Ultra-high performance concrete building plate damage assessment and early warning method based on multi-source sensing data

Abstract

The invention discloses an ultra-high performance concrete building plate damage assessment early warning method based on multi-source sensing data, and relates to the technical field of concrete building plate damage assessment, and the method comprises the following steps: calculating a change amplitude deviation value, a change rate distribution index and a change duration length index of recognized high fluctuation data; judging whether violent fluctuation is formed or not through a multi-dimensional threshold value, and marking the violent fluctuation as an abnormal signal source; the change direction consistency, the fitting slope similarity and the fitting residual range are calculated for other sensor data in the time period corresponding to the marked abnormal signal source, the change direction consistency, the fitting slope similarity and the fitting residual range are used for generating a trend stability score value, and other sensors refer to sensors which are not marked as abnormal signal sources. According to the method, the problem of abnormal signal misjudgment in multi-source sensing data fusion is solved, abnormal signal dynamic identification and weight suppression are realized, and the accuracy and reliability of structural damage early warning are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of concrete panel damage assessment, and particularly relates to a method for damage assessment and early warning of ultra-high performance concrete panel based on multi-source sensing data. BACKGROUND

[0002] The damage assessment and early warning of ultra-high performance concrete (UHPC) panel based on multi-source sensing data is a method for obtaining multi-dimensional and multi-temporal physical parameters during the operation of the structure through multi-type sensors (such as strain gauges, accelerometers, temperature and humidity sensors, vibration sensors, etc.) arranged on the surface or inside of the panel structure, and for performing fusion processing on the collected data, thereby realizing real-time assessment of the health state of the structure and early warning of potential damage trends. The existing technology usually realizes this goal through the following several links: firstly, scientific arrangement and calibration of multi-source sensors to ensure coverage of key stress positions and guarantee data reliability; secondly, synchronous acquisition and preprocessing of multi-source data, including denoising, normalization and data alignment steps; then, feature extraction and fusion modeling, using data fusion algorithms or machine learning models to comprehensively analyze multiple dimensions such as stress response, vibration frequency change, temperature influence, etc.; then, state recognition is performed by setting damage recognition indicators or training models to recognize possible cracks, fatigue or other forms of damage signs; finally, risk assessment and early warning output based on the recognition results, including early warning level determination, alarm information release and structure maintenance suggestions, etc. Through this complete process, the running state of the UHPC panel structure can be comprehensively mastered and early warning can be realized, thereby improving the safety and service life of the infrastructure.

[0003] The existing technology has the following deficiencies: In the process of damage assessment and early warning of ultra-high performance concrete slab based on multi-source sensing data, if a strain sensor produces a non-structural sudden signal due to poor installation contact in a specific collection period, while the data collected by other types of sensors (such as cracks, acoustic emission, temperature, etc.) remain stable, the abnormal signal will be given an unreasonable evaluation weight in the data fusion link due to its amplitude being significantly higher than that of other data sources, and then the comprehensive damage assessment result will be abnormally high, triggering a false high-level warning. Since the current fusion evaluation mechanism mainly relies on fixed sensor weight distribution method, it lacks the ability to dynamically compare and identify the trend stability of other data sources under the condition of sudden fluctuation of a single data source, so the system cannot determine whether the sudden signal is a non-structural interference, thereby suppressing the triggering of false warning. Therefore, the existing damage assessment and early warning technology of ultra-high performance concrete slab based on multi-source sensing data cannot reasonably judge the credibility of the signal under the condition of sudden and severe fluctuation of a single signal source in the fusion evaluation process according to the trend stability of other data sources, and suppress the false triggering behavior caused by abnormal data, which may eventually cause the structure to be misjudged as severely damaged, and then trigger unnecessary maintenance scheduling, traffic control or operation interruption, while reducing the sensitivity of the system to real damage, and reducing the reliability of the warning and the efficiency of the engineering response.

[0004] The above information disclosed in the BACKGROUND section merely to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0005] The purpose of the present application is to provide a multi-source sensing data based damage assessment and early warning method for ultra-high performance concrete slab to solve the problems in the background.

[0006] In order to achieve the above-mentioned purpose, the present application provides the following technical solution: a multi-source sensing data based damage assessment and early warning method for ultra-high performance concrete slab, specifically comprising the following steps: S1, constructing continuous data obtained by a plurality of sensors in a unified collection period into a time sequence response trajectory set, and extracting a change rate index of each sensor for identifying high fluctuation data; S2, calculating the change amplitude deviation value, change rate distribution index and change duration length index of the identified high fluctuation data, and judging whether it constitutes a severe fluctuation and is marked as an abnormal signal source through multi-dimensional threshold; S3, calculating the change direction consistency, fitting slope similarity and fitting residual range of other sensor data in the time period corresponding to the marked abnormal signal source for generating a trend stability score value, wherein the other sensors refer to sensors that are not marked as abnormal signal sources; S4, the severe fluctuation judgment result is combined with the trend stability score value to be mapped into a credibility value, which is used as the participation reliability basis of the abnormal signal source; S5, the participation weight of the abnormal signal source in the fusion evaluation calculation is adjusted based on the credibility value, the evaluation parameters of each sensor are fused by weighting, the structure damage evaluation result is generated, and the warning output is completed.

[0007] Preferably, S2 specifically comprises the following steps: S201, the difference between the maximum value and the minimum value of each identified high fluctuation data is calculated, and compared with the average difference value of the corresponding data of the same type sensor in the historical acquisition period, to obtain the change amplitude deviation value; S202, the numerical value change interval between adjacent data points of each identified high fluctuation data is counted, a change rate distribution map is constructed, and the frequency concentration interval and the change interval width are extracted as the change rate distribution index; S203, the time length of each identified high fluctuation data remaining higher than the change rate judgment reference in the continuous sampling period is calculated as the change duration length index; S204, the change amplitude deviation value is compared with the change amplitude deviation threshold value, the change rate distribution index is compared with the change rate distribution threshold value, and the change duration length is compared with the change duration time threshold value, when the three comparison results are all greater than the relationship, it is determined that the high fluctuation data constitutes a severe fluctuation, and the corresponding sensor is marked as an abnormal signal source.

[0008] Preferably, S201 specifically comprises: The maximum value and the minimum value of each identified high fluctuation data in the current acquisition period are extracted, and the difference value is calculated as the current change amplitude value; A plurality of time periods with the same length as the current acquisition period are selected from the historical data of the same type sensor, the maximum value and the minimum value difference in each time period is calculated, and the arithmetic average value of these difference values is calculated as the average difference value; The current change amplitude value is subtracted from the average difference value, and the obtained value is used as the change amplitude deviation value for subsequent severe fluctuation judgment.

[0009] Preferably, S202 specifically comprises: The numerical difference between each two adjacent data points of each identified high fluctuation data is calculated in sequence, and the change interval sequence composed of all the difference values is counted; The change interval sequence is counted and a histogram distribution graph is drawn to generate a change rate distribution map; The continuous numerical interval with the highest frequency in the change rate distribution map is identified as the frequency concentration interval, and the numerical difference between the minimum change value and the maximum change value in the distribution map is calculated as the change interval width.

[0010] Preferably, S203 specifically comprises: Point-by-point comparison is made between the change rate sequence corresponding to each identified high fluctuation data segment and the change rate determination reference, and a sequence of continuous sampling points with values greater than the change rate determination reference is marked; The number of consecutive adjacent sampling points is counted in each marked sampling point sequence, and the number is multiplied by the sampling interval to obtain the time length in each sequence that maintains a change rate greater than the change rate determination reference; The maximum value is selected from all time lengths as the change duration indicator of the currently identified high fluctuation data.

[0011] Preferably, S3 specifically comprises the following steps: S301, selecting sensor data corresponding to a time period of a sensor marked as an abnormal signal source and not marked as an abnormal signal source, constructing time sequence response trajectories of each sensor in the time period, and extracting the change direction of each response trajectory; S302, point-by-point direction comparison is made between the change direction of each sensor not marked as an abnormal signal source and the change direction of the sensor marked as an abnormal signal source, and the proportion of the number of data points with consistent directions is counted as the change direction consistency; S303, linear least squares regression is performed on all sensor response trajectories not marked as an abnormal signal source respectively, the fitting slope and fitting residual are obtained, and numerical comparison is made with the fitting results of the sensor marked as an abnormal signal source, and the fitting slope similarity and fitting residual difference are calculated respectively; S304, according to the change direction consistency, fitting slope similarity and fitting residual difference, weight coefficients are assigned respectively, and a trend stability score value is generated in a weighted average manner, which is used to determine whether to perform weight suppression processing on the abnormal signal source.

[0012] Preferably, S302 specifically comprises: The change direction sequence of adjacent data points of the sensor marked as an abnormal signal source in the corresponding time period is extracted, and the change direction is determined by the difference sign of adjacent data points; The change direction sequence of the sensor not marked as an abnormal signal source in the same time period is extracted, and the time points are one-to-one corresponding to the direction sequence of the abnormal signal source; The directions of each pair of corresponding data points in the two change direction sequences are compared, the number of data points with consistent directions is counted, and the number is divided by the total number of data points to calculate the consistency proportion of the directions, which is used as the change direction consistency indicator.

[0013] Preferably, S303 specifically comprises: Applying linear least square fitting to the response trajectory of each sensor not marked as an abnormal signal source in the target time period, the corresponding fitting slope and residual mean square value are obtained; The same fitting processing is performed on the response trajectory of the sensor marked as an abnormal signal source, and the corresponding fitting slope and residual mean square value are obtained as a comparison reference; The difference between the fitting slope of all sensors not marked as an abnormal signal source and the comparison reference slope is calculated as the fitting slope similarity, and the maximum difference between the fitting residual of all sensors not marked as an abnormal signal source and the comparison reference residual is extracted as the fitting residual range.

[0014] Preferably, S4 specifically is: The fluctuation judgment result is mapped to a numerical magnitude, and is set to a first preset numerical value when the fluctuation judgment is established, and is set to a second preset numerical value when the fluctuation judgment is not established, to form a fluctuation numerical factor; The trend stability score value is subjected to interval normalization processing, and the score value is converted into a trend stability factor of a unified standard, to ensure that it can be additively fused with the fluctuation numerical factor under the same numerical scale; The fluctuation numerical factor and the trend stability factor are taken as joint input variables, a weighted linear combination function is executed, and a credibility numerical value is output, which is used as a participation reliability basis of the corresponding abnormal signal source in subsequent information fusion calculation.

[0015] Preferably, S5 specifically is: The participation weight of the sensor marked as an abnormal signal source in the fusion evaluation calculation is adjusted based on the credibility numerical value, the participation weight of the sensor not marked as an abnormal signal source is set to a full weight value, and the participation weight of the sensor marked as an abnormal signal source is equal to the corresponding credibility numerical value; The evaluation parameters of all sensors in the current evaluation period are extracted, the evaluation parameter of each sensor is multiplied by the corresponding participation weight to obtain the weighted evaluation value of each sensor, and all weighted evaluation values are subjected to weight normalization aggregation to generate a structure damage evaluation result; The structure damage evaluation result is compared with a damage judgment threshold value, when the structure damage evaluation result is greater than the damage judgment threshold value, a structure damage warning signal is output, and the structure positioning information, damage level numerical value and participation weight information in the current evaluation period are also output, to complete the warning output.

[0016] In the above technical solution, the present application provides the following technical effects and advantages: 1、The application can effectively solve the misjudgment problem caused by individual sensor signal mutation in structural damage assessment by constructing an abnormality recognition and dynamic fusion mechanism based on multi-source sensing data. This method not only fuses the data of multiple sensors such as strain, crack, acoustic emission and temperature, but also introduces multi-dimensional indexes such as change amplitude deviation, change rate distribution and change duration to accurately identify high volatility data, and establishes a violent fluctuation judgment logic by comparing with historical data, realizing the preliminary screening of abnormal signal sources. On this basis, by calculating the change direction consistency, fitting slope similarity and fitting residual difference between the abnormal signal and other sensors, the trend stability score is formed, and then the score and the fluctuation judgment result are jointly mapped into the reliability, realizing the quantification of the reliability of the abnormal signal, thereby providing a scientific basis for subsequent weighted fusion.

[0017] 2、The application adjusts the participation weight of the abnormal signal source in real time through the reliability value, and executes weighted aggregation combined with other sensor evaluation parameters, effectively suppressing the interference of non-structural violent fluctuation signals on the final damage assessment result, and avoiding false alarm of high-level warning. At the same time, the method can provide multi-dimensional information such as structure positioning, damage level and participation weight at the same time when outputting the structure damage assessment result, improving the explanation and decision value of the warning, greatly enhancing the sensitivity and overall warning reliability of the system to the real structure damage. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0019] Figure 1 The flowchart of the method for evaluating and warning the damage of super high performance concrete panel based on multi-source sensing data of the present application. DETAILED DESCRIPTION

[0020] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations, however, can be implemented in many different ways and should not be construed as limited to the examples set forth herein; rather, these example implementations are provided so that this disclosure will be thorough and complete, and will fully convey the inventive concept to those skilled in the art.

[0021] The present application provides a method for evaluating and warning the damage of super high performance concrete panel based on multi-source sensing data as shown in Figure 1 The method comprises the following steps: S1, construct continuous data obtained by multiple sensors in a unified collection period into a time sequence response track set, and extract the change rate index of each sensor for identifying high fluctuation data; In this embodiment, S1 is specifically: The continuous data obtained by multiple sensors in a unified collection period is bidirectionally sorted according to sensor categories and collection time, a time sequence corresponding to each category of sensors is constructed, and a time sequence response track set is formed by combination; In the process of bidirectionally sorting the continuous data obtained by multiple sensors in a unified collection period according to sensor categories and collection time, constructing a time sequence corresponding to each category of sensors, and combining to form a time sequence response track set, a unified collection period boundary is first set, and the original data returned by each sensor is preliminarily classified according to sensor number, each category representing a physical quantity type data source, such as strain, vibration, temperature or acoustic emission. Subsequently, within each category of sensor data, the time stamp field in each data record is read, and ascending order is arranged according to time sequence to ensure that the order of each sensor data point on the time axis is complete and consistent. Then, the time sequences of multiple sensors under the same physical quantity category are horizontally spliced to form a response vector for each time, and finally a complete time sequence matrix of the category of sensors is constructed. For multiple categories of data, the above time sequence matrix is arranged according to the physical quantity category one by one, and is integrated vertically into a multi-dimensional time-sensor response structure, and is organized and stored in the form of two-dimensional or three-dimensional array to form a complete time sequence response track set across categories and time axes, which is used for subsequent change rate analysis and high fluctuation identification. In the whole process, the data can be efficiently processed and accurately constructed through timestamp alignment algorithm, multi-dimensional array splicing operation and data structure index mechanism.

[0022] The time sequence corresponding to each category of sensors is extracted from the time sequence response track set, and a change rate index sequence is constructed through the numerical difference between adjacent collection time points; In the process of extracting the time series corresponding to each type of sensor from the set of time series response trajectories and constructing the change rate index sequence through the numerical difference of adjacent collection time points, first, the time series response trajectories set needs to be traversed by sensor type to extract the time series matrix corresponding to each type of sensor. Each row in the matrix represents the continuous data of a sensor in a unified collection period, and each column corresponds to a standardized time sampling point. Then, for each row of time series data, the difference algorithm is used to calculate the difference between the numerical values of adjacent collection time points, i.e., the (n+1)th sampling value minus the nth sampling value, to obtain the continuous change value sequence. This sequence is the change rate index sequence, which quantifies the change rate of the sensor response in the time dimension. This operation can be achieved through a sliding window mechanism with a window length of 2 and a step size of 1 to ensure consistency in the length of the difference sequence and the original sequence, and the result sequence is standardized to adjust the uniform dimension, making the change rate index between different sensors comparable. Throughout the process, the change rate index sequence not only retains the dynamic trend characteristics of the original signal, but also effectively eliminates the information redundancy caused by low-frequency slow changes, facilitating the accurate identification of sudden changes by the subsequent high volatility identification module.

[0023] The mean and standard deviation of the change rate index sequence of each type of sensor are calculated, and the continuous segment with a change rate greater than the sum of the mean and standard deviation is selected as high volatility data, and the identification of high volatility data is completed.

[0024] In the process of calculating the mean and standard deviation of the change rate index sequence of each type of sensor and selecting the continuous segment with a change rate greater than the sum of the mean and standard deviation as high volatility data, first, the change rate index sequence needs to be statistically processed to calculate the arithmetic mean to reflect the overall change trend, and the standard deviation to measure the distribution of the sequence. Then, set the volatility identification threshold to the sum of the mean and standard deviation as the limit for judging abnormal volatility. By traversing the change rate index sequence point by point, compare each change rate value with the threshold. If the change rate of multiple consecutive sampling points is greater than the threshold, the continuous segment is extracted as high volatility data. This operation can be completed by combining Boolean markers and sliding windows. If the number of consecutive points in the window that meet the conditions reaches the set lower limit, they are identified as high volatility segments. The design logic of this method is that the combined threshold of mean and standard deviation takes into account both the overall trend and the local volatility, which can adapt to the changes in the response characteristics of different sensors, while avoiding misidentifying normal small fluctuations as abnormal, which helps to improve the identification accuracy and accurately extract atypical violent fluctuation behavior, providing a reliable foundation for subsequent credibility assessment and early warning control.

[0025] S2, calculate the change amplitude deviation value, the change rate distribution index and the change duration index of the identified high fluctuation data, and determine whether it constitutes a violent fluctuation and mark it as an abnormal signal source through multi-dimensional threshold value judgment; In this embodiment, S2 specifically includes the following steps: S201, calculate the difference between the maximum value and the minimum value of each identified high fluctuation data, and compare it with the average difference of the corresponding data of the same type sensor in the historical collection period to obtain the change amplitude deviation value; S202, count the numerical value change interval between adjacent data points of each identified high fluctuation data, construct a change rate distribution map, and extract the frequency concentration interval and the change interval width as the change rate distribution index; S203, calculate the time length of each identified high fluctuation data remaining above the change rate judgment reference in the continuous sampling period as the change duration index; S204, compare the change amplitude deviation value with the change amplitude deviation threshold value, compare the change rate distribution index with the change rate distribution threshold value, and compare the change duration with the change duration threshold value. When the three comparison results are all greater than the relationship, it is determined that the high fluctuation data constitutes a violent fluctuation, and the corresponding sensor is marked as an abnormal signal source.

[0026] In the violent fluctuation judgment process, in order to ensure the multi-dimensional robustness of the judgment result, three key indexes need to be included in the judgment logic. The specific way is: first, compare the change amplitude deviation value of each high fluctuation data with the preset change amplitude deviation threshold value; second, compare the frequency concentration interval width extracted from the corresponding change rate distribution index with the change rate distribution threshold value; third, compare the calculated change duration with the change duration threshold value. When the three comparison results are all greater than the relationship, it means that the data in the amplitude deviation degree, rate fluctuation amplitude and time dimension all reach a significant abnormal level, which comprehensively reflects the strong abnormal fluctuation characteristics. At this time, it can be determined that the high fluctuation data constitutes a violent fluctuation, and its corresponding sensor is marked as an abnormal signal source in the current collection period for subsequent credibility judgment and warning weight adjustment. By introducing three types of indexes at the same time and setting a strict multi-threshold joint judgment mechanism, it can effectively avoid misjudgment due to single index abnormality, and improve the accuracy and anti-interference ability of violent fluctuation identification.

[0027] The change amplitude deviation threshold, the change rate distribution threshold and the change duration threshold are respectively used to determine whether the high fluctuation data reaches the standard of severe fluctuation in the three dimensions of fluctuation intensity, rate discreteness and time duration. The change amplitude deviation threshold can be calculated by the difference between the maximum value and the minimum value of the same type of sensor in each collection period under normal state, and a certain multiple of the deviation from the mean value is taken as the judgment basis. The change rate distribution threshold is based on the change interval sequence of historical fluctuation data, and by analyzing the distribution characteristics of the change interval width in the frequency distribution map, the upper limit of the acceptable fluctuation range is selected as the standard combined with engineering experience. The change duration threshold is set as the lower limit value that can stably reflect the duration of abnormality by simulating a variety of known normal and abnormal states of high fluctuation samples, extracting the time length of continuous high change rate segments, and combining the tolerance of false triggering. The three thresholds are derived from the cooperative calculation of historical data statistics, experimental verification and warning false alarm tolerance, ensuring the engineering adaptability and judgment accuracy of the judgment standard.

[0028] In order to realize the accurate judgment of whether the high fluctuation data constitutes a severe fluctuation, it is necessary to quantitatively determine from three complementary dimensions. The change amplitude deviation value is used to measure the deviation degree of the current data fluctuation intensity relative to the historical level, reflecting the suddenness of abnormal fluctuation; the change rate distribution index reflects the severity of data fluctuation in a short time through the frequency concentration interval and the change interval width, evaluating the discreteness of dynamic change; the change duration length measures whether the high intensity fluctuation has continuity in the time dimension, revealing the stable abnormal characteristics of fluctuation. Comparing the three indexes with the set change amplitude deviation threshold, change rate distribution threshold and change duration threshold, a full-dimensional judgment of the intensity, rate and continuity of a single fluctuation segment can be formed. Only when all three indexes exceed the corresponding threshold, it can be determined as a truly disturbing severe fluctuation, so as to mark the corresponding sensor as an abnormal signal source, avoiding false judgment due to local random fluctuation. This multi-dimensional cross-validation method enhances the accuracy of judgment and the anti-noise ability of the system, and is especially suitable for damage assessment and warning scenarios with complex interference factors in multi-source sensor data.

[0029] In this embodiment, S201 is specifically: The maximum value and the minimum value in the current collection period are extracted for each identified high fluctuation data, and the difference between them is calculated as the current change amplitude value; After obtaining the identified high fluctuation data of each segment, the start and end time points of the current collection period are first determined, all measurement values in the segment are extracted within the time period, the maximum and minimum values are extracted by traversing, and the maximum value is subtracted from the minimum value to calculate the current change amplitude value. The purpose of this operation is to quantify the response fluctuation range of the data segment within the time interval. For example, if a sensor records data [3.2, 4.1, 3.8, 5.6, 4.9] in a certain collection period, the maximum value is 5.6, the minimum value is 3.2, and the difference between the two is 2.4, i.e. the current change amplitude value of the segment is 2.4. This amplitude value can be used for subsequent deviation analysis with the historical average fluctuation level to determine whether the segment belongs to the abnormal fluctuation range. In this technical feature, the "current collection period" defines the analysis time range, the "difference between the maximum value and the minimum value" constitutes a quantitative discrimination index, and the "change amplitude value" serves as an evaluation benchmark for subsequent credibility derivation. This process can be achieved through sliding window traversal, efficient numerical extraction algorithm or matrix screening method, and is suitable for various sensor types and data structures.

[0030] Select multiple time periods of the same length as the current collection period from the historical data of the same type of sensor, calculate the difference between the maximum and minimum values in each time period, and calculate the arithmetic mean of these differences as the average difference. When analyzing the amplitude deviation of high fluctuation data, a reference standard based on historical data needs to be established. First, divide the historical records of the same type of sensor into multiple non-overlapping historical periods according to the same length as the current collection period. Extract the maximum and minimum values of the sensor in each period and calculate their difference as the historical fluctuation amplitude of that period. Then, sum up the fluctuation amplitude values of all historical periods and divide by the total number of periods to obtain the arithmetic mean, which is used for subsequent comparison with the current change amplitude value. For example, if the current collection period is 5 minutes, select three consecutive 5-minute periods from the historical data, and the difference between the maximum and minimum values of each period is 1.5, 2.0, and 1.8, respectively, then the average difference is (1.5+2.0+1.8) / 3 = 1.77. In the technical feature, "the same type of sensor" ensures consistent data sources, "the same length of the collection period" ensures time scale alignment, "the difference between the maximum value and the minimum value" serves as a local fluctuation measure, and "the arithmetic mean" provides a stable reference benchmark. This calculation process can be achieved through window sliding, extreme value function extraction and mean value operation module to ensure calculation accuracy and processing efficiency.

[0031] Subtract the average difference from the current change amplitude value to obtain the change amplitude deviation value, which is used for subsequent severe fluctuation determination.

[0032] To quantify the deviation of the current data fluctuation from the historical normal fluctuation level, the current change amplitude value is subtracted from the historical average difference value. The specific implementation is as follows: first, obtain the difference between the maximum value and the minimum value of the identified high fluctuation data in the current collection period as the current change amplitude value; then obtain the average fluctuation amplitude of the same type of sensor in multiple equal time periods from the historical data as the average difference value; subtract the average difference value from the current change amplitude value, and the calculated value is the change amplitude deviation value, which is used to represent the abnormality of the current fluctuation. For example, if the current change amplitude value is 2.5 and the average difference value is 1.8, the change amplitude deviation value is 2.5-1.8=0.7. The larger the value, the more intense the current fluctuation compared to the historical level, and the more likely it is an abnormal signal. In the technical features, the "current change amplitude value" is derived from the target collection period, and the "average difference value" is derived from the statistical features of the historical data. The difference between the two is an important parameter for determining the significance of fluctuation. This calculation process can be completed through basic arithmetic subtraction operation and is the core input basis for subsequent multi-dimensional threshold judgment.

[0033] In this embodiment, S202 is specifically: The numerical difference between each two adjacent data points of each identified high fluctuation data is calculated in sequence, and the change interval sequence composed of all the difference values is counted. To extract the change characteristics of high fluctuation data in the time sequence, the difference between adjacent points of each identified high fluctuation data is calculated to construct the change interval sequence. The implementation is as follows: arrange the data in time sequence, select each two adjacent data points in sequence, calculate the difference between the values, and keep the positive and negative directions of the difference value, record one by one, and finally form a complete change interval sequence. For example, the continuous measurement value of a certain segment of high fluctuation data is [2.1, 2.4, 2.0, 1.9, 2.3], the difference between adjacent data points is [+0.3, -0.4, -0.1, +0.4], and the difference value sequence is the change interval sequence of the segment of high fluctuation data. The sequence not only reflects the amplitude of data fluctuation, but also embodies the directionality of change, which can be used for subsequent frequency statistics and distribution analysis. In the technical features, the "difference between adjacent data points" reflects the micro change trend, and the "change interval sequence" is the basis for constructing the change rate distribution map and has the ability to quantitatively describe high-frequency fluctuation behavior. This calculation process can be realized in the data processing program through loop difference function or sliding window subtraction.

[0034] The change interval sequence is counted and a histogram distribution graph is drawn to generate a change rate distribution map. To comprehensively describe the change rate characteristics of high-volatility data, it is necessary to count the frequencies of the change interval sequence and generate a change rate distribution map. The specific implementation is as follows: first, set a fixed numerical interval width, divide each difference value in the change interval sequence into the corresponding interval according to its absolute value, and count the number of difference values in each interval to obtain the frequency value of each interval; then, take the interval as the horizontal axis and the frequency as the vertical axis to draw a histogram and form a frequency distribution graph of the change rate. For example, if the difference value range is from -0.6 to +0.6 and the interval width is set to 0.2, the statistical interval is [-0.6~-0.4], [-0.4~-0.2], …, [+0.4~+0.6], and the number of difference values in these intervals is counted respectively. Through this method, the concentration trend and volatility amplitude of the difference value can be intuitively displayed, and a complete change rate distribution map can be generated. In the technical features, “frequency counting” is used to quantify the frequency of occurrence of different rates of change, “histogram distribution graph” provides an intuitive expression form of rate distribution, and “change rate distribution map” provides a structured basis for subsequent extraction of the concentration interval and change width. This process can be completed by combining the histogram calculation algorithm and the visualization drawing function.

[0035] In the change rate distribution map, the continuous numerical interval with the highest frequency is identified as the frequency concentration interval, and the numerical difference between the minimum change value and the maximum change value in the distribution map is calculated as the change interval width.

[0036] In the change rate distribution map, to extract the statistical region that best reflects the concentration characteristics of data volatility, the continuous numerical interval with the highest frequency needs to be identified and the change interval width needs to be calculated. The specific implementation is as follows: first, according to the frequency value of each numerical interval in the histogram, traverse all adjacent interval combinations, find the set of continuous intervals with the maximum cumulative frequency value, and take the numerical range corresponding to this set of intervals as the frequency concentration interval. Then identify the minimum boundary value and the maximum boundary value in this set of intervals, and calculate the difference between the two as the change interval width. For example, if the frequency concentration interval is [-0.2, 0.2], the change interval width is 0.4. In the technical features, “the continuous numerical interval with the highest frequency” is used to reflect the main distribution range of data volatility, and “the numerical difference between the minimum change value and the maximum change value” represents the extension degree of volatility. This process can be realized by using the sliding window and frequency accumulation judgment strategy to accurately define the core region of the rate of change.

[0037] In this embodiment, S203 specifically comprises: Compare the change rate sequence corresponding to each identified high-volatility data segment with the change rate determination reference point by point, and mark the continuous sampling point sequence with a value greater than the change rate determination reference; In order to determine the duration characteristics of high-volatility data, the change rate sequence corresponding to each identified high-volatility data segment needs to be compared point by point with a preset change rate judgment criterion. The specific implementation is: each value in the change rate sequence is read in turn and compared with the set change rate judgment criterion, if the value is greater than the judgment criterion, the corresponding time point is marked as a high change point; if a plurality of consecutive sampling points satisfy the condition, the sequence of consecutive time points is marked as a high change sampling point sequence. In the technical features, the "change rate sequence" reflects the time sequence change speed of the sensor data, the "change rate judgment criterion" is a quantitative standard for high-volatility behavior, the "point-by-point comparison" ensures that each sampling time is accurately evaluated, and the marking of the "sequence of consecutive sampling points" provides a basis for subsequent calculation of the volatility duration. This process can be realized by setting logical conditions for iterative comparison and using a Boolean array for marking.

[0038] The number of consecutive adjacent sampling points in each marked sampling point sequence is counted, and the number is multiplied by the sampling interval to obtain the time length in each sequence that maintains a change rate greater than the change rate judgment criterion; In order to calculate the time length in each marked sampling point sequence that the change rate is continuously higher than the change rate judgment criterion, the number of adjacent points of each group of continuously marked "high change" sampling points needs to be counted, and the specific way is: scan the Boolean marking array, when encountering a continuous "true" value paragraph, record its start and end position, count the number of sampling points contained in the paragraph, and multiply the sampling interval time to obtain the duration of the paragraph. For example, if a high change segment consists of 5 adjacent sampling points, and the sampling interval is 0.1 seconds, the duration of the segment is 0.5 seconds. In the technical features, the "number of consecutive adjacent sampling points" represents the number of continuous sampling points in the high-volatility state, and the "sampling interval" reflects the data acquisition frequency, and their product can be converted to a physical time scale, thereby realizing accurate quantification of the duration of the violent fluctuation. This statistical process can be realized by using the loop accumulation method combined with state transition judgment logic.

[0039] The maximum value is selected from all time lengths as the change duration index of the currently identified high-volatility data.

[0040] After the corresponding change duration is calculated for each sequence of continuous high-change sampling points identified in the high-volatility data segment, a representative value needs to be selected from these time lengths to reflect the persistence of the violent fluctuations. The specific way is: traverse all the calculated duration lengths, compare their numerical values, and select the largest one as the change duration length index of the high-volatility data segment. This index reflects the longest time interval of continuous violent change in the entire high-volatility process, and is an important basis for judging whether it constitutes violent fluctuations. In the technical features, the "time length" comes from the previous change rate and threshold comparison result, and the "maximum value" as the upper limit of change persistence helps to avoid underestimating the volatility intensity due to short-term abnormalities. Quick extraction can be achieved through array traversal or built-in maximum value function, ensuring the accuracy and timeliness of subsequent judgment.

[0041] S3, calculating change direction consistency, fitting slope similarity and fitting residual range in the time period corresponding to the marked abnormal signal source for other sensor data, other sensors referring to sensors not marked as abnormal signal sources; In this embodiment, S3 specifically includes the following steps: S301, selecting sensor data not marked as abnormal signal sources in the time period corresponding to the sensor marked as abnormal signal source, constructing time sequence response trajectories of each sensor in the time period, and extracting the change direction of each response trajectory; In a unified collection period, first, the time period corresponding to the sensor that has been marked as an abnormal signal source needs to be determined, and then all sensor data not marked as an abnormal signal source is selected in the time period. For each sensor not marked as an abnormal signal source, the continuous data points in the time period are extracted in the order of sampling time, and a time sequence response trajectory with time as the horizontal axis and sensor data as the vertical axis is constructed. The trajectory is used to describe the response change process of the sensor in the time period. After all the trajectories are constructed, the numerical difference between adjacent sampling points in each trajectory is extracted one by one, and the change direction between each two sampling points is judged according to the positive and negative signs of the difference, wherein a positive value indicates an upward direction, a negative value indicates a downward direction, and a zero value indicates that the direction remains unchanged. Finally, the change direction sequence of each sensor is retained for subsequent comparison and analysis with the change direction of the abnormal signal source. The whole process can provide direction feature basis for trend consistency calculation, while ensuring that the constructed time sequence response trajectories have comparability and consistency in time and data dimensions.

[0042] S302, comparing the change direction of each sensor not marked as an abnormal signal source with the change direction of the sensor marked as an abnormal signal source point by point, and counting the proportion of data points with consistent direction as the change direction consistency; S303, respectively, on all the sensor response trajectory which is not marked as an abnormal signal source, linear least squares regression is performed, the fitting slope and the fitting residual are obtained, and then compared with the fitting result of the sensor marked as an abnormal signal source, the fitting slope similarity and the fitting residual difference are calculated respectively; S304, according to the change direction consistency, the fitting slope similarity and the fitting residual difference, the weight coefficient is assigned, and the trend stability score value is generated in the weighted average way, which is used to judge whether to perform weight inhibition processing on the abnormal signal source.

[0043] When generating the trend stability score value, first, the change direction consistency, the fitting slope similarity and the fitting residual difference three indexes need to be unified dimension processing, so that their value range is in the same range, for example, by normalizing to 0 to 1 interval. Subsequently, each index is assigned a weight coefficient, which can be set according to the relative importance of each index to the trend judgment in actual application, for example, the change direction consistency weight is 0.4, the fitting slope similarity weight is 0.3, and the fitting residual difference weight is 0.3. By multiplying each normalized index value by the corresponding weight coefficient and summing, the weighted average value is calculated as the trend stability score value. For example, if the normalized values of the three indexes are 0.9, 0.7 and 0.6 respectively, the score value is 0.9x0.4 + 0.7x0.3 + 0.6x0.3 = 0.78. The trend stability score value is used to quantify the consistency of the abnormal signal source and other sensors in the trend behavior, the higher the value, the stronger the overall consistency, the higher the reliability; if the score value is lower than the preset threshold, it indicates that the abnormal signal may deviate significantly from other data sources, and it is suggested that the evaluation weight of the abnormal signal should be inhibited in the subsequent fusion calculation, so as to reduce the misjudgment risk caused by local abnormality and improve the stability and accuracy of the early warning system.

[0044] In this embodiment, S302 is specifically: The change direction sequence of the adjacent data points of the sensor marked as an abnormal signal source in the corresponding time period is extracted, and the change direction is determined by the difference sign of the adjacent data points; The continuous response data collected by the sensor marked as an abnormal signal source in a certain time period is extracted and arranged in time sequence. For each pair of adjacent data points in the time sequence, the numerical difference between the current data point and the previous data point is calculated, and the change direction is determined by the sign of the difference: if the difference is positive, the sensor value shows an upward trend; if the difference is negative, the sensor value shows a downward trend; if the difference is zero, the sensor value remains unchanged in this period. The change direction between each pair of adjacent sampling points is recorded in sequence to form a complete change direction sequence. The change direction sequence is used to describe the dynamic change trend of the abnormal signal source in the target time period, and serves as the basis for subsequent consistency comparison with other sensors. The accuracy and direction judgment accuracy directly affect the judgment result of trend stability. Through this way, the fine expression of the response change mode of the abnormal signal source in the time dimension can be realized.

[0045] The change direction sequence of the sensor not marked as an abnormal signal source in the same time period is extracted, and the time points are in one-to-one correspondence with the abnormal signal source direction sequence; In a certain time period, the response data of all sensors not marked as an abnormal signal source is extracted one by one, and the data of each sensor is arranged in time sequence to construct the corresponding time sequence. The difference value operation between adjacent data points is performed for each time sequence, and the change direction is determined based on the difference value sign to obtain the corresponding change direction sequence. The direction points in each change direction sequence must be strictly aligned with the change direction sequence of the sensor marked as an abnormal signal source on the time axis, ensuring that the data direction of different sensors at the same time point has a one-to-one correspondence. This one-to-one correspondence can be realized through a unified sampling index or time stamp matching mechanism, ensuring that the change trend of each sensor in the same time window has comparability in the subsequent direction consistency comparison. This method ensures the accuracy and logical integrity of the direction consistency index statistics, and is a prerequisite for building the trend stability score.

[0046] The directions of each pair of corresponding data points in the two change direction sequences are compared, the number of data points with consistent directions is counted, and the number is divided by the total number of data points to calculate the direction consistency proportion, which is used as the change direction consistency index.

[0047] For each pair of change direction sequences, by comparing the change directions of the corresponding data points of the two sensors at the same time point, it is judged whether the directions are consistent. The condition for the directions to be consistent is that the difference values of the two data points have the same sign, that is, both are positive or both are negative. The number of data points that satisfy the consistent condition at all time points is counted and recorded as the consistent number. Then, the consistent number is divided by the total number of data points in the time period to calculate the proportion of the direction consistency. For example, in a time period containing 10 sampling points, if the change directions of the two sensors are completely consistent at 8 time points, then the proportion of the direction consistency is 8 divided by 10, and the result is 0.8. The proportion value is used as a quantitative index to measure the synchronization degree of the change trend between different sensors. The higher the value, the closer the abnormal signal is to the trend direction of other sensors, which helps to judge the trend stability subsequently. The comparison method can be realized by comparing the symbol arrays, which has the characteristics of clear calculation logic and simple operation.

[0048] In this embodiment, S303 is specifically: A linear least squares fitting is applied to the response trajectory of each sensor that is not marked as an abnormal signal source in the target time period to obtain the corresponding fitting slope and residual mean square value; When performing linear least squares fitting on the response trajectory collected by the sensor that is not marked as an abnormal signal source in the target time period, first, a set of time-value pair coordinate sequences is constructed with the sampling time as the independent variable and the sensor data value as the dependent variable. In linear least squares fitting, the goal is to determine a straight line y=ax+b that minimizes the sum of the squares of the perpendicular distances of all points to the line. The slope a can be obtained by calculating the covariance and variance of the sequence, and the intercept b can be determined in combination with the mean value relationship. After fitting, the difference between each actual sampling value and the estimated value at the corresponding time point on the fitted straight line is used to calculate the square of all differences and the average value, which is the residual mean square value. For example, in the case of a time sequence of [1, 2, 3, 4, 5] and response data of [2.1, 2.5, 2.9, 3.2, 3.8] for a sensor, the slope obtained by linear fitting is about 0.42, and the residual mean square value reflects the degree of deviation between the data and the linear fitting. The smaller the value, the closer the data is to the linear trend. This method has a clear mathematical foundation and repeatability, and is suitable for trend extraction of continuous time series data.

[0049] The same fitting process is performed on the response trajectory of the sensor marked as an abnormal signal source to obtain the corresponding fitting slope and residual mean square value as a comparison reference; When fitting the sensor response trajectory marked as an abnormal signal source, first extract its continuous sampling data in the target time period, take the sampling time as the independent variable and the response value as the dependent variable, and construct the time-value data point set. Apply linear least squares method to establish the fitting model, calculate the covariance and variance of the time series and data series, and obtain the fitting slope, that is, the rate of change of the dependent variable caused by the change of the independent variable per unit. Then, combined with the mean value relationship, determine the fitting intercept to form a complete linear expression. Subsequently, according to the difference between the actual sampling point value and the fitting value, square the difference point by point and take the average to obtain the residual mean square value, which is used to reflect the deviation degree of the fitting. For example, if the time series of a certain abnormal sensor is [1, 2, 3, 4, 5] and the response value is [5.0, 7.0, 6.5, 9.0, 8.2], the fitting slope may be 0.8 and the residual mean square value is 0.36, indicating that the data fluctuates relatively small but still has slight instability on the overall linear trend. The fitting slope and residual mean square value serve as the comparison benchmark in the subsequent multi-source sensor trend comparison, reflecting the change pattern of the abnormal signal source in the time series.

[0050] Calculate the average of the difference between the fitting slope of all sensors not marked as abnormal signal sources and the comparison benchmark slope as the fitting slope similarity; extract the maximum difference between the fitting residual of all sensors not marked as abnormal signal sources and the comparison benchmark residual as the fitting residual range.

[0051] When calculating the average of the difference between the fitting slope of all sensors not marked as abnormal signal sources and the comparison benchmark slope, first perform linear least squares fitting on each sensor not marked as abnormal signal source to obtain the corresponding fitting slope set, then calculate the difference between each fitting slope and the fitting slope of the sensor marked as abnormal signal source, take the absolute value of the difference and calculate the arithmetic mean as the fitting slope similarity, which is used to measure the closeness of the overall trend between the two slope sets. For example, the comparison benchmark slope is 0.8 and the fitting slopes of the other three sensors are 0.7, 0.6 and 0.9 respectively, the absolute value of the difference is 0.1, 0.2 and 0.1 respectively, and the average is 0.133, which is taken as the fitting slope similarity. When calculating the fitting residual range, first obtain the residual mean square value of each sensor not marked as abnormal signal source, then calculate the absolute value of the difference between these values and the comparison benchmark residual mean square value, and extract the maximum value as the fitting residual range. For example, the benchmark residual is 0.36 and the other residuals are 0.40, 0.55 and 0.33 respectively, the absolute value of the difference is 0.04, 0.19 and 0.03 respectively, and the maximum value is 0.19, which is taken as the fitting residual range, used to measure the maximum deviation of the fluctuation degree of each sensor. The two indicators are used together to judge the similarity of the abnormal signal source in trend and fitting error with other sensors.

[0052] S4, mapping the result of the severe fluctuation judgment and the trend stability score value to a credibility value as a basis for the participation reliability of the abnormal signal source; In this embodiment, S4 is specifically: The result of the severe fluctuation judgment is mapped to a numerical magnitude, and is set to a first preset numerical value when the severe fluctuation judgment is established, and is set to a second preset numerical value when the severe fluctuation judgment is not established, forming a severe fluctuation numerical factor. The result of the severe fluctuation judgment is usually represented in Boolean logic form, that is, true when judged as severe fluctuation, otherwise false. In order to use this result for subsequent credibility value calculation, the Boolean result needs to be mapped to a numerical form. The specific way is to set two different preset numerical values, for example, map the true result to the numerical value 1, and map the false result to the numerical value 0. In this way, the logical judgment result can be converted into a quantifiable numerical input. Taking a sensor data as an example, if its change amplitude deviation value, change rate distribution index and change duration length index are all greater than the corresponding threshold, it is judged as severe fluctuation, and the corresponding mapping numerical value is 1; if it does not meet the judgment condition, the mapping numerical value is 0. Define this numerical result as a severe fluctuation numerical factor, which can be one of the input parameters for joint calculation of credibility, so that the identification result of severe fluctuation participates in the subsequent weighted calculation logic in the form of a numerical value. This numerical factor has clear distinguishability, which is conducive to reflecting the severity of signal anomaly in the subsequent scoring model.

[0053] Interval normalization processing is performed on the trend stability score value to convert the score value into a uniform standard trend stability factor, ensuring that it is additively fused with the severe fluctuation numerical factor under the same numerical scale; The trend stability score value is usually derived from the weighted average calculation of multiple characteristic indexes (change direction consistency, fitting slope similarity, fitting residual range). Its original value may be in a non-uniform numerical range, for example, between 0 and 10. In order to make the trend stability score value and the severe fluctuation numerical factor (such as 0 or 1) fused under the same numerical scale, interval normalization processing needs to be performed on the trend stability score value. The implementation of interval normalization can adopt a linear normalization formula: subtract the minimum score value from the original score value, and divide by the difference between the maximum and minimum values, so as to compress the result to the standard interval of 0 to 1. For example, if the minimum historical score is 2 and the maximum score is 8, and the current score is 5, the normalized result is (5-2) / (8-2)=0.5, and finally the trend stability score value is converted into a standardized trend stability factor. This factor has the same scale characteristics as the severe fluctuation numerical factor, ensuring that both have additivity and relative weighting in subsequent credibility fusion calculation. Uniform numerical scale helps to ensure the numerical consistency of the reliability judgment model, and improves the interpretability and controllability of the credibility value.

[0054] The severe fluctuation value factor and the trend stability factor are taken as joint input variables, a weighted linear combination function is performed, and a reliability value is output. The reliability value is used as the basis for the reliability of the corresponding abnormal signal source in subsequent information fusion calculation.

[0055] The severe fluctuation value factor and the trend stability factor respectively reflect the reliability evaluation basis of the abnormal signal source under the mutation characteristics and trend background. By taking these two factors as joint input variables, a reliability value model can be constructed through a weighted linear combination function, wherein each factor corresponds to a weight parameter for adjusting the influence degree on the final reliability result. The weighted linear combination function can be expressed as: reliability value = α × severe fluctuation value factor + β × trend stability factor, wherein α and β are preset weight coefficients, satisfying α + β = 1, for controlling the relative contribution proportion of the two types of information. For example, when the severe fluctuation value factor is 1 (indicating strong fluctuation) and the trend stability factor is 0.3 (indicating unstable trend), and α = 0.7 and β = 0.3 are set, then the reliability value = 0.7 × 1 + 0.3 × 0.3 = 0.79. This value can be used as the participation weight basis for the reliability of the abnormal signal source data in subsequent multi-source information fusion calculation, realizing quantitative measurement and dynamic regulation of the signal source reliability. By linearly combining two evaluation factors of different sources, the accuracy and robustness of abnormal identification decision can be improved.

[0056] S5, based on the reliability value, adjusting the participation weight of the abnormal signal source in the fusion evaluation calculation, generating the structure damage evaluation result by weighted fusion of the evaluation parameters of each sensor, and completing the warning output.

[0057] In this embodiment, S5 is specifically: Based on the reliability value, the participation weight of the sensor marked as an abnormal signal source in the fusion evaluation calculation is adjusted, the participation weight of the sensor not marked as an abnormal signal source is set to a full weight value, and the participation weight of the sensor marked as an abnormal signal source is equal to the corresponding reliability value. This process is to control the influence of abnormal signal source on the evaluation result of structure damage by dynamic adjustment of the participation weight of sensor, the purpose is to improve the reliability of overall evaluation. In the implementation, first of all, each sensor needs to be classified in the current evaluation period, identify which sensors have been marked as abnormal signal source. For sensors not marked as abnormal signal source, directly give a fixed full weight value, for example 1.0, indicating that its data is fully accepted in the fusion calculation. For sensors marked as abnormal signal source, according to the credibility value generated in the previous process, the value is used as the participation weight in the fusion. For example, if the credibility of an abnormal sensor is 0.72, its evaluation parameter only participates in the weighted fusion of structure damage index with a proportion of 0.72. This mechanism ensures that the reference value of abnormal signal is retained in information fusion, and the interference caused by its uncertainty is controlled, which improves the accuracy and stability of structure damage identification.

[0058] In the implementation, first of all, all sensors are numbered and a one-to-one weight vector is established, then the abnormal marking list and the corresponding credibility result in the current period are read, the sensor set is traversed in turn, the unmarked sensors are directly assigned to 1.0, and the marked sensors are extracted from the credibility mapping table to fill in the weight vector. After the weight vector is updated, it is multiplied by the current evaluation parameter of each sensor to generate a weighted evaluation value vector, which is used as the input parameter of the subsequent structure damage index fusion calculation. In the whole process, the credibility value and the state marking of the sensor constitute the core basis of weight adjustment, which supports the system to maintain the stability and credibility of the overall evaluation result when facing local abnormal signal source interference.

[0059] Extract all sensor evaluation parameters in the current evaluation period, multiply each sensor evaluation parameter with its corresponding participation weight to obtain the weighted evaluation value of each sensor, and aggregate all weighted evaluation values by weight normalization to generate the structure damage evaluation result; In structural damage assessment, in order to ensure that the fusion result fully reflects the numerical contribution and participation credibility of each sensor, the evaluation parameters and corresponding participation weights of all sensors need to be weighted and fused. The specific way is to first extract the evaluation parameter values calculated by each sensor in the current evaluation period, and construct a set of evaluation parameter vectors, and prepare a participation weight vector of the same length, which contains the fixed value of the full weight sensor and the credibility value of the abnormal signal source sensor. By multiplying each evaluation parameter and its corresponding weight one by one, the weighted evaluation value of each sensor is obtained. In order to ensure that the final fusion result is consistent in numerical scale and has comparability, it is necessary to normalize and aggregate all weighted evaluation values, that is, divide the weighted sum of all weighted evaluation values by the total weight sum, so as to generate a standardized structural damage assessment result, which can be used to judge the health state level of the target structure and drive subsequent warning decisions.

[0060] In a specific implementation, two one-dimensional arrays of the same length are constructed to store the original evaluation parameters of each sensor in the current period and their participation weights, for example, the parameter array is P1, P2,..., Pn, and the weight array is W1, W2,..., Wn. The corresponding weighted evaluation value array is P1xW1, P2xW2,..., PnxWn. Then, the weighted evaluation value array is summed, denoted as the total weighted evaluation value, and the weight array is summed, denoted as the total weight value. The final structural damage assessment result is equal to the total weighted evaluation value divided by the total weight value. For example, if the evaluation values of three sensors are [0.8, 0.6, 0.4], the corresponding weights are [1.0, 0.85, 0.72], the weighted values are [0.8, 0.51, 0.288], the weighted sum is 1.598, the total weight is 2.57, and the structural damage assessment result is 1.598 divided by 2.57, which is about 0.622. This mechanism can ensure that the fusion result not only considers the evaluation value itself, but also incorporates the credibility control logic of the data source.

[0061] The structural damage assessment result is compared with the damage judgment threshold value. When the structural damage assessment result is greater than the damage judgment threshold value, the structural damage warning signal is output, and the structural positioning information, damage level value and participation weight information in the current evaluation period are also output, completing the warning output.

[0062] In structural health monitoring, in order to respond to potential risks in time, it is necessary to compare the current period of structural damage assessment results with the pre-set damage judgment threshold. The judgment threshold is determined according to historical data, structural material properties, environmental factors and empirical models, and represents the maximum acceptable damage tolerance of the structure. Once the structural damage assessment result exceeds the threshold, it is considered that the current structure is in a potentially dangerous state, and the early warning mechanism needs to be triggered immediately. At this time, the key data related to the assessment results need to be output synchronously, including the spatial positioning information of the structure corresponding to the evaluation time point, which is used to indicate the specific damage area; the damage level value evaluated, which is used to quantify the risk degree; and the participation weight information of each sensor in this fusion, which is used to further assist the credibility traceability and subsequent analysis. This mechanism constitutes a complete closed loop of early warning output, which not only provides the judgment result, but also supports the response decision.

[0063] In the specific implementation, first, the calculated structural damage assessment result is denoted as a numerical value D, and the pre-set damage judgment threshold is T. When D is greater than T, the system calls the early warning submodule to trigger the alarm signal generation. The positioning information is traced back through the sensor layout area index and timestamp, combined with the geographical identification of the identified abnormal signal source, to output the accurate structural damage area number; the damage level value can be mapped to the qualitative level (such as mild, moderate, severe) according to the segmentation rule in the 0~1 interval; the participation weight information is composed of the weight values of each sensor participating in the fusion in the current period, which is output in a structured array manner and can be used to assist interface display or subsequent feedback learning optimization. In this mechanism, the accuracy, explainability and operability of the early warning response are ensured.

[0064] The above-described embodiments can be implemented in part or in whole through software, hardware, firmware or any combination thereof. When implemented in software, the above-described embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When loaded and executed by a computer, the computer instructions or computer programs cause the computer to perform all or part of the processes or functions described in the embodiments of the present application. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable apparatus. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through a wired or wireless (for example, infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0065] It should be understood that the size of the sequence number of each process described above in various embodiments of the present application does not mean the order of execution, and the execution order of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0066] Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0067] In several embodiments provided in the present application, it should be understood that the disclosed system and method can be implemented in other ways. For example, the above-described embodiments are only illustrative, for example, the division of units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed each other can be indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0068] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may also be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0069] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0070] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for damage assessment and early warning of ultra-high performance concrete masonry slabs based on multi-source sensor data, characterized in that, Specifically, the following steps are included: S1. Construct a time-series response trajectory set from the continuous data acquired by multiple sensors within a unified acquisition period, and extract the rate of change index of each sensor to identify high-fluctuation data. S2. Calculate the deviation of the change amplitude, the distribution index of the change rate, and the change duration index for the identified high-fluctuation data. Use multi-dimensional thresholds to determine whether it constitutes a violent fluctuation and mark it as an abnormal signal source. S3. Calculate the consistency of change direction, similarity of fitting slope, and range of fitting residuals for other sensor data within the time period corresponding to the marked abnormal signal source, and use them to generate a trend stability score. Other sensors refer to sensors that have not been marked as abnormal signal sources. S4. The results of the judgment of violent fluctuations are jointly mapped with the trend stability score to a credibility value, which serves as the basis for the reliability of participation of abnormal signal sources. S5. Adjust the participation weight of abnormal signal sources in the fusion evaluation calculation based on the credibility value, generate structural damage evaluation results by weighted fusion of the evaluation parameters of each sensor, and complete the early warning output.

2. The method for damage assessment and early warning of ultra-high performance concrete slabs based on multi-source sensor data according to claim 1, characterized in that, S2 specifically includes the following steps: S201. Calculate the difference between the maximum and minimum values ​​for each segment of highly fluctuating data, and compare it with the average difference of the corresponding data of the same type of sensor in the historical acquisition period to obtain the deviation value of the change range. S202. For each segment of highly fluctuating data, the numerical change interval between adjacent data points is statistically analyzed, a change rate distribution map is constructed, and the frequency concentration interval and the width of the change interval are extracted as change rate distribution indicators. S203. Calculate the length of time that each segment of identified high-fluctuation data remains above the rate of change judgment benchmark in a continuous sampling period, and use it as an indicator of the duration of change. S204. Compare the deviation value of the change amplitude with the change amplitude deviation threshold, compare the change rate distribution index with the change rate distribution threshold, and compare the change duration with the change duration threshold. When all three comparison results are greater than the threshold, the high fluctuation data is identified as a violent fluctuation, and the corresponding sensor is marked as an abnormal signal source.

3. The method for damage assessment and early warning of ultra-high performance concrete slabs based on multi-source sensor data according to claim 2, characterized in that, S201 specifically refers to: For each segment of highly volatile data, extract the maximum and minimum values ​​within the current acquisition period, and calculate the difference as the current change amplitude value; In the historical data of the same type of sensor, select multiple time periods with the same acquisition cycle length, calculate the difference between the maximum and minimum values ​​in each time period, and take the arithmetic mean of these differences as the average difference. Subtract the average difference from the current change value to obtain the change deviation value, which is used to determine subsequent drastic fluctuations.

4. The method for damage assessment and early warning of ultra-high performance concrete slabs based on multi-source sensor data according to claim 2, characterized in that, S202 specifically refers to: For each segment of highly volatile data, the numerical difference between every two adjacent data points is calculated sequentially, and the change interval sequence composed of all differences is statistically analyzed. The frequency statistics of the changing interval sequence are statistically analyzed and a histogram is plotted to generate a rate of change distribution map. In the rate of change distribution spectrum, the interval of continuous values ​​with the highest frequency is identified as the frequency concentration interval, and the difference between the minimum and maximum change values ​​in the distribution spectrum is calculated as the width of the change interval.

5. The method for damage assessment and early warning of ultra-high performance concrete slabs based on multi-source sensor data according to claim 2, characterized in that, S203 specifically refers to: Each segment of high-fluctuation data is compared point by point with the rate of change judgment benchmark, and the continuous sampling point sequence with values ​​greater than the rate of change judgment benchmark is marked. In each sequence of marked sampling points, the number of consecutive adjacent sampling points is counted, and this number is multiplied by the sampling interval to obtain the length of time in each sequence where the rate of change is greater than the threshold for determining the rate of change. The maximum value among all time periods is selected as the indicator of the duration of change in the currently identified high-volatility data.

6. The method for damage assessment and early warning of ultra-high performance concrete slabs based on multi-source sensor data according to claim 1, characterized in that, S3 specifically includes the following steps: S301. Select the sensor data that is not marked as an abnormal signal source within the time period corresponding to the sensor marked as an abnormal signal source, construct the time-series response trajectory of each sensor within the time period, and extract the change direction of each response trajectory. S302. Compare the direction of change of each sensor that is not marked as an abnormal signal source with the direction of change of the sensor that is marked as an abnormal signal source point by point, and count the percentage of data points with the same direction as the consistency of the direction of change. S303. Perform linear least squares regression on the response trajectories of all sensors that are not marked as abnormal signal sources, obtain the fitting slope and fitting residual, and then compare the fitting results with those of sensors marked as abnormal signal sources. Calculate the similarity of fitting slope and the range of fitting residual respectively. S304. Based on the consistency of the direction of change, the similarity of the fitting slope, and the range of the fitting residual, weight coefficients are assigned respectively, and a trend stability score is generated by weighted averaging to determine whether to perform weighted suppression processing on abnormal signal sources.

7. The method for damage assessment and early warning of ultra-high performance concrete slabs based on multi-source sensor data according to claim 6, characterized in that, S302 specifically refers to: Extract the sequence of changes in the direction of adjacent data points of the sensor marked as an abnormal signal source within the corresponding time period. The direction of change is determined by the sign of the difference between adjacent data points. Extract the change direction sequence of sensors that are not marked as abnormal signal sources within the same time period, and maintain a one-to-one correspondence with the time points of the direction sequence of abnormal signal sources; Compare the directions of each pair of corresponding data points in the two change direction sequences, count the number of data points with the same direction, divide the number by the total number of data points, calculate the direction consistency ratio, and use it as the change direction consistency index.

8. The method for damage assessment and early warning of ultra-high performance concrete slabs based on multi-source sensor data according to claim 6, characterized in that, S303 specifically refers to: For each sensor that was not marked as an anomalous signal source, the response trajectory within the target time period was fitted using the linear least squares method to obtain the corresponding fitting slope and residual mean square value. The same fitting process was performed on the sensor response trajectories marked as abnormal signal sources, and the corresponding fitting slope and residual mean square value were obtained as a comparison benchmark. The mean difference between the fitted slopes of all unmarked anomalous signal sources and the baseline slope is calculated as the fitted slope similarity. Extract the maximum difference between the fitting residuals of all unmarked signal sources and the baseline residuals, and use it as the fitting residual range.

9. The method for damage assessment and early warning of ultra-high performance concrete slabs based on multi-source sensor data according to claim 1, characterized in that, S4 specifically refers to: The result of the drastic fluctuation judgment is mapped to a numerical order of magnitude. When the drastic fluctuation judgment is true, it is set to the first preset value, and when the drastic fluctuation judgment is false, it is set to the second preset value, thus forming a drastic fluctuation numerical factor. The trend stability score is normalized by interval, and the score is converted into a trend stability factor with a unified standard to ensure its additive integration with the drastic fluctuation numerical factor on the same numerical scale. Using the drastic fluctuation numerical factor and the trend stability factor as joint input variables, a weighted linear combination function is executed to output a confidence value. The confidence value serves as the reliability basis for the participation of the corresponding abnormal signal source in subsequent information fusion calculations.

10. The method for damage assessment and early warning of ultra-high performance concrete slabs based on multi-source sensor data according to claim 1, characterized in that, S5 specifically refers to: The participation weight of sensors marked as anomalous signal sources in the fusion evaluation calculation is adjusted based on the confidence level value. The participation weight of sensors not marked as anomalous signal sources is set to the full weight value, and the participation weight of sensors marked as anomalous signal sources is equal to the corresponding confidence level value. Extract the evaluation parameters of all sensors in the current evaluation period, multiply the evaluation parameters of each sensor with their corresponding participation weights to obtain the weighted evaluation values ​​of each sensor, and then normalize and aggregate all weighted evaluation values ​​to generate structural damage evaluation results. The structural damage assessment result is numerically compared with the damage judgment threshold. When the structural damage assessment result is greater than the damage judgment threshold, a structural damage early warning signal is output. At the same time, the structural location information, damage level value and participation weight information within the current assessment period are output to complete the early warning output.

Images