Bridge abnormal state detection method, system, equipment and medium

Abstract

The present invention discloses a bridge abnormal state detection method, system, equipment and medium, including: acquiring real-time monitoring data of a bridge to be tested, constructing a spatiotemporal distribution matrix of the real-time monitoring data of the bridge, and performing data preprocessing operations to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after preprocessing; calculating a safety threshold of the mean value of the monitoring data of the bridge to be tested in a healthy state; using a time sliding window to traverse each row of monitoring data in the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after preprocessing, and obtaining the mean value of the monitoring data in each time sliding window; comparing the mean value of the monitoring data in each time sliding window with the safety threshold of the mean value of the monitoring data of the bridge to be tested in a healthy state, and obtaining the abnormal state detection result of the bridge to be tested; the present invention statistically analyzes the real-time monitoring data to realize rapid detection of abnormal states of the bridge, and the analysis result is interpretable, thereby ensuring safe operation of the bridge.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge safety detection, and in particular relates to a bridge abnormal state detection method, system, equipment and medium. Background Art

[0002] As an important transportation facility, the safe and healthy operation of bridges is related to the lifeline of the national economy. During the operation of in-service bridges, due to the long-term effects of body static loads, traffic dynamic loads, environmental loads, and sudden vehicle and ship collisions, their strength and stiffness will continue to decrease, which may lead to a series of potential structural safety risks such as material corrosion and aging, material performance degradation, and structural damage and disease. When the bridge develops to the point where it is partially or completely unstable and is unable to bear the traffic load, it may lead to safety accidents such as bridge damage and collapse, causing a large amount of property losses and even threatening human life safety.

[0003] At present, methods such as modified bridge finite element models, data-driven and deep learning are commonly used for bridge structure damage identification. Among them, the damage identification method based on the modified bridge finite element model requires an accurate bridge structure finite element model. By modifying the model, the mechanical properties of the modified structural model are kept as consistent as possible with the test results, so as to discover or identify the actual structural damage. However, the actual bridge structure is relatively complex, and multi-objective parallel optimization is often required when modifying the finite element model. There are too many optimization parameters, the optimization process is difficult to converge, and the amount of calculation is too large. However, there are too few bridge finite element models, and the difference between the mechanical properties of the structural model and the actual situation gradually increases with the increase of service time, so this method is rarely used. In addition, the damage identification method based on data-driven analyzes the structural health monitoring data to find outliers and then identify bridge structure anomalies. However, the damage identification method based on data-driven can directly establish a mathematical model and analyze the monitoring data obtained by the health monitoring system, without the need to establish a finite element model, saving time and economic costs, and the intuitiveness and interpretability of the identification results are not high. As artificial intelligence, deep learning and other methods are increasingly applied in the field of civil engineering and have achieved good results; machine learning is good at solving nonlinear multi-classification problems, and its application in the damage identification of complex large-span bridge structures has great scientific value and engineering significance; however, the existing bridge health monitoring systems mostly obtain one-dimensional data, which are characterized by time series, massiveness, and nonlinearity. The complex structural relationship of the bridge leads to complex corresponding relationships between bridge health monitoring parameters, such as strong coupling, non-equivalence, and multi-source heterogeneity. The structural damage identification method based on deep learning has the problems of low damage identification efficiency and poor interpretability of analysis results. Summary of the invention

[0004] In view of the technical problems existing in the prior art, the present invention provides a bridge abnormal state detection method, system, equipment and medium to solve the technical problems of low damage identification efficiency and poor interpretability of analysis results in the existing bridge structure damage identification method.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] The present invention provides a method for detecting abnormal state of a bridge, comprising the following steps:

[0007] Acquire real-time monitoring data of the bridge to be tested, and construct a spatiotemporal distribution matrix of the real-time monitoring data of the bridge; wherein, in the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, each column represents a spatial distribution sequence of all monitoring data at a certain moment, and each row represents a time distribution sequence of monitoring data at a certain measuring point on the bridge to be tested;

[0008] Performing data preprocessing on the spatiotemporal distribution matrix of the real-time monitoring data of the bridge to obtain a preprocessed spatiotemporal distribution matrix of the real-time monitoring data of the bridge;

[0009] Using the historical monitoring data of the bridge to be tested in a healthy state, a safety threshold of the mean value of the monitoring data of the bridge to be tested in a healthy state is calculated;

[0010] Using a time sliding window, traverse each row of monitoring data in the spatiotemporal distribution matrix of the preprocessed real-time monitoring data of the bridge, and perform statistical processing on the monitoring data in each time sliding window to obtain the mean value of the monitoring data in each time sliding window;

[0011] The mean value of the monitoring data in each time sliding window is compared with the safety threshold value of the mean value of the monitoring data of the bridge to be tested in the healthy state to obtain the abnormal state detection result of the bridge to be tested.

[0012] Furthermore, the process of obtaining the real-time sensor data of the bridge to be tested and constructing the spatiotemporal distribution matrix of the real-time monitoring data of the bridge is as follows:

[0013] Deflection sensors are preset at different positions of the bridge to be tested, and a set of real-time monitoring data of the deflection sensors is obtained by sampling once every sampling interval τ, and the spatiotemporal distribution matrix X of the real-time monitoring data of the bridge is constructed;

[0014] The spatiotemporal distribution matrix X of the real-time monitoring data of the bridge is:

[0015]

[0016] Among them, x i,jis the real-time monitoring data of the i-th deflection sensor at the j-th moment; m is the total number of preset deflection sensors at different positions of the bridge to be tested; and n is the total number of samples of each deflection sensor.

[0017] Furthermore, the process of performing data preprocessing operation on the spatiotemporal distribution matrix of the bridge real-time monitoring data to obtain the preprocessed spatiotemporal distribution real-time monitoring data matrix is ​​as follows:

[0018] In the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, each row of monitoring data is traversed to determine whether each row of monitoring data is all zero; if one row of monitoring data is all zero, the monitoring data of the corresponding row is removed from the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after division by zero;

[0019] Using the most probable value filling method, data filling is performed on the time-space distribution matrix of the bridge real-time monitoring data after the division by zero, so as to obtain the time-space distribution matrix of the bridge real-time monitoring data after the data filling;

[0020] Using a wavelet denoising method, the spatiotemporal distribution matrix of the bridge real-time monitoring data after the data filling is subjected to denoising, so as to obtain the spatiotemporal distribution matrix of the bridge real-time monitoring data after the denoising;

[0021] Normalizing the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after the noise reduction processing to obtain a normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge;

[0022] The real-time monitoring data with zero-point drift in the normalized time-space distribution matrix of the real-time monitoring data of the bridge is corrected to obtain the time-space distribution matrix of the real-time monitoring data of the bridge after preprocessing.

[0023] Furthermore, the monitoring data with zero-point drift in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge is corrected to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after preprocessing, which is specifically as follows:

[0024] Obtain historical monitoring data of the bridge to be tested in a zero-point drift state and historical monitoring data of the bridge to be tested in a normal state, respectively calculate the slope and mean of the historical monitoring data of the bridge to be tested in the zero-point drift state, and calculate the slope and mean of the historical monitoring data of the bridge to be tested in a normal state;

[0025] According to the slope and mean of the historical monitoring data of the bridge to be tested in the zero drift state, a slope threshold and a variance threshold for determining whether the monitoring data of the bridge to be tested has zero drift are obtained;

[0026] Using a time sliding window, traverse each row of monitoring data in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge, and calculate the slope and variance of the monitoring data in each time sliding window;

[0027] Comparing the slope of the monitoring data in each time sliding window with a slope threshold of whether the monitoring data of the bridge to be tested has a zero drift, or comparing the variance of the monitoring data in each time sliding window with a variance threshold of whether the monitoring data of the bridge to be tested has a zero drift;

[0028] If the slope of the monitoring data in each time sliding window exceeds the slope threshold of whether the monitoring data of the bridge to be tested has zero drift, or the variance of the real-time monitoring data in each time sliding window exceeds the variance threshold of whether the monitoring data of the bridge to be tested has zero drift, then the corresponding monitoring data has zero drift;

[0029] For the monitoring data with zero point drift, the mean of the monitoring data before the zero point drift occurs is calculated to obtain a first mean; the mean of the monitoring data after the zero point drift occurs is calculated to obtain a second mean; and a data correction value is calculated based on the first mean and the second mean;

[0030] The data correction value is used to correct the monitoring data in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge, so as to obtain the corrected spatiotemporal distribution matrix of the real-time monitoring data of the bridge, that is, to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after preprocessing; wherein, the correction process is specifically as follows: the monitoring data with zero point drift is subtracted from the data correction value to obtain the correction value, and the correction value is used to replace the monitoring data with zero point drift.

[0031] Furthermore, the process of calculating the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state by using the historical monitoring data of the bridge to be tested in the healthy state is as follows:

[0032] Obtain historical monitoring data of the bridge to be tested in a healthy state;

[0033] Calculate the mean μ and variance σ of the historical monitoring data of the bridge to be tested in a healthy state;

[0034] According to the mean μ and variance σ of the historical monitoring data of the bridge to be tested in the healthy state, a safety threshold value of the mean of the monitoring data of the bridge to be tested in the healthy state is obtained;

[0035] The safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state is (μ-4σ, μ+4σ).

[0036] Furthermore, after comparing the mean value of the monitoring data in each time sliding window with the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state, and obtaining the abnormal state detection result of the bridge to be tested, the method further includes an abnormal state locating step;

[0037] The abnormal state location step is as follows:

[0038] According to the normal distribution principle, a data classifier is constructed to describe the operation status of the bridge.

[0039] According to the data classifier for describing the operation status of the bridge, the spatiotemporal distribution matrix of the pre-processed real-time monitoring data of the bridge is subjected to classification processing to obtain a classification result;

[0040] According to the classification results, construct a real-time monitoring data status matrix of the bridge;

[0041] According to a preset coloring standard, the real-time monitoring data state matrix of the bridge is colorized to obtain a dynamic standard color atlas;

[0042] Performing grayscale processing on the dynamic standard color atlas to obtain a grayscale atlas; generating a histogram under the grayscale atlas according to the grayscale atlas;

[0043] Binarizing the histogram under the grayscale spectrum to obtain a damage spectrum;

[0044] According to the damage map, the abnormal state positioning result of the bridge to be tested is obtained.

[0045] Furthermore, the process of coloring the bridge real-time monitoring data state matrix according to the preset coloring standard to obtain a dynamic standard color atlas is as follows:

[0046] Determine the numerical boundary of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge; divide the numerical boundary of the monitoring data at each measuring point into intervals according to the total number of colors in the preset color phase space in the preset coloring standard, and obtain the real-time monitoring data state matrix of the bridge after the interval division;

[0047] According to the correspondence between the value of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge and the color of the preset color phase space, the real-time monitoring data state matrix of the bridge after the interval division is mapped to the preset color phase space to obtain a dynamic standard color map; wherein, the correspondence between the value of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge and the color of the preset color phase space is:

[0048]

[0049] Among them, x:p ij is the pixel value of the i-th row and j-th column in the two-dimensional map; x ij is the monitoring data of the i-th row and j-th column in the real-time monitoring data state matrix of the bridge; N is the total number of colors in the preset color phase space; x max is the maximum value in the bridge real-time monitoring data state matrix; x min It is the minimum value in the real-time monitoring data status matrix of the bridge.

[0050] The present invention also provides a bridge abnormal state detection system, comprising:

[0051] A time-space matrix module is used to obtain the real-time monitoring data of the bridge to be tested and construct a time-space distribution matrix of the bridge real-time monitoring data; wherein, in the time-space distribution matrix of the bridge real-time monitoring data, each column represents the spatial distribution sequence of all monitoring data at a certain moment, and each row represents the time distribution sequence of monitoring data at a certain measuring point on the bridge to be tested;

[0052] A data preprocessing module, used to perform data preprocessing operations on the spatiotemporal distribution matrix of the real-time monitoring data of the bridge to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after preprocessing;

[0053] A threshold module, used to calculate a safety threshold of the mean value of the monitoring data of the bridge to be tested in a healthy state by using the historical monitoring data of the bridge to be tested in a healthy state;

[0054] A data statistics module is used to traverse each row of monitoring data in the spatiotemporal distribution matrix of the preprocessed real-time monitoring data of the bridge using a time sliding window, and perform statistical processing on the monitoring data in each time sliding window to obtain the mean value of the monitoring data in each time sliding window;

[0055] The abnormality detection module is used to compare the mean value of the monitoring data in each time sliding window with the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state, so as to obtain the abnormal state detection result of the bridge to be tested.

[0056] The present invention also provides a bridge abnormal state detection device, comprising:

[0057] Memory for storing computer programs;

[0058] A processor is used to implement the steps of the bridge abnormal state detection method when executing the computer program.

[0059] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the bridge abnormal state detection method are implemented.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] The present invention provides a bridge abnormal state detection method and system, which uses a time sliding window to perform statistical processing on each row of monitoring data in the spatiotemporal distribution matrix of the preprocessed real-time monitoring data of the bridge, compares the mean of the monitoring data in each time sliding window with the safety threshold of the mean of the monitoring data of the bridge to be tested in the healthy state, and judges whether damage has occurred based on the comparison result; through statistical analysis of the real-time monitoring data of the bridge, rapid detection of abnormal states of the bridge is achieved, the damage identification efficiency is high, the analysis results are interpretable, and the safety and operational level of the bridge are ensured.

[0062] Furthermore, during the data preprocessing process, the monitoring data with zero-point drift in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge is corrected, which effectively improves the accuracy of the data in the spatiotemporal distribution matrix of the preprocessed real-time monitoring data of the bridge and ensures the accuracy of the detection results.

[0063] Furthermore, the spatiotemporal distribution matrix of the preprocessed real-time monitoring data of the bridge is hierarchically processed to reconstruct a graph structure with rich information meaning, intuitive images and strong interpretability. The image is analyzed using digital image processing technology, which greatly improves the efficiency and accuracy of bridge structure damage identification, increases the readability of bridge structure data results, and effectively improves the efficiency and accuracy of bridge abnormal state detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is a flow chart of the bridge abnormal state detection method according to the present invention;

[0065] Figure 2 This is a graph showing normalization and mean statistical results of the monitoring data of the second deflection sensor of the bridge 1 in Example 1;

[0066] Figure 3 A dynamic standard color atlas of the monitoring data of the second deflection sensor of the bridge 1 in Example 1;

[0067] Figure 4 is a dynamic standard color atlas of the bridge 1 in Example 1;

[0068] Figure 5 is the grayscale map of bridge 1 in Example 1;

[0069] Figure 6 is the damage map of the bridge 1 in Example 1;

[0070] Figure 7This is a graph showing normalization and mean statistical results of the monitoring data of the fourth deflection sensor of the bridge 2 in Example 2;

[0071] Figure 8 A dynamic standard color atlas of the monitoring data of the fourth deflection sensor of the bridge 2 in Example 2;

[0072] Fig. 9 is the dynamic standard color atlas of bridge 2 in Example 2;

[0073] Fig.10 is the grayscale map of bridge 2 in Example 2;

[0074] Fig.11 This is the damage map of Bridge 2 in Example 2. DETAILED DESCRIPTION

[0075] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention more clearly understood, the present invention is further described in detail in the following specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0076] As attached Figure 1 As shown, the present invention provides a method for detecting abnormal state of a bridge, comprising the following steps:

[0077] Step 1. According to the main defects of small and medium-span bridges, as well as the monitoring cost and operability, deflection sensors are deployed at preset positions of the bridges in accordance with TCCES 15-2020 "Technical Specifications for Selection and Deployment of Bridge Health Monitoring Sensors", and the deflection of the bridge is monitored in real time. The real-time monitoring data of the bridge to be tested is obtained, and a spatiotemporal distribution matrix of the real-time monitoring data of the bridge is constructed; wherein, in the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, each column represents the spatial distribution sequence of all monitoring data at a certain moment, and each row represents the time distribution sequence of the monitoring data at a certain measuring point on the bridge to be tested.

[0078] The real-time monitoring data of the bridge to be tested is the deflection data collected in real time by the bridge health detection system; specifically: presetting deflection sensors at different positions of the bridge to be tested, sampling once every sampling interval τ; and sampling once every sampling interval τ to obtain a group of real-time monitoring data of the deflection sensors, to form an m×n-order spatiotemporal distribution matrix X of the real-time monitoring data of the bridge, so as to reflect the health status of the bridge to be tested during the detection period.

[0079] Among them, the spatiotemporal distribution matrix X of the real-time monitoring data of the bridge is:

[0080]

[0081] Among them, xi,j is the real-time monitoring data of the i-th deflection sensor at the j-th moment; m is the total number of preset deflection sensors at different positions of the bridge to be tested, that is, the number of measuring points of the bridge to be tested; n is the total number of samples of each deflection sensor.

[0082] Step 2: Perform data preprocessing on the spatiotemporal distribution matrix of the real-time monitoring data of the bridge to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after preprocessing; the specific process is as follows:

[0083] In the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, each row of monitoring data is traversed to determine whether each row of monitoring data is all zero; if one row of monitoring data is all zero, the monitoring data of the corresponding row is removed from the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after division by zero;

[0084] Using the most probable value filling method, data filling is performed on the time-space distribution matrix of the bridge real-time monitoring data after the division by zero, so as to obtain the time-space distribution matrix of the bridge real-time monitoring data after the data filling;

[0085] Using a wavelet denoising method, the spatiotemporal distribution matrix of the bridge real-time monitoring data after the data filling is subjected to denoising, so as to obtain the spatiotemporal distribution matrix of the bridge real-time monitoring data after the denoising;

[0086] The time-space distribution matrix of the real-time monitoring data of the bridge after the noise reduction processing is normalized to map the value of each row in the time-space distribution matrix to the range of [0, 1] to obtain the normalized time-space distribution matrix of the real-time monitoring data of the bridge.

[0087] Step 3: correcting the monitoring data with zero-point drift in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge, to obtain a corrected spatiotemporal distribution matrix of the real-time monitoring data of the bridge; wherein the specific steps are as follows:

[0088] Obtain historical monitoring data of the bridge to be tested in a zero-point drift state and historical monitoring data of the bridge to be tested in a normal state, respectively calculate the slope and mean of the historical monitoring data of the bridge to be tested in the zero-point drift state, and calculate the slope and mean of the historical monitoring data of the bridge to be tested in a normal state;

[0089] According to the slope and mean of the historical monitoring data of the bridge to be tested in the zero drift state, a slope threshold and a variance threshold for determining whether the monitoring data of the bridge to be tested has zero drift are obtained;

[0090] Using a time sliding window, traverse each row of monitoring data in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge, and calculate the slope and variance of the monitoring data in each time sliding window;

[0091] Comparing the slope of the monitoring data in each time sliding window with a slope threshold of whether the monitoring data of the bridge to be tested has a zero drift, or comparing the variance of the monitoring data in each time sliding window with a variance threshold of whether the monitoring data of the bridge to be tested has a zero drift;

[0092] If the slope of the monitoring data in each time sliding window exceeds the slope threshold of whether the monitoring data of the bridge to be tested has zero drift, or the variance of the real-time monitoring data in each time sliding window exceeds the variance threshold of whether the monitoring data of the bridge to be tested has zero drift, then the corresponding monitoring data has zero drift;

[0093] For the monitoring data with zero point drift, the mean of the monitoring data before the zero point drift occurs is calculated to obtain a first mean; the mean of the monitoring data after the zero point drift occurs is calculated to obtain a second mean; and a data correction value is calculated based on the first mean and the second mean;

[0094] The data correction value is used to correct the monitoring data in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge, so as to obtain the corrected spatiotemporal distribution matrix of the real-time monitoring data of the bridge, that is, to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after preprocessing; wherein, the correction process is specifically as follows: the monitoring data with zero point drift is subtracted from the data correction value to obtain the correction value, and the correction value is used to replace the monitoring data with zero point drift.

[0095] Step 4: Calculate the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state by using the historical monitoring data of the bridge to be tested in the healthy state; wherein the calculation process of the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state is as follows:

[0096] Obtain historical monitoring data of the bridge to be tested in a healthy state;

[0097] Calculate the mean μ and variance σ of the historical monitoring data of the bridge to be tested in a healthy state;

[0098] According to the mean μ and variance σ of the historical monitoring data of the bridge to be tested in the healthy state, a safety threshold value of the mean of the monitoring data of the bridge to be tested in the healthy state is obtained;

[0099] The safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state is (μ-4σ, μ+4σ).

[0100] Step 5: Use the time sliding window to traverse each row of monitoring data in the corrected spatiotemporal distribution real-time monitoring data matrix, and perform statistical processing on the monitoring data in each time sliding window to obtain the mean of the real-time monitoring data in each time sliding window.

[0101] Step 6. Compare the mean of the monitoring data in each time sliding window with the safety threshold of the mean of the monitoring data of the bridge to be tested in the healthy state to obtain the abnormal state detection result of the bridge to be tested; specifically, if the mean of the monitoring data in each time sliding window does not exceed the safety threshold of the mean of the monitoring data of the bridge to be tested in the healthy state, it means that the real-time monitoring data of the corresponding bridge is normal; if the mean of the monitoring data in each time sliding window exceeds the safety threshold of the mean of the monitoring data of the bridge to be tested in the healthy state, it means that the real-time monitoring data of the corresponding bridge is abnormal.

[0102] Step 7: Based on the normal distribution, construct a data classifier for describing the operating status of the bridge; according to the data classifier for describing the operating status of the bridge, perform classification processing on the spatiotemporal distribution matrix of the preprocessed real-time monitoring data of the bridge to obtain a classification result; and based on the classification result, construct a real-time monitoring data status matrix of the bridge.

[0103] The process of constructing the real-time monitoring data status matrix of the bridge is as follows:

[0104] Acquire historical monitoring data of the bridge to be tested in a healthy state, and calculate the mean μ and variance σ of the historical monitoring data of the bridge to be tested in the healthy state;

[0105] In the present invention, 9 different levels of status areas are defined according to the normal distribution, so as to obtain the data classifier for describing the operating status of the bridge; wherein the 9 different levels of status areas include the first level area (-∞, μ-4σ), the second level area (μ-4σ, μ-3σ), the third level area (μ-3σ, μ-2σ), the fourth level area (μ-2σ, μ-σ), the fifth level area (μ-σ, μ+σ), the sixth level area (μ+σ, μ+2σ), the seventh level area (μ+2σ, μ+3σ), the eighth level area (μ+3σ, μ+4σ) and the ninth level area (μ+4σ, +∞).

[0106] According to the data classifier, the time-space distribution matrix of the modified real-time monitoring data of the bridge is graded and divided, and the monitoring data divided into the same level of status area are replaced with the same number to obtain the real-time monitoring data status matrix of the bridge.

[0107] Step 8: Color the bridge real-time monitoring data status matrix according to a preset coloring standard to obtain a dynamic standard color atlas.

[0108] The process of coloring the bridge real-time monitoring data state matrix to obtain a dynamic standard color map specifically includes the following steps:

[0109] Determine the numerical boundary of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge; divide the numerical boundary of the monitoring data at each measuring point into intervals according to the total number of colors in the preset color phase space in the preset coloring standard, and obtain the real-time monitoring data state matrix of the bridge after the interval division;

[0110] According to the correspondence between the value of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge and the color of the preset color phase space, the real-time monitoring data state matrix of the bridge after the interval division is mapped to the preset color phase space to obtain a dynamic standard color map; wherein, the correspondence between the value of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge and the color of the preset color phase space is:

[0111]

[0112] Among them, x:p ij is the pixel value of the i-th row and j-th column in the dynamic standard color atlas; x ij is the monitoring data of the i-th row and j-th column in the real-time monitoring data state matrix of the bridge; N is the total number of colors in the preset color phase space; x max is the maximum value in the bridge real-time monitoring data state matrix; x min It is the minimum value in the real-time monitoring data status matrix of the bridge.

[0113] Step 9: grayscale processing is performed on the dynamic standard color map to obtain a grayscale map; and a histogram under the grayscale map is generated according to the grayscale map.

[0114] Step 10: Binarize the histogram under the grayscale spectrum to obtain a damage spectrum.

[0115] Step 11: According to the position and time when the black area appears in the damage map, the positioning result of the abnormal state of the bridge can be obtained.

[0116] The bridge abnormal state detection method described in the present invention performs statistical processing on the real-time monitoring data of the bridge after data correction and data preprocessing through a time sliding window, and uses its mean value to judge whether the real-time monitoring data of the bridge has been damaged; the present invention can quickly determine whether the bridge has been damaged through data analysis, and then determine the location and time of the bridge damage based on the dynamic standard color map; wherein, damage identification first traverses the monitoring data through the time sliding window, and judges whether the damage has occurred through the mean value in the time sliding window; when no damage has occurred, the next monitoring data is identified; damage location can judge the location and time of the damage through the map formed by the real-time monitoring data of the bridge, and when a continuous and large area of ​​abnormal color appears in the map, it is judged as a damaged area, and the location can be accurately near the corresponding actual location of the sensor in the bridge, and in order to make the damage more obvious, image processing means are used to generate a damage map.

[0117] In the present invention, the bridge health monitoring data is reconstructed into a dynamic standard color atlas with rich information, intuitive images, and strong interpretability. The powerful image analysis ability of digital image processing technology is utilized to greatly improve the efficiency and accuracy of bridge structure damage identification. It has important scientific significance and engineering value for quickly and accurately identifying abnormal data and damage degree of bridge structure and improving bridge operation safety and operation level.

[0118] The present invention also provides a bridge abnormal state detection system, including a space-time matrix module, a data preprocessing module, a data statistics module and an abnormality detection module; the space-time matrix module is used to obtain the real-time monitoring data of the bridge to be tested and construct a space-time distribution matrix of the bridge real-time monitoring data; wherein, in the space-time distribution matrix of the bridge real-time monitoring data, each column represents the spatial distribution sequence of all monitoring data at a certain moment, and each row represents the time distribution sequence of monitoring data at a certain measuring point on the bridge to be tested; the data preprocessing module is used to perform data preprocessing operations on the space-time distribution matrix of the bridge real-time monitoring data to obtain the space-time distribution matrix of the bridge real-time monitoring data after preprocessing. A threshold module is used to calculate the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state by using the historical monitoring data of the bridge to be tested in the healthy state; a data statistics module is used to traverse each row of monitoring data in the spatiotemporal distribution matrix of the preprocessed real-time monitoring data of the bridge by using a time sliding window, and perform statistical processing on the monitoring data in each time sliding window to obtain the mean value of the monitoring data in each time sliding window; an anomaly detection module is used to compare the mean value of the monitoring data in each time sliding window with the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state to obtain the abnormal state detection result of the bridge to be tested.

[0119] The present invention also provides a bridge abnormal state detection device, comprising: a memory for storing a computer program; a processor for implementing the steps of the bridge abnormal state detection method when executing the computer program. Alternatively, the processor implements the functions of each module in the above bridge abnormal state detection system when executing the computer program.

[0120] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of completing preset functions, and the instruction segments are used to describe the execution process of the computer program in the bridge abnormal state detection device.

[0121] The bridge abnormal state detection device may be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The bridge abnormal state detection device may include, but is not limited to, a processor and a memory. Those skilled in the art may understand that the above are examples of bridge abnormal state detection devices and do not constitute a limitation on the bridge abnormal state detection device. The bridge abnormal state detection device may include more components than the above, or a combination of certain components, or different components. For example, the bridge abnormal state detection device may also include input and output devices, network access devices, buses, etc.

[0122] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc. The processor is the control center of the bridge abnormal state detection device, and uses various interfaces and lines to connect various parts of the entire bridge abnormal state detection device.

[0123] The memory can be used to store the computer program and / or module, and the processor implements various functions of the bridge abnormal state detection device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory.

[0124] The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0125] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the bridge abnormal state detection method are implemented.

[0126] If the modules / units integrated in the bridge abnormal state detection system are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0127] Based on such understanding, the present invention implements all or part of the processes in the above-mentioned bridge abnormal state detection method, and can also be completed by instructing related hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above-mentioned bridge abnormal state detection method can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or preset intermediate form, etc.

[0128] The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0129] It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable storage media do not include electrical carrier signals and telecommunication signals.

[0130] Example 1

[0131] Taking the abnormal state detection process of a bridge 1 as an example, data from the actual project of the bridge 1 is collected at a frequency of once per hour. The deflection data of the bridge is collected, and a total of 9 months of historical data is collected as the research object; wherein, there are three deflection sensors on the bridge 1, and the failure data of the first deflection sensor is all zero; each deflection sensor has a total of 6562 sampling points, totaling 13124 data.

[0132] As attached Figure 1 As shown, this embodiment 1 provides a bridge abnormal state detection method, which specifically includes the following steps:

[0133] Step 1. According to the main diseases of small and medium-span bridges, monitoring costs and operability, deflection sensors are deployed at preset positions of the bridge in accordance with TCCES 15-2020 "Technical Specifications for Selection and Deployment of Bridge Health Monitoring Sensors" to monitor the deflection of the bridge in real time.

[0134] Step 2: Acquire the real-time monitoring data of the bridge 1 to be tested, and construct a spatiotemporal distribution matrix of the real-time monitoring data of the bridge; wherein, in the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, each column represents the spatial distribution sequence of all monitoring data at a certain moment, and each row represents the time distribution sequence of monitoring data at a certain measuring point on the bridge to be tested.

[0135] The spatiotemporal distribution matrix X of the real-time monitoring data of the bridge is an m×n-order data matrix; the spatiotemporal distribution matrix X of the real-time monitoring data of the bridge is, specifically:

[0136]

[0137] Among them, x 1,j is the monitoring data of the first sensor at the jth moment; x 2,j is the monitoring data of the second sensor at the jth moment.

[0138] Step 3, traverse each row of monitoring data in the spatiotemporal distribution matrix of the real-time monitoring data of the bridge to determine whether the monitoring data in this row are all zero. If they are all zero, it means that the corresponding deflection sensor has failed, and the monitoring data in this row is eliminated to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after dividing by zero.

[0139] Step 4: Using the best possible value filling method, through regression analysis and Bayesian statistical inference method, solve the maximum possible value of the missing value of the specific attribute in the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after division by zero, and fill it in to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after data filling.

[0140] Step 5: Use the wavelet denoising method to perform denoising on the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after the data filling, and obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after the denoising. Specifically, perform four-layer wavelet decomposition through the db30 wavelet, use the hard threshold method to denoise, and perform wavelet reconstruction to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after the denoising.

[0141] In this embodiment 1, the wavelet denoising method includes signal decomposition, threshold processing and signal reconstruction, which are as follows:

[0142] Signal decomposition: Select appropriate wavelet basis functions and decomposition functions to perform wavelet decomposition on the signal at different scales; wherein the wavelet basis functions and decomposition functions are as follows:

[0143]

[0144] h 0 (k)=<φ 1,0 ,φ 0,k >

[0145]

[0146] Among them, c j,k is the kth actual scale coefficient under the jth scale; h 0 (k),h 1 (k) is a pair of orthogonal mirror filter banks; n is the nth sampling point; c j-1,k is the kth actual scale coefficient at the j-1th scale; ω j,k is the kth actual wavelet coefficient at the jth scale; ω j-1,k is the kth actual wavelet coefficient at the j-1th scale; M is the number of discrete sampling points; φ 1,0 and φ 0,k All are scale functions; and They are all wavelet functions.

[0147] Threshold processing: After wavelet analysis, a suitable threshold is rotated; wherein the threshold is used to perform threshold processing on each layer of high frequency coefficients.

[0148] Signal reconstruction: reconstruct the pre-processed wavelet coefficients; wherein the reconstruction formula is:

[0149]

[0150] Among them, h 0 (n-2k) and h 1 (n-2k) is a pair of orthogonal mirror filter banks.

[0151] Step 6: Perform linear transformation on all elements in the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after the noise reduction processing, so that the linear transformation result of each element falls into the interval of [0,1], and obtain the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge.

[0152] Step 7: traverse each row of monitoring data in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge to determine whether the sensor has zero drift; wherein the specific steps are:

[0153] Obtain historical monitoring data of the bridge to be tested in a zero-point drift state and historical monitoring data of the bridge to be tested in a normal state, respectively calculate the slope and mean of the historical monitoring data of the bridge to be tested in the zero-point drift state, and calculate the slope and mean of the historical monitoring data of the bridge to be tested in a normal state;

[0154] According to the slope and mean of the historical monitoring data of the bridge to be tested in the zero drift state, a slope threshold and a variance threshold for determining whether the monitoring data of the bridge to be tested has zero drift are obtained;

[0155] Using a time sliding window, traverse each row of monitoring data in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge, and calculate the slope and variance of the monitoring data in each time sliding window;

[0156] Comparing the slope of the monitoring data in each time sliding window with a slope threshold of whether the monitoring data of the bridge to be tested has a zero drift, or comparing the variance of the monitoring data in each time sliding window with a variance threshold of whether the monitoring data of the bridge to be tested has a zero drift;

[0157] If the slope of the monitoring data within each time sliding window exceeds the slope threshold of whether the monitoring data of the bridge to be tested has zero drift, or the variance of the real-time monitoring data within each time sliding window exceeds the variance threshold of whether the monitoring data of the bridge to be tested has zero drift, then the corresponding monitoring data has zero drift.

[0158] For the monitoring data with zero point drift, the mean of the monitoring data before the zero point drift occurs is calculated to obtain a first mean; the mean of the monitoring data after the zero point drift occurs is calculated to obtain a second mean; and a data correction value is calculated based on the first mean and the second mean;

[0159] The data correction value is used to correct the monitoring data in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge, so as to obtain the corrected spatiotemporal distribution matrix of the real-time monitoring data of the bridge; wherein the correction process is specifically as follows: the monitoring data with zero point drift is subtracted from the data correction value to obtain the correction value, and the monitoring data with zero point drift is replaced by the correction value.

[0160] Step 8: Calculate the safety threshold of the mean value of the monitoring data of the bridge to be tested in a healthy state by using the historical monitoring data of the bridge to be tested in a healthy state.

[0161] The construction process of the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state is as follows:

[0162] Obtain historical monitoring data of the bridge to be tested in a healthy state;

[0163] Calculate the mean μ and variance σ of the historical monitoring data of the bridge to be tested in a healthy state;

[0164] According to the mean μ and variance σ of the historical monitoring data of the bridge to be tested in the healthy state, a safety threshold value of the mean of the monitoring data of the bridge to be tested in the healthy state is obtained;

[0165] The safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state is (μ-4σ, μ+4σ).

[0166] Among them, the mean μ and variance σ are respectively:

[0167]

[0168]

[0169] Step 9: Use the time sliding window to traverse each row of monitoring data in the time-space distribution matrix of the corrected real-time bridge monitoring data, and perform statistical processing on the monitoring data in each time sliding window to obtain the mean of the monitoring data in each time sliding window.

[0170] Among them, the design principle of the time sliding window is: initially a time sliding window is fixed, and only the data in the window is processed each time; with the help of the time sliding window, the mean, slope and variance of the data in each time sliding window are calculated, and then the time sliding window slides backward by a step distance, and the data in the current window is the new data to be processed, and continues to slide after statistical processing; and so on, until the data processing in the last time sliding window is completed.

[0171] Step 10: Compare the mean value of the monitoring data in each time sliding window with the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state to obtain the abnormal state detection result of the bridge to be tested.

[0172] Specifically, if the mean value of the monitoring data within each time sliding window does not exceed the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state, it means that the real-time monitoring data of the corresponding bridge is normal; if the mean value of the monitoring data within each time sliding window exceeds the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state, it means that there is an abnormality in the real-time monitoring data of the corresponding bridge.

[0173] Step 11: Based on the normal distribution, a data classifier is constructed to describe the operation status of the bridge; according to the data classifier, the normalized spatiotemporal distribution matrix is ​​classified to obtain the classification results, and a real-time monitoring data status matrix of the bridge is constructed.

[0174] Among them, the process of building a real-time monitoring data status matrix of the bridge; the specific process is as follows:

[0175] Acquire historical monitoring data of the bridge to be tested in a healthy state, and calculate the mean μ and variance σ of the historical monitoring data of the bridge to be tested in the healthy state;

[0176] According to the normal distribution, 9 different levels of state areas are defined, that is, the data classifier for describing the operation state of the bridge is obtained; wherein the 9 different levels of state areas include the first level area (-∞, μ-4σ), the second level area (μ-4σ, μ-3σ), the third level area (μ-3σ, μ-2σ), the fourth level area (μ-2σ, μ-σ), the fifth level area (μ-σ, μ+σ), the sixth level area (μ+σ, μ+2σ), the seventh level area (μ+2σ, μ+3σ), the eighth level area (μ+3σ, μ+4σ) and the ninth level area (μ+4σ, +∞);

[0177] According to the data classifier, the time-space distribution matrix of the modified real-time monitoring data of the bridge is graded and divided, and the monitoring data divided into the same level of status area are replaced with the same number to obtain the real-time monitoring data status matrix of the bridge.

[0178] If the data in the preprocessed spatiotemporal distribution matrix is ​​located in the second-level area to the eighth-level area, the corresponding data is in a safe state; if the data in the preprocessed spatiotemporal distribution matrix is ​​located in the first-level area or the ninth-level area, the corresponding data is abnormal; wherein, when the data in the same level status area are replaced with the same number, the data in the first-level area to the ninth-level area are replaced with numbers 1-9 in sequence;

[0179] The bridge real-time monitoring data state matrix X′ is:

[0180]

[0181] Step 12: color the bridge real-time monitoring data status matrix according to a preset coloring standard, and obtain a dynamic standard color map after computer image processing; the dynamic standard color map can be a dynamic standard color map that characterizes the system abnormality location and the abnormality occurrence time, so as to realize the identification of bridge structure abnormalities.

[0182] The specific process is as follows:

[0183] Determine the numerical boundary of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge [x min ,x max ]; According to the total number of colors N in the preset color phase space, the numerical boundary of the monitoring data at each measuring point is divided into intervals to obtain the real-time monitoring data state matrix of the bridge after interval division; the color phase space follows a gradient from blue to yellow.

[0184] Among them, the variable interval length is:

[0185]

[0186] According to the correspondence between the value of the monitoring data at each measuring point in the real-time monitoring data status matrix of the bridge and the color of the preset color phase space, the real-time monitoring data status matrix of the bridge after the interval division is mapped to the preset color phase space to obtain a dynamic standard color atlas.

[0187] The corresponding relationship between the value of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge and the color of the preset color phase space is:

[0188]

[0189] Among them, x:p ij is the pixel value of the i-th row and j-th column in the dynamic standard color atlas; x ij is the monitoring data of the i-th row and j-th column in the real-time monitoring data state matrix of the bridge; N is the total number of colors in the preset color phase space; x max is the maximum value in the bridge real-time monitoring data state matrix; x min It is the minimum value in the real-time monitoring data status matrix of the bridge.

[0190] Step 13: grayscale processing is performed on the dynamic standard color map to obtain a grayscale map; and a histogram under the grayscale map is generated according to the grayscale map.

[0191] Step 14: preset a custom threshold and perform binarization on the histogram under the grayscale map, that is, convert the abnormal area in the histogram under the grayscale map into a black area, and convert the remaining parts into white, to obtain a damage map; wherein the abnormal area is the black area in the damage map.

[0192] Step 15: Obtain the positioning result of the abnormal state of the bridge according to the damage map; specifically, the positioning result of the abnormal state of the bridge can be obtained according to the position and time when the black area in the damage map appears.

[0193] As attached Figure 2-3 As shown, attached Figure 2 The normalization and mean statistical result diagram of the monitoring data of the second deflection sensor of the bridge 1 in Example 1 are given in FIG. Figure 3 The dynamic standard color atlas of the monitoring data of the second deflection sensor of the bridge 1 in Example 1 is given in the figure; Figure 2-3 It can be seen that when the monitoring data is abnormal, the mean value will exceed the threshold range and the color of the map at the corresponding position will also change abnormally.

[0194] As attached Figure 4 As shown, attached Figure 4 The dynamic standard color atlas of the bridge 1 in Example 1 is given in the figure. Figure 4 It can be seen that the second and third sensors of Bridge 1 both turned blue at about 3500 hours and remained blue in the later period. Therefore, Bridge 1 may have structural damage after 3500 hours, and the damage location is the corresponding position of the second and third sensors.

[0195] As attached Figure 5-6 As shown, attached Figure 5 The grayscale map of bridge 1 in Example 1 is given. Figure 6 The damage map of the bridge 1 in Example 1 is given in FIG. Figure 5-6 It can be seen that the abnormal color area in the dynamic standard color atlas is the damaged area, and the damage atlas can show the damaged area more clearly.

[0196] Example 2

[0197] Taking the abnormal state detection process of a bridge 2 as an example, the real-time monitoring data of the actual project of the bridge 2 is collected at a frequency of once per hour, and the deflection data of the bridge is collected. A total of 8 months of historical data is collected as the research object; among them, there are 28 deflection sensors on the bridge 2, and if the failure data of 7 deflection sensors are all zero, then this row of data is directly eliminated; each deflection sensor has a total of 5840 sampling points, totaling 122640 data.

[0198] This embodiment 2 provides a bridge abnormal state detection method, which specifically includes the following steps:

[0199] Step 1: collect the deflection data of the bridge health detection system in real time, and construct an m×n-order spatiotemporal distribution matrix, that is, construct the spatiotemporal distribution matrix of the bridge real-time monitoring data; wherein, in the spatiotemporal distribution matrix of the bridge real-time monitoring data, the columns represent the spatial series distribution of all sensor monitoring data at the current moment, and the rows represent the time series distribution of the sensor monitoring data.

[0200] Step 2: traverse each row of data in the preprocessed space-time matrix to determine whether the row of data is all zero. If all is zero, it means that the sensor fails and the row of data is removed.

[0201] Step 3: Using the most likely value filling method, through regression analysis and Bayesian statistical inference method, solve the maximum possible value of the missing value of the specific attribute in the spatiotemporal distribution matrix, and fill it in to obtain the spatiotemporal distribution matrix after data filling.

[0202] Step 4: Use db30 wavelet to perform four-layer wavelet decomposition, use hard threshold method to denoise, and perform wavelet reconstruction to obtain the space-time distribution matrix after denoising.

[0203] Step 5: Perform a linear transformation on all elements in the space-time distribution matrix after the denoising process so that the linear transformation result of each element falls within the interval [0, 1], thereby obtaining the space-time distribution matrix after the preprocessing.

[0204] Step 6: traverse each row of data in the preprocessed space-time matrix to determine whether the sensor has zero drift and correct the data.

[0205] Step 7: Collect historical monitoring data of the bridge health status and calculate the safety threshold of the mean value under the bridge health status.

[0206] Step 8: Use the time sliding window to traverse each row of data in the preprocessed spatiotemporal distribution matrix and calculate the mean of the data in each time sliding window.

[0207] Step 9: Compare the mean of the bridge real-time monitoring data with the mean safety threshold. If it does not exceed the mean safety threshold, it means that the corresponding bridge real-time monitoring data is normal. If it exceeds the mean safety threshold, it means that the corresponding bridge real-time monitoring data is abnormal.

[0208] Step 10: Based on the normal distribution, a data classifier is constructed to describe the operation status of the bridge; according to the data classifier, the preprocessed spatiotemporal distribution matrix is ​​classified to obtain a classification result, and a real-time monitoring data status matrix of the bridge is constructed.

[0209] Step 11: color the bridge real-time monitoring data status matrix according to a preset coloring standard, and obtain a dynamic standard color map after computer image processing; the dynamic standard color map can be a dynamic standard color map that characterizes the system fault location and fault degree to realize the identification of abnormal data.

[0210] Step 12: convert the dynamic standard color map into a grayscale map, and generate a histogram under the grayscale map.

[0211] Step 13, preset a custom threshold, and binarize the histogram under the grayscale spectrum, that is, convert the abnormal area in the histogram under the grayscale spectrum into a black area, and convert the remaining parts into white, to obtain a damage spectrum; wherein the abnormal area is the black area in the damage spectrum.

[0212] Step 14: obtaining the positioning result of the abnormal state of the bridge according to the damage map; specifically, obtaining the positioning result of the abnormal state of the bridge according to the position and time when the black area in the damage map appears.

[0213] As attached Figure 7-8 As shown, attached Figure 7 The normalization and mean statistical result diagram of the monitoring data of the fourth deflection sensor of Bridge 2 in Example 2 are given in FIG. Figure 8 The dynamic standard color atlas of the monitoring data of the fourth deflection sensor of the bridge 2 in Example 2 is given in the figure; Figure 7-8 It can be seen that when the data is abnormal, the mean value will exceed the threshold range, and the color of the map at the corresponding position will also change abnormally.

[0214] As attached Fig. 9 As shown, attached Fig. 9 The dynamic standard color atlas of the bridge 2 in Example 2 is given in FIG. Fig. 9 It can be seen that the fourth and fifth sensors have been showing yellow since about 3000h, which means the damaged position is the position on the bridge corresponding to the fourth and fifth sensors, and on-site investigation is needed; the 14th sensor turned yellow at about 4000h, which means the damaged position is the position on the bridge corresponding to the 14th sensor, and on-site investigation is needed.

[0215] As attached Figure 10-11 As shown, attached Fig.10 The grayscale map of bridge 2 in Example 2 is given in FIG. Fig.11 The damage map of bridge 2 in Example 2 is given in FIG. Figure 10-11It can be seen that the abnormal color area in the dynamic standard color atlas is the damaged area, and the damage atlas can show the damaged area more clearly.

[0216] The description of the relevant parts of a bridge abnormal state detection system, device and computer-readable storage medium provided in Examples 1 and 2 can be found in the detailed description of the corresponding parts of a bridge abnormal state detection method described in Example 1 or 2, and will not be repeated here.

[0217] The bridge abnormal state detection method and system described in the present invention can quickly analyze data to determine whether damage has occurred, and then determine the approximate location and time of the damage based on the map; damage identification first uses a sliding window to traverse the monitoring data and judge whether damage has occurred by the mean value in the sliding window. When no damage has occurred, the next monitoring data is identified; damage location can use the map formed by the overall data to determine the location and time of the damage. When continuous and large-area abnormal colors appear in the map, it is judged as a damaged area. The location can be accurate to the vicinity of the actual location corresponding to the sensor in the bridge. In order to make the damage more obvious, image processing methods are used to generate a damage map.

[0218] The above embodiment is only one of the implementation methods that can realize the technical solution of the present invention. The scope of protection claimed by the present invention is not limited only to this embodiment, but also includes changes, replacements and other implementation methods that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed by the present invention.

Claims

1. A bridge abnormal state detection method, characterized in that: The following steps are involved: Acquire real-time monitoring data of the bridge to be tested, and construct a spatiotemporal distribution matrix of the real-time monitoring data of the bridge; wherein, in the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, each column represents a spatial distribution sequence of all monitoring data at a certain moment, and each row represents a time distribution sequence of monitoring data at a certain measuring point on the bridge to be tested; Performing data preprocessing on the spatiotemporal distribution matrix of the real-time monitoring data of the bridge to obtain a preprocessed spatiotemporal distribution matrix of the real-time monitoring data of the bridge; Using the historical monitoring data of the bridge to be tested in a healthy state, a safety threshold of the mean value of the monitoring data of the bridge to be tested in a healthy state is calculated; Using a time sliding window, traverse each row of monitoring data in the spatiotemporal distribution matrix of the preprocessed real-time monitoring data of the bridge, and perform statistical processing on the monitoring data in each time sliding window to obtain the mean value of the monitoring data in each time sliding window; The mean value of the monitoring data in each time sliding window is compared with the safety threshold value of the mean value of the monitoring data of the bridge to be tested in the healthy state to obtain the abnormal state detection result of the bridge to be tested.

2. A bridge abnormal state detection method according to claim 1, characterized in that: The process of obtaining the real-time sensor data of the bridge to be tested and constructing the spatiotemporal distribution matrix of the bridge real-time monitoring data is as follows: Deflection sensors are preset at different positions of the bridge to be tested, and a set of real-time monitoring data of the deflection sensors is obtained by sampling once every sampling interval τ, and the spatiotemporal distribution matrix X of the real-time monitoring data of the bridge is constructed; The spatiotemporal distribution matrix X of the real-time monitoring data of the bridge is: Among them, x i,j is the real-time monitoring data of the i-th deflection sensor at the j-th moment; m is the total number of preset deflection sensors at different positions of the bridge to be tested; and n is the total number of samples of each deflection sensor.

3. A bridge abnormal state detection method according to claim 1, characterized in that: The process of performing data preprocessing operation on the spatiotemporal distribution matrix of the bridge real-time monitoring data to obtain the preprocessed spatiotemporal distribution real-time monitoring data matrix is ​​as follows: In the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, each row of monitoring data is traversed to determine whether each row of monitoring data is all zero; if one row of monitoring data is all zero, the monitoring data of the corresponding row is removed from the spatiotemporal distribution matrix of the real-time monitoring data of the bridge, to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after division by zero; Using the most probable value filling method, data filling is performed on the time-space distribution matrix of the bridge real-time monitoring data after the division by zero, so as to obtain the time-space distribution matrix of the bridge real-time monitoring data after the data filling; Using a wavelet denoising method, the spatiotemporal distribution matrix of the bridge real-time monitoring data after the data filling is subjected to denoising, so as to obtain the spatiotemporal distribution matrix of the bridge real-time monitoring data after the denoising; Normalizing the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after the noise reduction processing to obtain a normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge; The real-time monitoring data with zero-point drift in the normalized time-space distribution matrix of the real-time monitoring data of the bridge is corrected to obtain the time-space distribution matrix of the real-time monitoring data of the bridge after preprocessing.

4. A bridge abnormal state detection method according to claim 3, characterized in that: The process of correcting the monitoring data with zero point drift in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge to obtain the preprocessed spatiotemporal distribution matrix of the real-time monitoring data of the bridge is as follows: Obtain historical monitoring data of the bridge to be tested in a zero-point drift state and historical monitoring data of the bridge to be tested in a normal state, respectively calculate the slope and mean of the historical monitoring data of the bridge to be tested in the zero-point drift state, and calculate the slope and mean of the historical monitoring data of the bridge to be tested in a normal state; According to the slope and mean of the historical monitoring data of the bridge to be tested in the zero drift state, a slope threshold and a variance threshold for determining whether the monitoring data of the bridge to be tested has zero drift are obtained; Using a time sliding window, traverse each row of monitoring data in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge, and calculate the slope and variance of the monitoring data in each time sliding window; Comparing the slope of the monitoring data in each time sliding window with a slope threshold of whether the monitoring data of the bridge to be tested has a zero drift, or comparing the variance of the monitoring data in each time sliding window with a variance threshold of whether the monitoring data of the bridge to be tested has a zero drift; If the slope of the monitoring data in each time sliding window exceeds the slope threshold of whether the monitoring data of the bridge to be tested has zero drift, or the variance of the real-time monitoring data in each time sliding window exceeds the variance threshold of whether the monitoring data of the bridge to be tested has zero drift, then the corresponding monitoring data has zero drift; For the monitoring data with zero point drift, the mean of the monitoring data before the zero point drift occurs is calculated to obtain a first mean; the mean of the monitoring data after the zero point drift occurs is calculated to obtain a second mean; and a data correction value is calculated based on the first mean and the second mean; The data correction value is used to correct the monitoring data in the normalized spatiotemporal distribution matrix of the real-time monitoring data of the bridge, so as to obtain the corrected spatiotemporal distribution matrix of the real-time monitoring data of the bridge, that is, to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after preprocessing; wherein, the correction process is specifically as follows: the monitoring data with zero point drift is subtracted from the data correction value to obtain the correction value, and the correction value is used to replace the monitoring data with zero point drift.

5. A bridge abnormal state detection method according to claim 1, characterized in that: The process of calculating the safety threshold of the mean value of the monitoring data of the bridge under test in a healthy state by using the historical monitoring data of the bridge under test in a healthy state is as follows: Obtain historical monitoring data of the bridge to be tested in a healthy state; Calculate the mean μ and variance σ of the historical monitoring data of the bridge to be tested in a healthy state; According to the mean μ and variance σ of the historical monitoring data of the bridge to be tested in the healthy state, a safety threshold value of the mean of the monitoring data of the bridge to be tested in the healthy state is obtained; The safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state is (μ-4σ, μ+4σ).

6. A bridge abnormal state detection method according to claim 1, characterized in that: After comparing the mean value of the monitoring data in each time sliding window with the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state to obtain the abnormal state detection result of the bridge to be tested, the abnormal state positioning step is also included; The abnormal state location step is as follows: According to the normal distribution, a data classifier is constructed to describe the operation status of the bridge; According to the data classifier for describing the operation status of the bridge, the spatiotemporal distribution matrix of the pre-processed real-time monitoring data of the bridge is subjected to classification processing to obtain a classification result; According to the classification results, construct a real-time monitoring data status matrix of the bridge; According to a preset coloring standard, the real-time monitoring data state matrix of the bridge is colorized to obtain a dynamic standard color atlas; Performing grayscale processing on the dynamic standard color map to obtain a grayscale map; generating a histogram under the grayscale map according to the grayscale map; Binarizing the histogram under the grayscale spectrum to obtain a damage spectrum; According to the damage map, the abnormal state positioning result of the bridge to be tested is obtained.

7. A bridge abnormal state detection method according to claim 6, characterized in that: According to the preset coloring standard, the real-time monitoring data state matrix of the bridge is colorized to obtain a dynamic standard color map, which is specifically as follows: Determine the numerical boundary of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge; divide the numerical boundary of the monitoring data at each measuring point into intervals according to the total number of colors in the preset color phase space in the preset coloring standard, and obtain the real-time monitoring data state matrix of the bridge after the interval division; According to the correspondence between the value of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge and the color of the preset color phase space, the real-time monitoring data state matrix of the bridge after the interval division is mapped to the preset color phase space to obtain a dynamic standard color map; wherein, the correspondence between the value of the monitoring data at each measuring point in the real-time monitoring data state matrix of the bridge and the color of the preset color phase space is: Among them, x:p ij is the pixel value of the i-th row and j-th column in the two-dimensional map; x ij is the monitoring data of the i-th row and j-th column in the real-time monitoring data state matrix of the bridge; N is the total number of colors in the preset color phase space; x max is the maximum value in the bridge real-time monitoring data state matrix; x min It is the minimum value in the real-time monitoring data status matrix of the bridge.

8. A bridge abnormal state detection system, characterized in that: include: A time-space matrix module is used to obtain the real-time monitoring data of the bridge to be tested and construct a time-space distribution matrix of the bridge real-time monitoring data; wherein, in the time-space distribution matrix of the bridge real-time monitoring data, each column represents the spatial distribution sequence of all monitoring data at a certain moment, and each row represents the time distribution sequence of monitoring data at a certain measuring point on the bridge to be tested; A data preprocessing module, used for performing data preprocessing operation on the spatiotemporal distribution matrix of the real-time monitoring data of the bridge to obtain the spatiotemporal distribution matrix of the real-time monitoring data of the bridge after preprocessing; A threshold module, used to calculate a safety threshold of the mean value of the monitoring data of the bridge to be tested in a healthy state by using the historical monitoring data of the bridge to be tested in a healthy state; A data statistics module is used to traverse each row of monitoring data in the spatiotemporal distribution matrix of the preprocessed real-time monitoring data of the bridge using a time sliding window, and perform statistical processing on the monitoring data in each time sliding window to obtain the mean value of the monitoring data in each time sliding window; The abnormality detection module is used to compare the mean value of the monitoring data in each time sliding window with the safety threshold of the mean value of the monitoring data of the bridge to be tested in the healthy state, so as to obtain the abnormal state detection result of the bridge to be tested.

9. A bridge abnormal state detection device, characterized in that: include: Memory for storing computer programs; A processor is used to implement the steps of the bridge abnormal state detection method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the bridge abnormal state detection method according to any one of claims 1 to 7 are implemented.

Images