, and the sum of the vertical coordinates is 450+448+452+451+449+453+449+450+452+448=4502. The center of gravity coordinates (600.3, 450.2) are obtained by dividing the above coordinates by the total number 10. The center of gravity coordinates of all abnormal areas are calculated in this way for each period to generate a sequence of mutation point centers of gravity.

[0103] The path construction submodule connects the center of gravity coordinates of each period in time sequence based on the sequence of mutation point centers of gravity, forms a continuous center of gravity point trajectory, constructs an evolution path of the center of gravity point, analyzes the actual stress path in the target structure, and establishes a path fitting sequence.

[0104] Based on the sequence of mutation point centers of gravity, for example, the center of gravity coordinates obtained in period 1 to period 5 are (600.3, 450.2), (601.0, 451.1), (601.7, 452.0), (602.5, 452.7), and (603.2, 453.5), respectively. These coordinates are connected in time sequence to form a continuous center of gravity point trajectory, for example, the center of gravity connection trajectory from period 1 to period 2 is from (600.3, 450.2) to (601.0, 451.1), the connection trajectory from period 2 to period 3 is from (601.0, 451.1) to (601.7, 452.0), and so on until period 5. At the same time, the actual stress path marked in the structure design drawing (for example, the stress path coordinates of a certain area in the design drawing are (600, 450), (600.8, 450.9), (601.5, 451.7), (602.3, 452.4), and (603.0, 453.2) in 5 periods) is connected in the same way to form a design path trajectory, thereby obtaining a path fitting sequence of the actual observation and the design structure.

[0105] The direction offset calculation submodule calls the established path fitting sequence and compares it with the stress path in the target structure design model using the formula:

[0106] ;

[0107] Calculate the degree of path consistency, analyze the consistency of direction change trends and offset angles, and generate offset direction information;

[0108] in, is the degree of path consistency, is the number of path segments involved in the comparison, The cycle index number of the path segment is used to identify the The path vector within a period, For the The direction vector of the design model path within a cycle is obtained by extracting the stress direction coordinate sequence of the cycle in the target structure design model and performing vector processing. For the The direction vector of the actual path within a cycle is calculated by the coordinate difference of the center of gravity of the chromatic aberration mutation point in consecutive cycles. For the mid-cycle and The angle between the two vectors is obtained by calculating the inverse cosine function of the angle between the two vectors. is the reference threshold of the path deviation angle, which is obtained by setting the upper limit of the allowable deviation angle set in the design standard;

[0109] Call the path fitting sequence and compare the continuous path segments one by one to calculate the path consistency. With design model path vector Calculate for input, for example, the actual path vector of the first cycle is (601.0-600.3,451.1-450.2)=(0.7,0.9), the designed path vector is (600.8-600,450.9-450)=(0.8,0.9), the angle between the two vectors is The calculation process is:

[0110] ;

[0111] Similarly, the vector and angle of each periodic path segment are calculated to obtain the specific data in Table 5:

[0112] Table 5 Path segment angle calculation results:

[0113] ;

[0114] As shown in Table 5, the degree of path consistency is calculated by substituting the formula:

[0115] ;

[0116] Substitute the angle between the specific example path segment into the calculation (reference threshold 10°):

[0117] ;

[0118] Wherein, the degree of path consistency represents the consistency of the structural stress evolution path and the design target path direction change trend in a plurality of continuous cycles. The closer the value is to 1, the higher the actual path and the target path direction match, and the expansion trend of the crack or abnormal area is stable and controlled; the lower the value, the more obvious the path deviation, which may indicate that the local stress state of the structure changes or the abnormal expansion trend. The index can be used as an important reference parameter for subsequent structure state evaluation, deviation diagnosis and early warning decision, and provides a quantitative basis for directional abnormal evolution under the color difference recognition mechanism of pressure-sensitive paint. The calculation result 0.797 indicates that the consistency of the actual observed path and the design path is high (according to the degree of path consistency, more than 0.7 is high consistency, 0.5-0.7 is medium consistency, and less than 0.5 is low consistency), which means that the direction change trend between the actual gravity evolution path and the design path is consistent, and there is a certain deviation angle, which generates deviation direction information for further processing by the subsequent early warning judgment module.

[0119] Please refer to Figure 6 , the early warning judgment module includes:

[0120] The frequency analysis submodule calls the deviation direction information, analyzes the color difference change frequency of the color difference abnormal area in the pressure-sensitive paint area in a plurality of continuous cycles, extracts the frequency value according to the continuous cycle sequence, and obtains the color difference frequency change sequence;

[0121] After calling the offset direction information, the system obtains the period path offset angle and direction change trend data identified in the previous stage, and locates the corresponding color difference abnormal area in the coating image on this basis, and then takes the area as the target to analyze the image sequence cycle by cycle, and calculates the color difference response frequency in each cycle. Specifically, the extraction of color difference change frequency is based on the number of frame-by-frame color difference mutation points in the image frame sequence in each cycle. For example, assuming that the image in each cycle is one frame per second, and a total of 30 frames are collected, the system identifies the color difference mutation peak point by analyzing the fluctuation of the number of mutation pixel points in the abnormal area of each image, and takes the number of mutations in each cycle as the basis for frequency calculation. For example, in the frame sequence in the first cycle, when the number of mutation pixels reaches a local maximum (such as greater than the previous and next frames), it is recorded as a peak value. If a total of 4 local extreme points are detected, the frequency is 4. Repeat the above operation to extract the frequency in the next 5 consecutive cycles, and obtain the frequency change sequence, such as [4, 5, 7, 6, 9, 11]. This sequence indicates that in the color difference abnormal area, as the cycle advances, the frequency of mutation occurrence is increasing. All frequency values belong to a continuous cycle sequence, forming a complete color difference frequency change sequence, which serves as a quantitative input basis for subsequent mutation behavior.

[0122] The mutation extraction submodule identifies the color difference gradient of the corresponding area in each cycle according to the color difference frequency change sequence, detects the change inflection point based on the curve shape of the gradient value change with the cycle, and locates the occurrence cycle to obtain the gradient mutation time point sequence.

[0123] Based on the color difference frequency change sequence obtained in the previous step, the system synchronously extracts the color difference gradient value of each cycle abnormal area in the corresponding cycle. The gradient calculation is based on the difference between the maximum and minimum color difference values in each cycle divided by the frame distance, for example, in the first cycle, the maximum color difference value of the abnormal area is 80, the minimum value is 50, and the frame distance is 30 frames, then the gradient value of this cycle is (80-50) / 30≈1.0. Arrange all the color difference gradient values in order, for example, [1.0, 1.2, 1.8, 1.6, 2.4, 3.1], and then the system analyzes the change trend of the gradient sequence, extracts the curve trend using the first-order difference method, and calculates the increment value of the gradient of adjacent cycles, for example, the second cycle increases by 0.2 compared to the first cycle, the third cycle increases by 0.6 compared to the second cycle, the fourth cycle decreases by 0.2, the fifth cycle increases by 0.8, and the sixth cycle increases by 0.7. The obtained increment sequence is [+0.2, +0.6, -0.2, +0.8, +0.7], and further judge the gradient change inflection point position, that is, the cycle point where the increment and decrement symbols change is the mutation occurrence cycle. In this example, the third cycle changes from positive increment to negative, and the fourth cycle changes from negative to positive, so the third and fourth cycles are judged as change inflection points, corresponding to mutation points, and the final gradient mutation time point sequence is [3, 4].

[0124] The synchronous evaluation sub-module calls the gradient mutation time point sequence, analyzes the frequency change trend and the degree of synchronization and consistency of the gradient mutation time point by extracting the frequency increase rate and the mutation density in each time period, identifies the color difference response accelerated evolution state of the region, and generates crack trend synchronization information;

[0125] The gradient mutation time point sequence in the previous step is called, and the system analyzes it in combination with the color difference frequency sequence in the corresponding time period to evaluate its synchronization and consistency. The specific operation is as follows: first, the frequency increase rate before and after the mutation point is calculated, for example, the frequency of the 3rd period is 7, and the frequency of the 2nd period is 5, then the increase rate is (7-5) / 5=0.4; at the same time, the frequency of the 4th period is 6, and the increase rate is (6-7) / 7≈-0.14, then the mutation density in the surrounding ±1 period of the mutation period is counted respectively, for example, the number of mutation points in the 3 periods before and after the 3rd period is [0, 1, 1], and the density is 2 / 3≈0.67. The frequency increase rate and the mutation density corresponding to each mutation time point are combined and analyzed, if the two are synchronous and consistent in change direction in multiple periods (such as the mutation point increases and the frequency accelerates, or the mutation point decreases and the frequency slows down), then it is defined as "synchronous and consistent". If the mutation density is high but the frequency increase rate is negative or no increase, then it is "asynchronous". In this way, the proportion of synchronous and consistent periods in the whole period is counted, for example, 3 times are synchronous and consistent in 6 periods, and the proportion is 50%. If the proportion exceeds 60%, it is judged as "accelerated evolution", otherwise it is considered as a stable or fluctuating state. In this example, it is slightly lower than the threshold value, so it does not constitute an accelerated evolution state, only the "local synchronous enhancement trend" is recorded, and the corresponding crack trend synchronization information sequence is generated to provide input parameters for the subsequent structure warning model.

[0126] The above is only a preferred embodiment of the present application, and does not limit the form of the present application in other forms. Any skilled person in the art can use the disclosed technical content to make changes or modifications to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made according to the technical essence of the present application to the above embodiments without departing from the technical solution content of the present application still belongs to the protection scope of the present application.

Claims

1. A structural stress monitoring system based on color difference recognition of pressure-sensitive coatings, characterized in that: The system comprises: The image analysis module collects images of the paint area, analyzes the color gradient changes of each pixel and its neighboring pixels in the image, identifies pixel mutation points and uses them as the center to establish multi-radius concentric ring areas, analyzes the distribution density trend of mutation points, detects color abnormality areas, and generates density aggregation trends; The frequency identification module uses the density aggregation trend to obtain response data of the color difference abnormal area using stress and temperature sensors, records the number of peak occurrences as the response frequency value, analyzes the consistency between the stress and temperature response frequencies, evaluates the regional resonance response state, and establishes response frequency information; The rhythm extraction module calls the response frequency information, identifies the paint color difference in adjacent frames in the image sequence of the color difference abnormal area, identifies the trend of color change, evaluates the trend consistency of regional color evolution in the time series, and generates evolution rhythm information; The path comparison module calls the evolution rhythm information, constructs the center of gravity evolution path based on the coordinates of the center of gravity of the mutation point in the abnormal area in each cycle, and compares it with the stress path in the target structure design model, analyzes the consistency of the direction change trend and the offset angle, and generates offset direction information; The density aggregation trend includes the distribution density of pixel mutation points, the scale level of concentric ring areas, and the density change ratio sequence; the response frequency information includes the stress response frequency, the temperature response frequency, and the frequency consistency state; the evolution rhythm information includes the color difference change time series, the color change trend between image frames, and the time rhythm consistency level; the offset direction information specifically includes the color difference center of gravity evolution path, the theoretical stress path direction, and the center of gravity path offset angle.

2. The structural stress monitoring system based on color difference recognition of pressure-sensitive coatings according to claim 1 is characterized in that: The image analysis module includes: The pixel gradient extraction submodule collects images of the pressure-sensitive coating area, obtains the RGB channel value of each pixel in the image, calculates the RGB difference between the pixel point and the neighboring pixels, analyzes the color gradient change, identifies the pixel mutation point and records the image coordinates of the mutation point, and obtains the pixel mutation point distribution data; The ring density calculation submodule calls the pixel mutation point distribution data, constructs multiple groups of multi-radius concentric ring areas with each mutation point as the center, counts the number of mutation points in each ring area, calculates the number and density of mutation points in each ring area, constructs a density value sequence in increasing radius order, and obtains the pixel mutation point ring density sequence; The anomaly detection and evaluation submodule analyzes the distribution density trend of the mutation points based on the pixel mutation point ring density sequence, calculates the abnormality score value of each area, detects the color difference abnormal area, and generates the density aggregation trend.

3. The structural stress monitoring system based on color difference recognition of pressure-sensitive coatings according to claim 2 is characterized in that: The frequency identification module includes: The data acquisition submodule calls the density aggregation trend, collects stress response data and temperature response data in the color difference abnormal area, extracts the cycle peak point sequence recorded by the stress sensor and the cycle peak point sequence recorded by the temperature sensor in each cycle, and obtains a sensor response sequence group; The frequency extraction submodule counts the peak values ​​of stress and temperature responses in each cycle according to the sensor response sequence group and uses the peak values ​​as the response frequency values ​​to obtain a cycle-frequency matching pair sequence; The consistency analysis submodule calls the periodic frequency matching pair sequence, analyzes the numerical variation difference of stress frequency and temperature frequency in each period, calculates the frequency synchronization convergence value, evaluates the regional resonance response state, and establishes the response frequency information.

4. The structural stress monitoring system based on color difference recognition of pressure-sensitive coatings according to claim 3 is characterized in that: The rhythm extraction module includes: The amplitude sequence extraction submodule calls the response frequency information, analyzes the image sequence of the color difference abnormal area, identifies the color difference value of the corresponding pixel between consecutive image frames, calculates the color difference change amplitude between adjacent frames, and obtains the change amplitude sequence; The change direction analysis submodule analyzes the change direction of the color difference values ​​between adjacent frames based on the change amplitude sequence, analyzes the direction change of each pixel in multiple frame sequences, identifies the direction state of color change in continuous cycles, and generates a change trend sequence; The trend consistency evaluation submodule compares the trend directions of multiple pixels in each cycle according to the change trend sequence, evaluates the trend consistency of regional color evolution in the time series, and obtains evolution rhythm information.

5. The structural stress monitoring system based on color difference recognition of pressure-sensitive coatings according to claim 4 is characterized in that: The path comparison module includes: The center of gravity extraction submodule uses the evolution rhythm information to locate the color difference abnormal area of ​​the pressure-sensitive coating area in consecutive cycles, extracts the set of mutation points in the abnormal area in each cycle, calculates the average value of the pixel coordinates in the set and uses it as the center of gravity coordinate to generate a sequence of the center of gravity of the mutation points; The path construction submodule connects the center of gravity coordinates of each period in chronological order based on the mutation point center of gravity sequence to form a continuous center of gravity trajectory, constructs the center of gravity evolution path, analyzes the actual stress path in the target structure, and establishes a path fitting sequence; The direction offset calculation submodule calls the established path fitting sequence, calculates the path consistency by comparing it with the stress path in the target structure design model, analyzes the consistency of the direction change trend and the offset angle, and generates offset direction information.

6. The structural stress monitoring system based on color difference recognition of pressure-sensitive coatings according to claim 1 is characterized in that: The system also includes: The early warning judgment module calls the offset direction information, analyzes the color difference change frequency of the abnormal color difference area of ​​the pressure-sensitive coating area within multiple consecutive cycles, records and analyzes the frequency change trend, identifies the mutation time point of the color difference gradient within each cycle, and identifies the accelerated evolution state of the color difference response of the area by analyzing the synchronization and consistency of the frequency change trend and the gradient mutation time point, thereby generating crack trend synchronization information; The crack trend synchronization information specifically refers to the frequency increasing trend, the gradient mutation time point sequence, and the response synchronization level.

7. The structural stress monitoring system based on color difference recognition of pressure-sensitive coatings according to claim 6 is characterized in that: The early warning determination module includes: The frequency analysis submodule calls the offset direction information to analyze the color difference change frequency of the color difference abnormal area in the pressure-sensitive coating area within multiple continuous cycles, extracts the frequency value according to the continuous cycle sequence, and obtains the color difference frequency change sequence; The mutation extraction submodule identifies the color difference gradient of the corresponding area in each cycle according to the color difference frequency change sequence, detects the inflection point of the change and locates the occurrence cycle based on the curve shape of the gradient value changing with the cycle, and obtains the gradient mutation time point sequence; The synchronous evaluation submodule calls the gradient mutation time point sequence, extracts the frequency increase rate and mutation density in each time period, analyzes the synchronization and consistency of the frequency change trend and the gradient mutation time point, identifies the accelerated evolution state of the color difference response of the region, and generates crack trend synchronization information.

Images