Crack state recognition method, system and device based on multi-source physical data fusion

Abstract

The application belongs to the field of electric digital data processing, and provides a crack state recognition method, system and equipment based on multi-source physical data fusion, which can normalize multi-physical quantity data of each sampling time into corresponding multi-physical quantity normalized feature vectors, and then select a neighborhood constraint range. For each sampling time, the distance values between the multi-physical quantity normalized feature vectors of each sampling time in the neighborhood constraint range are calculated, and the distance values are arranged in a neighborhood distance array according to the sequence. Based on the neighborhood distance array, a mutation intensity index and a trend intensity index are constructed. Within the neighborhood constraint range, layered threshold intervals can be generated according to the mutation intensity index and the trend intensity index, respectively, and the crack state can be recognized in stages according to the layered threshold intervals, and the results of the stage recognition are output, so as to improve the accuracy, stability and engineering usability of crack state recognition.

Description

Technical Field

[0001] This invention belongs to the field of electrical digital data processing, and specifically relates to a crack state identification method, system and device based on multi-source physical data fusion. Background Technology

[0002] Concrete structures are widely used in bridges, tunnels, rail transit, nuclear power plant containment structures, dams, and building construction. Cracks, as one of the most common and earliest forms of damage in concrete structures, exhibit abnormal opening and closing, misalignment, accompanying changes in orientation (e.g., tilt angle, deflection), and vibration response, often related to factors such as load changes, temperature shrinkage and creep, material degradation, and uneven foundation settlement. During engineering operation and maintenance, when cracks evolve from stable to continuously expanding or undergo sudden changes, timely identification and measures such as load limiting, increased inspection frequency, or repair and reinforcement are often necessary. Therefore, there is a widespread need in the industry for long-term online monitoring and automatic identification of crack conditions.

[0003] However, in linear engineering projects such as railway tracks and bridges, crack monitoring has long relied on manual inspections during designated windows, typically using tools like calipers for on-site measurements and manual recording. This method is susceptible to factors such as personnel experience, errors from repeated measurement point positioning, and oversights in recording, making it difficult to meet the needs of high-frequency, long-term, and traceable condition assessments. For example, patent document CN109425289A points out that current monitoring methods for cracks in track slabs and cement foundations primarily rely on manual monitoring. Measurement accuracy varies with each instance, and the numerous, tedious measurements are prone to errors and extremely slow. The document also describes engineering requirements for issuing warnings / alarms when cracks reach specific thresholds, such as a 3mm warning or a 6mm alarm threshold. It is evident that relying primarily on manual inspections and single-point measurements results in insufficient efficiency and consistency.

[0004] In existing automatic monitoring solutions, a large number of technologies focus on the precise measurement and long-term deployment of crack width. For example, patent document CN104457549B provides an automatic crack width monitoring device, aiming to achieve precise, automatic, and long-term monitoring of crack width in concrete structures. However, it also points out that commonly used vibrating wire strain sensors and capacitive displacement gauges in engineering are not dedicated crack monitoring devices. Crack width measurement results are affected by the temperature deformation of the concrete surface, and exposed mechanisms are prone to corrosion in rainy environments, making long-term stability and accuracy difficult to guarantee. Therefore, while there are many automatic monitoring devices that use crack width / displacement as a single core quantity, they are easily affected by environmental and structural deformations and struggle to distinguish crack stages. While these types of solutions can address the issues of whether cracks exist and their width in engineering, the technical approach, which can be directly observed from the device structure and monitoring targets, typically relies on a single quantity (e.g., width or displacement) for threshold or trend judgment. Consequently, when multiple sources of disturbance, such as temperature, structural attitude, and vibration loads, are superimposed, it is often difficult to further distinguish stable stages, continuous expansion stages, and rapid change stages of state evolution.

[0005] Another approach attempts to utilize temperature fields or distributed optical fibers for crack monitoring and location. For example, patent document CN110672657B proposes a temperature tracing system and monitoring method for crack monitoring in concrete structures. It lists several issues: multiple cracks or excessive loss in the same optical path may lead to blind spots in the monitoring section; to ensure sensitivity, only unarmored optical fibers can be used, resulting in low mechanical strength and difficulty in guaranteeing construction success rate; crack width calculation requires knowing the angle between the crack and the optical fiber, but crack propagation is uncertain and the crack surface is not planar, making accurate angle determination difficult and width calculation unreliable. Distributed optical fiber / temperature tracing solutions can cover a large area, but they suffer from blind spots, construction success rate issues, and reliability problems in width inversion due to geometric uncertainties. Therefore, while these solutions have advantages in coverage and line maintainability, they still have limitations in quantitative inversion under complex working conditions and engineering feasibility. Furthermore, they typically rely on a single physical field, such as temperature / optical fiber signals, making it difficult to directly form multi-source fusion criteria for crack state stages.

[0006] In summary, relying solely on crack width or displacement thresholds often only yields a simple judgment of whether the threshold has been exceeded or not, making it difficult to reliably distinguish between rapid changes caused by short-term impacts and expansion caused by long-term slow accumulation. It also struggles to provide consistent conclusions under multi-source disturbances. Fixed thresholds are ill-suited to adapting to shifting operating conditions. Temperature expansion and contraction, creep contraction, and slow structural attitude shifts alter the baseline and fluctuation levels. Fixed thresholds, such as those used for early warning / alarm based on a specific crack value, are prone to false alarms or missed alarms under different seasons, structures, and installation conditions. Existing patents also explicitly mention that crack measurements are affected by factors such as temperature deformation. Furthermore, common engineering monitoring issues include sensor transient spikes, communication packet loss corrections, and short-term impact vibrations that create outliers. If thresholds are generated using statistical methods like mean and standard deviation, these outliers can easily skew the threshold, causing distortion and instability. This problem is even more pronounced with multi-source data. Furthermore, existing solutions often either favor a single sensing mechanism, such as width, tilt angle, or temperature tracer, or tend to focus on engineering implementation that involves first collecting data, then uploading it, and finally comparing it to trigger alarms. They lack a fusion mechanism that can uniformly characterize the intensity of multi-source changes within the temporal neighborhood and map it to outputs in stages such as stable, expanding, and rapidly changing phases. Summary of the Invention

[0007] The purpose of this invention is to propose a crack state identification method, system, and device based on multi-source physical data fusion, so as to solve one or more technical problems existing in the prior art, and at least provide a beneficial option or create conditions.

[0008] To achieve the above objectives, according to one aspect of the present invention, a crack state identification method based on multi-source physical data fusion is provided, the method comprising the following steps:

[0009] Acquire multivariate physical quantity data of cracks at multiple sampling times, wherein the multivariate physical quantity data includes at least one or more of crack width, crack misalignment, structural tilt angle, structural vibration intensity, and ambient temperature;

[0010] The multivariate physical quantity data at each sampling time are converted into the corresponding multivariate physical quantity normalized feature vectors.

[0011] Select a neighborhood constraint range, and for each sampling time, calculate the distance value between the normalized feature vectors of each multivariate physical quantity relative to each sampling time within the neighborhood constraint range. Then, arrange the distance values ​​in order to form a neighborhood distance array A.

[0012] Based on the neighborhood distance array A, a mutation intensity index B_jump and a trend intensity index B_trend are constructed. The mutation intensity index B_jump represents the degree of sudden increase in the distance value between the current sampling time and the adjacent sampling time within the neighborhood constraint range. The trend intensity index B_trend represents the degree of asymmetry in the distance value between the sampling time before and after the current sampling time within the neighborhood constraint range.

[0013] Within the neighborhood constraint, hierarchical threshold intervals are generated for the mutation intensity index B_jump and the trend intensity index B_trend, respectively. Based on the hierarchical threshold intervals, the crack state is identified in stages, and the results of the stage identification are output.

[0014] In existing technologies, crack states are generally classified into three types: active cracks, developing cracks, and static cracks. In some embodiments, crack states are identified in stages based on the layered threshold intervals. Correspondingly, the results of stage identification may include the crack's rapid change stage, crack expansion stage, and crack stability stage. The distance value is preferably Euclidean distance. Compared with previous methods that only set fixed thresholds for single physical quantities such as crack width or dip angle, or that independently alarm for multiple physical quantities, the method of the present invention can uniformly convert multiple physical quantities into normalized feature vectors, calculate Euclidean distances within the neighborhood constraint range to form a neighborhood distance array A, further construct a sudden change intensity index B_jump and a trend intensity index B_trend from A, and generate layered threshold intervals based on neighborhood statistics to output the identification results of stable, expanding, and rapid change stages. This scheme solves the technical problems in existing technologies where single-variable thresholds are difficult to distinguish between rapid crack changes and slow expansion, fixed thresholds are difficult to adapt to temperature and load drift, and multi-source alarm results are prone to conflict and lack unified criteria. It achieves a fast, stable, and interpretable stage identification effect for crack states under multiple operating conditions. Among them, the normalized vector enables physical quantities of different dimensions to be integrated on the same scale, the neighborhood Euclidean distance compresses multi-source changes into a unified geometric difference measure, and the neighborhood distance array preserves the local temporal structure of nearest and far neighbors. Thus, B_jump can reflect local discontinuous abrupt changes, and B_trend can reflect slow evolution that is asymmetrical before and after. Combined with the hierarchical threshold, it can effectively distinguish between random fluctuations, continuous development and sudden changes.

[0015] Furthermore, the neighborhood constraint range is preferably 2, and the neighborhood distance array A is constructed into four elements according to the position order of adjacent sampling times, which respectively correspond to the Euclidean distance between the sampling times of the two preceding times, the sampling time of the previous time, the sampling time of the next time, and the sampling times of the two following times.

[0016] Furthermore, regarding the neighborhood constraint range, when the sampling time is located at the sequence boundary, resulting in insufficient forward or backward neighborhood, one or more of the following methods are employed to ensure that the length of the neighborhood distance array A remains consistent: mirroring, nearest-neighbor replication, or missing bit weight reduction. In some embodiments, the missing bit weight reduction algorithm may preferably include feature selection methods, weight reduction, and dimensionality reduction methods for missing bits, such as missing value ratio, low variance filtering, forward and backward feature elimination, etc.

[0017] Furthermore, the mutation intensity index B_jump is determined by two distance values ​​in the neighborhood distance array A corresponding to adjacent sampling times, and the mutation intensity index B_jump is the maximum or average of the two distance values.

[0018] Compared to previous methods that directly triggered alarms based on crack width increments or single-moment changes, this invention determines the mutation intensity index B_jump by using two distance values ​​from the neighborhood distance array A corresponding to adjacent sampling times, and takes the maximum or average value as the mutation intensity. This scheme solves the technical problems of existing technologies that rely solely on single-sided increments and are susceptible to noise spikes, and that mutations occurring on different sides can lead to missed detections or unstable localization. It achieves greater sensitivity to crack transient changes and more stable localization of mutation moments. Specifically, rapid changes cause a significant amplification of the differences between the multi-source feature vectors of the current state and adjacent states in nearby moments. Using two adjacent distances can simultaneously cover cases where mutations occur on the front or back side. Taking the maximum value captures the strongest mutations, while taking the average value suppresses occasional noise at single points, thereby improving the robustness and stability of mutation identification.

[0019] Furthermore, the trend strength index B_trend is determined by the mean of the forward neighborhood distance and the mean of the backward neighborhood distance, and the trend strength index B_trend is the absolute value of the difference between the mean of the forward neighborhood distance and the mean of the backward neighborhood distance.

[0020] The forward neighborhood distance includes the distance between the sampling time and each sampling time before it within the neighborhood constraint range, and the backward neighborhood distance includes the distance between the sampling time and each sampling time after it within the neighborhood constraint range.

[0021] Compared to previous methods that used univariate trend lines or long-window mean drift to determine crack development, the method described in this invention uses a trend intensity index, B_trend, determined by the mean of the forward and backward neighborhood distances, and takes the absolute value of the difference between the two to characterize the asymmetry between the forward and backward distances. This method solves the technical problems in existing technologies, such as slow expansion being difficult to identify in a timely manner due to small changes in short time, periodic or symmetrical fluctuations being easily misjudged as development trends, and long-window discrimination causing response delays. It achieves a sensitive identification effect of the continuous development trend of cracks within a shorter neighborhood and reduces false alarms caused by symmetrical fluctuations. Specifically, when crack expansion exhibits unidirectional evolution, the similarity between the current moment and the past and future neighborhoods will show systematic differences. This difference can be directly quantified by the asymmetry of the mean forward and backward distances. In contrast, during symmetrical perturbations, the mean values ​​tend to be consistent, thus preventing B_trend from increasing abnormally, thereby effectively distinguishing between slow expansion and symmetrical fluctuations.

[0022] Furthermore, the threshold in the stratified threshold interval is determined based on the median and / or dispersion statistics of the corresponding indicators (corresponding to the mutation intensity indicator B_jump and the trend intensity indicator B_trend, respectively) within the neighborhood constraint range. The dispersion statistics are either the absolute deviation of the median or the interquartile range. The stratified threshold interval is divided according to the threshold.

[0023] Compared to previous methods that used fixed thresholds or thresholds determined by the mean and standard deviation, this method determines the thresholds in its tiered threshold intervals by using the median and dispersion statistics of the corresponding indicators within the neighborhood constraint range. The dispersion statistics are the median absolute deviation or interquartile range, and the tiered threshold intervals are divided based on the threshold values. This approach solves the technical problems in existing technologies, such as the difficulty of fixed thresholds adapting to changes in temperature, load, and noise levels leading to operating condition drift; the susceptibility of the mean and standard deviation statistics to outliers or shock data causing threshold distortion; and the tendency for state fluctuations to occur near the threshold. It achieves a stage-based discrimination effect that is adaptive to thresholds and less sensitive to outliers, and can achieve progressive output of early warnings and alarms through tiered classification. Specifically, the median feature calculation can stably represent the typical level within the neighborhood, and the interquartile range or median absolute deviation can stably represent the typical fluctuation amplitude without being easily affected by a few outliers. This allows the threshold to adaptively adjust with background fluctuations. Simultaneously, the tiered thresholds provide different thresholds for different significance levels, thereby improving the robustness and engineering usability of the identification.

[0024] Further, the stratified threshold interval is determined as follows: the median and dispersion statistics of the corresponding indicators within the neighborhood constraint range are weighted and combined to form the corresponding thresholds, and the mutation intensity index B_jump or trend intensity index B_trend at the current sampling time is compared with the corresponding threshold to determine whether it is significantly abnormal; preferably, exceeding the threshold is determined to be significantly abnormal.

[0025] Furthermore, in the stage identification of crack state based on the aforementioned layered threshold interval, the criteria for the abrupt change stage of the crack include at least the following: the abrupt change intensity index B_jump at the current sampling time exceeds its corresponding threshold, and the trend intensity index B_trend exceeds its corresponding threshold within the neighborhood constraint range. For example, at the current sampling time, the abrupt change intensity index B_jump exceeds the abrupt change intensity index threshold, and the trend intensity index B_trend at the two sampling times before and after the current sampling time also exceeds the trend intensity index threshold.

[0026] Furthermore, in the stage identification of crack state based on the layered threshold interval, the criteria for the crack expansion stage include at least: the mutation intensity index B_jump at the current sampling time does not exceed its corresponding threshold, while the trend intensity index B_trend at the current sampling time exceeds its corresponding threshold, and the trend intensity index B_trend continues to exceed its corresponding threshold in subsequent consecutive sampling times, and the number of subsequent consecutive sampling times is not less than the number of sampling times contained in the neighborhood constraint range.

[0027] For example, the trend strength index B_trend continuously exceeds the neighborhood robustness threshold for K consecutive sampling times, where K can be an integer from 3 to 10; preferably, when the neighborhood constraint range of the current sampling time is 2, then without including the previous sampling time, the number of sampling times included in the neighborhood constraint range is 4, that is, K should be no less than 4.

[0028] Furthermore, in the stage identification of crack state based on the layered threshold interval, the criteria for the stable stage of crack include at least the following: the mutation intensity index B_jump and the trend intensity index B_trend at the current sampling time do not exceed the corresponding threshold, and the mutation intensity index B_jump remains below its corresponding threshold in subsequent consecutive sampling times.

[0029] This invention also provides a crack state identification system based on multi-source physical data fusion. The crack state identification system includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the crack state identification method based on multi-source physical data fusion. The crack state identification system based on multi-source physical data fusion can run on computing devices such as desktop computers, laptops, handheld computers, and cloud data centers. The runnable system may include, but is not limited to, processors, memory, and server clusters. The processor executes the computer program within the following system units:

[0030] The data acquisition unit is used to acquire multivariate physical quantity data of the crack at multiple sampling times and convert the multivariate physical quantity data at each sampling time into the corresponding multivariate physical quantity normalized feature vector.

[0031] The constraint selection unit is used to select the neighborhood constraint range R. For each sampling time, it determines its forward neighborhood sampling time set and backward neighborhood sampling time set, wherein the forward neighborhood sampling time set includes the R sampling times before the sampling time, and the backward neighborhood sampling time set includes the R sampling times after the sampling time.

[0032] The neighborhood calculation unit is used to calculate the Euclidean distance between the normalized feature vector of the multivariate physical quantity at the sampling time and the normalized feature vector of the multivariate physical quantity corresponding to each sampling time in the forward neighborhood sampling time set and the backward neighborhood sampling time set, and to form a neighborhood distance array A according to the time proximity order.

[0033] The indicator extraction unit is used to construct a mutation intensity indicator B_jump and a trend intensity indicator B_trend based on the neighborhood distance array A, wherein the mutation intensity indicator B_jump represents the degree of sudden increase in the distance value between the sampling time and its neighboring sampling times, and the trend intensity indicator B_trend represents the degree of asymmetry of the distance value between the sampling time and the forward and backward neighbors.

[0034] The state recognition unit is used to generate at least two levels of hierarchical threshold intervals based on the mutation intensity index B_jump and the trend intensity index B_trend within the neighborhood constraint range, and to perform stage recognition of the crack state based on the hierarchical threshold intervals, and output the recognition result of whether the crack is in the rapid change stage, the expansion stage or the stable stage.

[0035] Correspondingly, the present invention also provides an electronic device, a readable storage medium, and a computer program product:

[0036] An electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the crack state identification method based on multi-source physical data fusion and the method for each step therein.

[0037] A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to perform the crack state identification method based on multi-source physical data fusion and the methods for each step therein.

[0038] A computer program product includes a computer program that, when executed by a processor, implements the crack state identification method based on multi-source physical data fusion and the methods for each step therein.

[0039] The beneficial effects of this invention are as follows: This invention provides a crack state identification method, system, and device based on multi-source physical data fusion. It can achieve unified representation and robust fusion of heterogeneous data based on the acquisition of multi-source physical quantities such as crack width, misalignment, dip angle, vibration intensity, and temperature; it can interpretably decompose the abrupt changes and trends of the state within the local temporal neighborhood; and through a robust threshold mechanism that is insensitive to operating condition drift and outliers, it can automatically output the identification results of whether the crack is in a stable stage, an expansion stage, or a rapid change stage, thereby improving the accuracy, stability, and engineering usability of crack state identification. Attached Figure Description

[0040] The above and other features of the present invention will become more apparent from the detailed description of the embodiments shown in conjunction with the accompanying drawings. In the accompanying drawings, the same reference numerals denote the same or similar elements. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort. In the drawings:

[0041] Figure 1 The flowchart shown is a method for crack state identification based on multi-source physical data fusion.

[0042] Figure 2 The figure shows the system architecture of a crack state identification system based on multi-source physical data fusion. Detailed Implementation

[0043] The following will provide a clear and complete description of the concept, specific structure, and technical effects of the present invention in conjunction with the embodiments and accompanying drawings, so as to fully understand the purpose, solution, and effects of the present invention. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0044] In the description of this invention, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0045] like Figure 1 The diagram shown is a flowchart of the crack state identification method based on multi-source physical data fusion according to the present invention. The following is a summary of the process. Figure 1 This paper describes a crack state identification method, system, and device based on multi-source physical data fusion according to embodiments of the present invention.

[0046] This invention proposes a crack state identification method based on multi-source physical data fusion, the method specifically including the following steps:

[0047] Obtain multivariate physical quantity data of the crack at multiple sampling times, and convert the multivariate physical quantity data at each sampling time into the corresponding multivariate physical quantity normalized feature vector.

[0048] Select a neighborhood constraint range R. For each sampling time, determine its forward neighborhood sampling time set and backward neighborhood sampling time set, where the forward neighborhood sampling time set includes the R sampling times before the sampling time and the backward neighborhood sampling time set includes the R sampling times after the sampling time.

[0049] Calculate the Euclidean distance between the normalized feature vector of the multivariate physical quantity at the sampling time and the normalized feature vector of the multivariate physical quantity corresponding to each sampling time in the forward neighborhood sampling time set and the backward neighborhood sampling time set, and form a neighborhood distance array A according to the time proximity order.

[0050] Based on the neighborhood distance array A, a mutation intensity index B_jump and a trend intensity index B_trend are constructed. The mutation intensity index B_jump represents the degree of sudden increase in the distance value between the sampling time and its neighboring sampling times, and the trend intensity index B_trend represents the degree of asymmetry of the distance value between the sampling time and the forward and backward neighbors.

[0051] Within the neighborhood constraint, at least two levels of hierarchical threshold intervals are generated based on the mutation intensity index B_jump and the trend intensity index B_trend, respectively. The crack state is then identified based on the hierarchical threshold intervals, and the identification results of whether the crack is in the rapid change stage, the expansion stage, or the stable stage are output.

[0052] Furthermore, the neighborhood constraint range R is 2, and the neighborhood distance array A consists of four elements, which correspond in order to the Euclidean distances of the sampling times of the first two elements, the sampling time of the previous element, the sampling time of the next element, and the sampling times of the last two elements.

[0053] Furthermore, for the neighborhood constraint range, when the sampling time is located at the sequence boundary, resulting in insufficient forward or backward neighborhood, any one or more of the following methods are adopted: mirroring, nearest value copying, or missing bit weighting, so that the length of the neighborhood distance array A is consistent with the position order.

[0054] Furthermore, the mutation intensity index B_jump is determined by two distance values ​​in the neighborhood distance array A corresponding to adjacent sampling times, and the mutation intensity index B_jump is the maximum or average of the two distance values.

[0055] Further, the trend strength index B_trend is determined by the mean of the forward neighborhood distance and the mean of the backward neighborhood distance, and the trend strength index B_trend is the absolute value of the difference between the mean of the forward neighborhood distance and the mean of the backward neighborhood distance; wherein, the forward neighborhood distance includes the distance value between the sampling time and each sampling time before it within the neighborhood constraint range, and the backward neighborhood distance includes the distance value between the sampling time and each sampling time after it within the neighborhood constraint range.

[0056] Furthermore, the threshold in the stratified threshold interval is determined based on the median and dispersion statistics of the corresponding index within the neighborhood constraint range. The dispersion statistics are either the absolute deviation of the median or the interquartile range. Based on the threshold, the index values ​​are divided into at least two levels of stratified threshold intervals.

[0057] Furthermore, the threshold is determined as follows: the median of the corresponding indicator within the neighborhood constraint range is used as a benchmark, and the multiple of the dispersion statistic is added to form the threshold, where the multiple is a preset parameter; the mutation intensity indicator B_jump or trend intensity indicator B_trend at the current sampling time is compared with the corresponding threshold to determine whether there is a significant anomaly.

[0058] Furthermore, in the stage identification of crack state based on the layered threshold interval, the criteria for the abrupt change stage of the crack include at least: the abrupt change intensity index B_jump at the current sampling time exceeds its corresponding threshold, and the structural vibration intensity exceeds the vibration threshold within the corresponding time window; in some embodiments, the criteria for the abrupt change stage also include the trend intensity index B_trend exceeding its corresponding threshold.

[0059] Furthermore, in the stage identification of crack state based on the layered threshold interval, the criteria for the crack expansion stage include at least: the mutation intensity index B_jump at the current sampling time does not exceed its corresponding threshold, while the trend intensity index B_trend exceeds its corresponding threshold, and the trend intensity index B_trend continues to exceed its corresponding threshold in the subsequent K consecutive sampling times, where K is an integer from 3 to 10.

[0060] Furthermore, in the stage identification of crack state based on the layered threshold interval, the criteria for the stable stage of crack include at least the following: the mutation intensity index B_jump and the trend intensity index B_trend at the current sampling time do not exceed their corresponding thresholds, and the mutation intensity index B_jump remains below its corresponding threshold for multiple consecutive sampling times, and the trend intensity index B_trend remains below its corresponding threshold for multiple consecutive sampling times.

[0061] In some embodiments, continuous monitoring of cracks in a concrete component was conducted. The sampling window included seven sampling times, denoted sequentially as t0, t1, t2, t3, t4, t5, and t6. Multivariate physical quantity data were acquired at each sampling time, including at least the crack width w, crack misalignment s, structural tilt angle θ (degree), structural vibration intensity E, and ambient temperature T. The raw data for each sampling time are as follows:

[0062] t0: w is 0.300mm, s is 0.050mm, θ is 0.000deg, E is 0.020g_rms, T is 20.0℃;

[0063] t1: w = 0.305 mm, s = 0.050 mm, θ = 0.005 deg, E = 0.021 g_rms, T = 20.1 ℃;

[0064] t2: w = 0.312 mm, s = 0.055 mm, θ = 0.010 deg, E = 0.019 g_rms, T = 20.2 ℃;

[0065] t3: w = 0.326 mm, s = 0.062 mm, θ = 0.020 deg, E = 0.023 g_rms, T = 20.3 ℃;

[0066] t4: w = 0.335 mm, s = 0.065 mm, θ = 0.028 deg, E = 0.200 g_rms, T = 20.4 ℃;

[0067] t5: w = 0.336 mm, s = 0.066 mm, θ = 0.029 deg, E = 0.120 g_rms, T = 20.5℃;

[0068] t6: w = 0.365 mm, s = 0.090 mm, θ = 0.050 deg, E = 0.080 g_rms, T = 20.6 ℃.

[0069] The multivariate physical quantity data at each sampling time are converted into corresponding normalized feature vectors. To highlight the changes in crack state, this embodiment uses the increment between adjacent time moments as input features, and normalizes each feature dimension to form a five-dimensional vector. The five components of the five-dimensional vector correspond to: crack width increment, misalignment increment, tilt angle increment, structural vibration intensity, and temperature increment. The normalized feature vectors at each sampling time are as follows, from the first to the fifth dimension, specifically including:

[0070] t0: 0.0000, 0.0000, 0.0000, 0.0000, 0.0;

[0071] t1: 0.1000, 0.0000, 0.1667, 0.0056, 0.5;

[0072] t2: 0.1400, 0.1667, 0.1667, 0.0000, 0.5;

[0073] t3: 0.2800, 0.2333, 0.3333, 0.0167, 0.5;

[0074] t4: 0.1800, 0.1000, 0.2667, 1.0000, 0.5;

[0075] t5: 0.0200, 0.0333, 0.0333, 0.5556, 0.5;

[0076] t6: 0.5800, 0.8000, 0.7000, 0.3333, 0.5;

[0077] In this embodiment, the structural vibration intensity corresponds to the fourth dimension of the aforementioned vector, which can be used as a criterion for subsequent rapid change stages.

[0078] Preferably, the neighborhood constraint range R is set to 2. For each sampling time, the two sampling times before, the one before that, the one after that, and the two sampling times after that are selected as neighborhood sampling times, and the Euclidean distance between the normalized feature vector of the current sampling time and the normalized feature vector of the neighborhood sampling times is calculated. These four Euclidean distances are arranged in a fixed order to form a neighborhood distance array A, with the order corresponding to the Euclidean distances with the two before, the one before that, the one after that, and the two after that.

[0079] In addition, when the sampling time is located at the sequence boundary, resulting in insufficient forward or backward neighborhood, mirroring can be used to maintain the length of array A and the positional order. For example, for t0, the previous position is mirrored as t1, and the previous two positions are mirrored as t2; for t6, the next position is mirrored as t5, and the last two positions are mirrored as t4; for the insufficient first two positions of t1, mirroring is used to maintain the positional order, and for the insufficient last two positions of t5, mirroring is used to maintain the positional order.

[0080] Under the above rules, the neighborhood distance array A at each sampling time is as follows, which are the Euclidean distances with the two preceding units, the one preceding unit, the one following unit, and the two following units, respectively:

[0081] t0: 0.5702, 0.5365, 0.5365, 0.5702;

[0082] t1: 0.0000, 0.5365, 0.1715, 0.3387;

[0083] t2: 0.5702, 0.1715, 0.2282, 1.0080;

[0084] t3: 0.3387, 0.2282, 0.9995, 0.6986;

[0085] t4: 1.0080, 0.9995, 0.5311, 1.1324;

[0086] t5: 0.6986, 0.5311, 1.1812, 0.0000;

[0087] t6: 1.1324, 1.1812, 1.1812, 1.1324.

[0088] Then, two metrics are constructed based on the neighborhood distance array A, where:

[0089] The mutation intensity index B_jump is used to characterize the degree of sudden increase in the distance value between the current sampling time and the adjacent sampling time. Optionally, in this embodiment, the larger value between the distance to the previous value and the distance to the next value is taken as B_jump.

[0090] The trend strength index B_trend is used to characterize the degree of asymmetry in the distance values ​​at the sampling time relative to the forward and backward neighbors. In this embodiment, the mean values ​​of the two forward distances and the mean values ​​of the two backward distances are calculated respectively, and the absolute value of the difference between the two is taken as B_trend.

[0091] Based on this, B_jump and B_trend at each sampling time are obtained as follows:

[0092] t0: B_jump is 0.5365; B_trend is 0.0000;

[0093] t1: B_jump is 0.5365; B_trend is 0.0131;

[0094] t2: B_jump is 0.2282; B_trend is 0.2472;

[0095] t3: B_jump is 0.9995; B_trend is 0.5656;

[0096] t4: B_jump is 0.9995; B_trend is 0.1721;

[0097] t5: B_jump is 1.1812; B_trend is 0.0242;

[0098] t6: B_jump is 1.1812; B_trend is 0.0000.

[0099] To generate robust stratified threshold intervals, this embodiment selects t0, t1, and t2 as historical data samples for stable periods, and establishes stratified threshold intervals for B_jump, B_trend, and structural vibration intensity, respectively. Specifically, this embodiment uses the interquartile range as the dispersion statistic, and the threshold stratification can adopt a two-level threshold method, for example:

[0100] The first threshold is the median plus twice the interquartile range; the second threshold is the median plus four times the interquartile range; and based on these, three intervals are divided: normal interval, warning interval, and alarm interval.

[0101] In the stable segment samples, the values ​​of B_jump are 0.5365, 0.5365, and 0.2282. Therefore, the median of B_jump is 0.5365; the interquartile range is 0.1542; the first threshold is 0.8448; and the second threshold is 1.1531.

[0102] Therefore, the normal range for B_jump is no higher than 0.8448; the warning range for B_jump is higher than 0.8448 but no higher than 1.1531; and the alarm range for B_jump is higher than 1.1531.

[0103] In the stable segment samples, the values ​​of B_trend are 0.0000, 0.0131, and 0.2472. Therefore, the median of B_trend is 0.0131; the interquartile range is 0.1236; the first threshold is 0.2603; and the second threshold is 0.5076.

[0104] Therefore: the normal range of B_trend is no higher than 0.2603; the warning range of B_trend is higher than 0.2603 but no higher than 0.5076; and the alarm range of B_trend is higher than 0.5076.

[0105] Furthermore, in this embodiment, the structural vibration intensity utilizes the fourth dimension of the normalized eigenvector. The values ​​of the fourth dimension in the stable segment samples are 0.0000, 0.0056, and 0.0000. Therefore:

[0106] The median vibration intensity was 0.0000; the interquartile range was 0.0056.

[0107] The first vibration threshold is 0.0112; the second vibration threshold is 0.0224.

[0108] Therefore, the normal range of vibration intensity is no higher than 0.0112; the warning range of vibration intensity is higher than 0.0112 but no higher than 0.0224; and the alarm range of vibration intensity is higher than 0.0224.

[0109] In some embodiments, the crack state is further identified in stages based on a stratified threshold range; wherein, the criterion settings may also be as follows:

[0110] The criterion for the stable phase may include: both B_jump and B_trend at the current sampling time fall within their respective normal ranges; optionally, their vibration intensity also falls within the normal range.

[0111] The extended phase criteria may include: B_jump at the current sampling time falls into the normal range, while B_trend falls into the warning range or alarm range, and the above-mentioned abnormal trend continues to appear in multiple subsequent sampling times;

[0112] The criteria for the rapid change phase may include: the sudden change intensity index B_jump at the current sampling time exceeds its corresponding threshold; optionally, and its structural vibration intensity exceeds the vibration threshold.

[0113] Preferably, in this embodiment, the corresponding threshold is taken as the first threshold, that is: the threshold corresponding to B_jump is 0.8448; the vibration threshold is 0.0112. Therefore, as long as B_jump at the sampling time is higher than 0.8448 and its vibration intensity is higher than 0.0112, the criterion for the rapid change stage can also be met.

[0114] Preferably, the time window can be a time window of 2R or 2R+1 sampling points centered on the current sampling time, corresponding to the selected neighborhood constraint range R.

[0115] Next, under the aforementioned thresholds and criteria, the determination process for t0 to t6 can be specifically described as follows:

[0116] For sampling time t0: B_jump is 0.5365, which falls within the normal range of B_jump; B_trend is 0.0000, which falls within the normal range of B_trend; vibration intensity is 0.0000, which falls within the normal range of vibration; therefore, the output is in the stable phase.

[0117] For sampling time t1: B_jump is 0.5365, which falls within the normal range of B_jump; B_trend is 0.0131, which falls within the normal range of B_trend; vibration intensity is 0.0056, which falls within the normal range of vibration; therefore, the output is in a stable phase.

[0118] For sampling time t2: B_jump is 0.2282, which falls within the normal range of B_jump; B_trend is 0.2472, which falls within the normal range of B_trend; vibration intensity is 0.0000, which falls within the normal range of vibration; therefore, the output is in a stable phase.

[0119] For sampling time t3: B_jump is 0.9995, falling into the B_jump warning interval and exceeding the corresponding threshold of 0.8448; B_trend is 0.5656, falling into the B_trend alarm interval; vibration intensity is 0.0167, falling into the vibration warning interval and exceeding the vibration threshold of 0.0112; since B_jump is higher than the corresponding threshold and vibration intensity is higher than the vibration threshold, the criterion for the acute change stage is met, so the output is the acute change stage.

[0120] For sampling time t4: B_jump is 0.9995, falling into the B_jump warning range and exceeding the corresponding threshold of 0.8448; B_trend is 0.1721, falling into the normal range of B_trend; vibration intensity is 1.0000, falling into the vibration alarm range and exceeding the vibration threshold of 0.0112; since B_jump is higher than the corresponding threshold and vibration intensity is higher than the vibration threshold, the criteria for the acute change stage are met, so the output is the acute change stage.

[0121] For sampling time t5: B_jump is 1.1812, falling into the B_jump alarm range and exceeding the corresponding threshold of 0.8448; B_trend is 0.0242, falling into the normal range of B_trend; vibration intensity is 0.5556, falling into the vibration alarm range and exceeding the vibration threshold of 0.0112; since B_jump is higher than the corresponding threshold and vibration intensity is higher than the vibration threshold, the criteria for the acute change stage are met, so the output is the acute change stage.

[0122] For sampling time t6: B_jump is 1.1812, falling into the B_jump alarm range and exceeding the corresponding threshold of 0.8448; B_trend is 0.0000, falling into the normal range of B_trend; vibration intensity is 0.3333, falling into the vibration alarm range and exceeding the vibration threshold of 0.0112; since B_jump is higher than the corresponding threshold and vibration intensity is higher than the vibration threshold, the criteria for the acute change stage are met, so the output is the acute change stage.

[0123] This embodiment converts multivariate physical quantity data into normalized feature vectors, calculates Euclidean distances within neighborhood constraints to form a neighborhood distance array A, constructs a sudden change intensity index B_jump and a trend intensity index B_trend, and generates stratified threshold intervals based on the median and interquartile range to achieve stage identification of crack states. Especially in the criterion for abrupt change stages, the combined condition of B_jump exceeding the threshold and structural vibration intensity exceeding the threshold makes the abrupt change stage identification results more accurate.

[0124] The crack state identification system based on multi-source physical data fusion runs on any computing device, such as a desktop computer, laptop computer, handheld computer, or cloud data center. The computing device includes a processor, a memory, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps in the crack state identification method based on multi-source physical data fusion. The runnable system may include, but is not limited to, processors, memory, and server clusters.

[0125] The crack state identification system based on multi-source physical data fusion provided by the embodiments of the present invention, such as... Figure 2 As shown, the crack state identification system based on multi-source physical data fusion in this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the above-described crack state identification method embodiment based on multi-source physical data fusion. The processor executes the computer program in the following system unit:

[0126] The data acquisition unit is used to acquire multivariate physical quantity data of the crack at multiple sampling times and convert the multivariate physical quantity data at each sampling time into the corresponding multivariate physical quantity normalized feature vector.

[0127] The constraint selection unit is used to select the neighborhood constraint range R. For each sampling time, it determines its forward neighborhood sampling time set and backward neighborhood sampling time set, wherein the forward neighborhood sampling time set includes the R sampling times before the sampling time, and the backward neighborhood sampling time set includes the R sampling times after the sampling time.

[0128] The neighborhood calculation unit is used to calculate the Euclidean distance between the normalized feature vector of the multivariate physical quantity at the sampling time and the normalized feature vector of the multivariate physical quantity corresponding to each sampling time in the forward neighborhood sampling time set and the backward neighborhood sampling time set, and to form a neighborhood distance array A according to the time proximity order.

[0129] The indicator extraction unit is used to construct a mutation intensity indicator B_jump and a trend intensity indicator B_trend based on the neighborhood distance array A, wherein the mutation intensity indicator B_jump represents the degree of sudden increase in the distance value between the sampling time and its neighboring sampling times, and the trend intensity indicator B_trend represents the degree of asymmetry of the distance value between the sampling time and the forward and backward neighbors.

[0130] The state recognition unit is used to generate at least two levels of hierarchical threshold intervals based on the mutation intensity index B_jump and the trend intensity index B_trend within the neighborhood constraint range, and to perform stage recognition of the crack state based on the hierarchical threshold intervals, and output the recognition result of whether the crack is in the rapid change stage, the expansion stage or the stable stage.

[0131] In order to better unify the linear relationship and probabilistic connection between physical quantities with different units of measurement, dimensionless processing can be performed on different physical quantities.

[0132] Preferably, all undefined variables in this invention, if not explicitly defined, can be manually set thresholds.

[0133] The crack state identification system based on multi-source physical data fusion can run on computing devices such as desktop computers, laptops, handheld computers, and cloud data centers. The crack state identification system based on multi-source physical data fusion includes, but is not limited to, processors and memory. Those skilled in the art will understand that the examples described are merely illustrations of crack state identification methods, systems, and devices based on multi-source physical data fusion, and do not constitute a limitation on the crack state identification methods, systems, and devices based on multi-source physical data fusion. It may include more or fewer components, or combine certain components, or different components. For example, the crack state identification system based on multi-source physical data fusion may also include input / output devices, network access devices, buses, etc.

[0134] The present invention also provides an electronic device, a readable storage medium, and a computer program product:

[0135] An electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the crack state identification method based on multi-source physical data fusion and the method for each step therein.

[0136] A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to perform the crack state identification method based on multi-source physical data fusion and the methods for each step therein.

[0137] A computer program product includes a computer program that, when executed by a processor, implements the crack state identification method based on multi-source physical data fusion and the methods for each step therein.

[0138] The term "electronic device" is intended to refer to various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices can also refer to various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0139] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0140] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0141] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0142] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0143] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0144] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

[0145] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete component gate circuits, transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center of the crack state identification system based on multi-source physical data fusion, connecting various sub-regions of the system via various interfaces and lines.

[0146] The memory can be used to store the computer program and / or modules. The processor, by running or executing the computer program and / or modules stored in the memory, and by calling the data stored in the memory, realizes various functions of the crack state identification method, system, and device based on multi-source physical data fusion. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0147] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0148] This invention provides a method, system, and device for crack state identification based on multi-source physical data fusion. It normalizes multivariate physical quantity data at each sampling time into corresponding multivariate physical quantity normalized feature vectors, then selects a neighborhood constraint range. For each sampling time, it calculates the distance between the sampling time and the normalized feature vectors of each multivariate physical quantity at each sampling time within the neighborhood constraint range, and arranges these distance values ​​into a neighborhood distance array in rank. Based on this neighborhood distance array, it constructs a mutation intensity index and a trend intensity index. Within the neighborhood constraint range, it generates tiered threshold intervals based on the mutation intensity index and the trend intensity index, and performs stage identification of the crack state based on these tiered threshold intervals, outputting the stage identification results. Through a robust threshold mechanism that is insensitive to operating condition drift and outliers, it automatically outputs the identification results of whether the crack is in a stable stage, an expansion stage, or a rapid change stage, thereby improving the accuracy, stability, and engineering usability of crack state identification.

[0149] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A crack state identification method based on multi-source physical data fusion, characterized in that, The method includes: Acquire multivariate physical quantity data of cracks at multiple sampling times. The multivariate physical quantity data includes at least one or more of the following: crack width, crack misalignment, structural tilt angle, structural vibration intensity, and ambient temperature. The multivariate physical quantity data at each sampling time are normalized into the corresponding multivariate physical quantity normalized feature vector; Select a neighborhood constraint range, and for each sampling time, calculate the distance value between the normalized feature vectors of each multivariate physical quantity relative to each sampling time within the neighborhood constraint range. Then, arrange the distance values ​​into a neighborhood distance array in order of position. Based on the neighborhood distance array, a mutation intensity index and a trend intensity index are constructed, wherein: the mutation intensity index characterizes the degree of sudden increase in the distance value between the current sampling time and the adjacent sampling time within the neighborhood constraint range, and the trend intensity index characterizes the degree of asymmetry in the distance value between the sampling time before and after the current sampling time within the neighborhood constraint range; Within the neighborhood constraint range, hierarchical threshold intervals are generated based on the mutation intensity index and the trend intensity index, respectively. The crack state is then identified in stages based on the hierarchical threshold intervals, and the results of the stage identification are output. The neighborhood constraint range is 2, and the neighborhood distance array is constructed into four elements according to the position order of adjacent sampling times, which respectively correspond to the Euclidean distance between the sampling times of the first two positions, the sampling time of the previous position, the sampling time of the next position, and the sampling time of the last two positions. The mutation intensity index is determined by two distance values ​​corresponding to adjacent sampling times in the neighborhood distance array, and the mutation intensity index is the maximum or average of the two distance values; The trend strength index is determined by the mean of the forward neighborhood distance and the mean of the backward neighborhood distance. The trend strength index is the absolute value of the difference between the mean of the forward neighborhood distance and the mean of the backward neighborhood distance. The forward neighborhood distance includes the distance between the current sampling time and each sampling time before it within the neighborhood constraint range, and the backward neighborhood distance includes the distance between the current sampling time and each sampling time after it within the neighborhood constraint range.

2. The crack state identification method based on multi-source physical data fusion according to claim 1, characterized in that, For the neighborhood constraint range, when the sampling time is located at the sequence boundary, resulting in insufficient forward or backward neighborhood, any one or more of the following methods are used to ensure that the length of the neighborhood distance array remains consistent: mirroring, copying the nearest value, or reducing the weight of missing bits.

3. The crack state identification method based on multi-source physical data fusion according to claim 1, characterized in that, in, The threshold in the stratified threshold interval is determined based on the median and / or dispersion statistics of the corresponding indicators within the neighborhood constraint range. The dispersion statistics are either the absolute deviation of the median or the interquartile range. The stratified threshold interval is divided according to the threshold.

4. The crack state identification method based on multi-source physical data fusion according to claim 3, characterized in that, in, The stratified threshold interval is determined as follows: the median and dispersion statistics of the corresponding indicators within the neighborhood constraint range are weighted and combined to form the corresponding thresholds, and the mutation intensity index or trend intensity index at the current sampling time is compared with the corresponding threshold to determine whether there is a significant anomaly.

5. The crack state identification method based on multi-source physical data fusion according to claim 4, characterized in that, In the stage identification of crack state based on the layered threshold interval, the criteria for the crack abrupt change stage include at least: the abrupt change intensity index at the current sampling time exceeds its corresponding threshold, and the trend intensity index exceeds its corresponding threshold within the neighborhood constraint range.

6. The crack state identification method based on multi-source physical data fusion according to claim 4, characterized in that, In the stage identification of crack state based on the layered threshold interval, the criteria for crack expansion stage include at least: the abrupt change intensity index at the current sampling time does not exceed its corresponding threshold, while the trend intensity index at the current sampling time exceeds its corresponding threshold, and the trend intensity index continues to exceed its corresponding threshold in subsequent consecutive sampling times, and the number of subsequent consecutive sampling times is not less than the number of sampling times contained in the neighborhood constraint range.

7. The crack state identification method based on multi-source physical data fusion according to claim 4, characterized in that, In the stage identification of crack state based on the layered threshold range, the criteria for the stable stage of crack include at least the following: the mutation intensity index and the trend intensity index at the current sampling time do not exceed the corresponding threshold, and the mutation intensity index is continuously lower than its corresponding threshold in subsequent consecutive sampling times.

8. A crack state identification system based on multi-source physical data fusion, characterized in that, The crack state identification system based on multi-source physical data fusion operates on any computing device, such as a desktop computer, a laptop computer, or a cloud data center. The computing device includes a processor, a memory, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps in the crack state identification method based on multi-source physical data fusion as described in any one of claims 1 to 7.

9. An electronic device, comprising: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores instructions executable by the at least one processor, characterized in that the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

Images