A bridge health monitoring data processing method and system

Abstract

The present invention relates to the technical field of bridge health monitoring, and in particular to a method and system for processing bridge health monitoring data. The method comprises the following steps: collecting bridge health monitoring template data and bridge health status to form a bridge health monitoring template data set, and collecting bridge health monitoring data to be processed at the same time; using a feature detection method to identify and eliminate abnormal data in the bridge health monitoring data to be processed; after eliminating the abnormal data, using the maximum similarity method considering the association of indicators based on the bridge health monitoring template data to fill in the missing data, and obtain accurate bridge health monitoring data; constructing a bridge health prediction model, and then predicting the bridge health status based on the accurate bridge health monitoring data. The present invention can realize the identification of abnormal data without defining a training validation set, and further considers the relationship between various bridge health monitoring indicators to supplement the missing data, and has high accuracy, reliability and real-time performance.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge health monitoring, and in particular to a bridge health monitoring data processing method and system. Background Art

[0002] As an important transportation infrastructure, the safety and stability of bridges are of great significance to social and economic development and the safety of people's lives and property. In order to monitor the health status of bridges in real time, modern bridges are usually equipped with various sensors to collect data on key parameters such as stress, vibration, and temperature of bridges. However, in the actual data collection process, module failure, external interference, or errors in data transmission may cause abnormal values ​​that are significantly deviated from the normal range in the collected bridge health monitoring data. At the same time, sensor failure, data transmission interruption, module maintenance, or other unforeseen factors may cause missing values ​​in the collected bridge health monitoring data. The appearance of abnormal values ​​and missing values ​​will lead to deviations in subsequent data analysis and directly affect the integrity and continuity of the data set, so that the bridge health monitoring data cannot reflect the true state of the bridge, affecting the comprehensive and accurate assessment of the bridge status, and then leading to false warnings. Therefore, the elimination of abnormal values ​​and the supplementation of missing values ​​are the premise and important steps to judge the true health status of the bridge.

[0003] At present, the commonly used outlier identification methods can be divided into tests based on statistical methods, tests based on machine learning methods, and tests based on non-statistical and machine learning methods. Among them, tests based on statistical methods are usually based on certain statistical assumptions, such as data obeying normal distribution or having specific distribution characteristics, but these assumptions are often not established in practical applications, resulting in limited accuracy and effectiveness of the methods; tests based on machine learning methods often require the definition of training and validation sets to train and validate the model, which affects the practical application of the method; tests based on non-statistical and machine learning methods often rely on the subjective judgment and experience of analysts, resulting in subjective bias and uncertainty in the identification results. In addition, the currently commonly used missing value filling methods such as interpolation filling, mean or median filling, and model prediction filling are limited to filling missing values ​​based on the characteristics of a set of data itself, without considering the impact of changes in related factors on the data. Therefore, the supplemented data cannot reflect the characteristics of the original data to the greatest extent, and cannot accurately evaluate the health status of the bridge. Summary of the invention

[0004] In view of the defects in the prior art, the present invention provides a bridge health monitoring data processing method and system.

[0005] In order to achieve the above-mentioned purpose, in the first aspect, the present invention provides a method for processing bridge health monitoring data, the method comprising the following steps: collecting bridge health monitoring template data and bridge health status to form a bridge health monitoring template data set, and collecting bridge health monitoring data to be processed; using feature detection method to identify and eliminate abnormal data in the bridge health monitoring data to be processed; after eliminating the abnormal data, using the maximum similarity method considering the association of indicators based on the bridge health monitoring template data to fill in the missing data, and obtain accurate bridge health monitoring data; constructing a bridge health prediction model, and then predicting the bridge health status based on the accurate bridge health monitoring data. The present invention can realize the identification of abnormal data without defining a training validation set, and further considers the relationship between various bridge health monitoring indicators to supplement the missing data, and has high accuracy, reliability and real-time performance.

[0006] Optionally, the collecting of the bridge health monitoring template data and the bridge health status to form a bridge health monitoring template data set, and the collecting of the bridge health monitoring data to be processed comprises the following steps:

[0007] Determine bridge health monitoring indicators, and then collect bridge health conditions within multiple data collection cycles and bridge health monitoring template data of different bridge health monitoring indicators;

[0008] Taking the bridge health monitoring template data of all the bridge health monitoring indicators in a data collection cycle as a group of bridge health monitoring template data, and establishing a bridge health monitoring template data set in combination with the corresponding bridge health status;

[0009] The bridge health monitoring data of different bridge health monitoring indicators in a recent data collection cycle are collected as the bridge health monitoring data to be processed.

[0010] Establishing a bridge health monitoring template data set can provide data support for subsequent missing data supplementation.

[0011] Optionally, the identifying and removing abnormal data in the bridge health monitoring data to be processed by using a feature detection method comprises the following steps:

[0012] Taking the acquisition time of the bridge health monitoring data to be processed as the horizontal coordinate, the bridge health monitoring data to be processed of each bridge health monitoring indicator is marked in different plane rectangular coordinate systems, and the marked points are recorded as characteristic nodes;

[0013] Extracting key feature nodes from the feature nodes;

[0014] In the same plane rectangular coordinate system, two adjacent key feature nodes are connected by a straight line to obtain a key feature node line segment, and a key feature node line segment set of each bridge health monitoring indicator is obtained;

[0015] The abnormality index values ​​of the key characteristic node segments are calculated according to the key characteristic node segment set, so as to identify and eliminate the abnormal data in the bridge health monitoring data to be processed.

[0016] By extracting a set of key feature node segments to reflect the main change trends of various bridge health monitoring indicators, the interference of noise data can be effectively eliminated. Then, without defining a training and verification set, abnormal data can be efficiently identified by calculating the value of the abnormality degree index.

[0017] Optionally, extracting key feature nodes from the feature nodes comprises the following steps:

[0018] Obtaining a fitting curve of the bridge health monitoring data to be processed according to the characteristic nodes marked in the plane rectangular coordinate system;

[0019] Calculating characteristic values ​​of each characteristic node except the first and the last characteristic node in a plane rectangular coordinate system according to the fitting curve of the bridge health monitoring data to be processed;

[0020] A feature threshold is set. When the feature value is greater than the feature threshold, the feature node corresponding to the feature value is recorded as a key feature node, and the first and last feature nodes in the plane rectangular coordinate system are directly recorded as key feature nodes.

[0021] By extracting key feature nodes, a data basis is provided for obtaining a set of key feature node segments, and then, without the need to define a training validation set, abnormal data can be efficiently identified by calculating the value of the abnormality degree index.

[0022] Optionally, the calculating of abnormality index values ​​of key characteristic node segments according to the key characteristic node segment set, and then identifying and eliminating abnormal data in the bridge health monitoring data to be processed comprises the following steps:

[0023] Calculate the abnormality index value of the key feature node line segment according to the key feature node line segment set;

[0024] An abnormality index threshold is set. When the abnormality index value is greater than the abnormality index threshold, the bridge health monitoring data to be processed within the time period where the key characteristic node segment is located is taken as the abnormal data and deleted.

[0025] By extracting a set of key feature node segments to reflect the main change trends of various bridge health monitoring indicators, the interference of noise data can be effectively eliminated. Then, without defining a training and verification set, the abnormality degree index value can be calculated based on the key feature node segment set to efficiently identify abnormal data.

[0026] Optionally, the abnormality index value satisfies the following relationship:

[0027]

[0028] Among them, EX a is the abnormality index value of the ath key feature node segment, B is the number of key feature node segments adjacent to a key feature node segment, l a is the length of the ath key feature node segment, l b is the length of the bth key feature node segment adjacent to the ath key feature node segment, l c is the length of the cth key feature node segment adjacent to the bth key feature node segment, k a is the slope of the ath key feature node line segment, k b is the slope of the bth key feature node segment adjacent to the ath key feature node segment, k c is the slope of the cth key feature node segment adjacent to the bth key feature node segment.

[0029] The numerical calculation of the abnormal degree index is simple and convenient, which can improve the efficiency of abnormal data identification.

[0030] Optionally, after removing the abnormal data, filling the missing data using the maximum similarity method considering the correlation of indicators based on the bridge health monitoring template data to obtain accurate bridge health monitoring data includes the following steps:

[0031] After removing the abnormal data, marking the bridge health monitoring index with missing data in the bridge health monitoring data to be processed as an index to be filled, and calculating the standard deviation of the index to be filled;

[0032] Calculating the comprehensive similarity between the bridge health monitoring data to be processed and each group of bridge health monitoring template data in the bridge health monitoring template data set, and recording a group of bridge health monitoring template data corresponding to the largest comprehensive similarity as comparative monitoring data;

[0033] Using the data corresponding to the missing data in the comparative monitoring data as basic filling data;

[0034] Determine the filling value range of the missing data of each of the indicators to be filled using the basic filling data and the standard deviation of the indicators to be filled;

[0035] Continuously randomly select values ​​within the filling value range to fill the corresponding missing data and perform abnormal data identification until the abnormal data can no longer be identified.

[0036] By calculating the comprehensive similarity between the bridge health monitoring data to be processed and each group of bridge health monitoring template data in the bridge health monitoring template data set, the missing data can be supplemented while considering the mutual influence of various bridge health monitoring indicators, so that the filled data can reflect the characteristics of the original data to the greatest extent.

[0037] Optionally, the comprehensive similarity satisfies the following relationship:

[0038]

[0039] Wherein, S is the comprehensive similarity, N is the number of the bridge health monitoring indicators, M is the number of times each bridge health monitoring indicator is collected in one data collection cycle, and a ij is the bridge health monitoring data to be processed of the i-th bridge health monitoring indicator collected for the j-th time, b ij It is the bridge health monitoring template data of the i-th bridge health monitoring indicator collected for the j-th time within the data collection cycle.

[0040] Optionally, the constructing of a bridge health prediction model and then predicting the health status of the bridge based on the bridge health monitoring accurate data comprises the following steps:

[0041] The health status of the bridge is digitized, and the bridge health monitoring index is used as input, the bridge health status level is used as output, and a BP neural network is used to construct a bridge health prediction primitive model;

[0042] Using the bridge health monitoring template data set to train and verify the bridge health prediction original model to obtain a bridge health prediction model;

[0043] Based on the accurate bridge health monitoring data, the bridge health prediction model is used to predict the health status of the bridge.

[0044] After accurately identifying and eliminating abnormal data in the bridge health monitoring data to be processed and supplementing the missing data, accurate bridge health monitoring data can be obtained, and then the bridge health status can be predicted using the bridge health prediction model, which makes it easier for relevant personnel to maintain the bridge in a timely manner when necessary and improve bridge safety.

[0045] In the second aspect, the present invention also provides a bridge health monitoring data processing system, the system uses a bridge health monitoring data processing method provided by the present invention, the system includes: a data acquisition module, the data acquisition module is used to collect bridge health monitoring template data and bridge health status to form a bridge health monitoring template data set, and at the same time collect bridge health monitoring data to be processed; an abnormal identification and elimination module, the abnormal identification and elimination module is used to use a feature detection method to identify and eliminate abnormal data in the bridge health monitoring data to be processed; a missing supplement module, the missing supplement module is used to eliminate the abnormal data, and then fill the missing data based on the bridge health monitoring template data using the maximum similarity method considering the association of indicators to obtain accurate bridge health monitoring data; a health prediction module, the health prediction module is used to build a bridge health prediction model, and then predict the bridge health status based on the accurate bridge health monitoring data; a data storage module, the data storage module is used to store the bridge health monitoring template data set, the bridge health monitoring data to be processed, the bridge health monitoring accurate data and the predicted bridge health status; a data output module, the data output module is used to output the data stored in the data storage module. The system provided by the present invention can stably execute the method provided by the present invention, and improve the practical application ability of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 A schematic diagram of a flow chart of a bridge health monitoring data processing method according to an embodiment of the present invention;

[0048] Figure 2 The figure is a schematic diagram of a framework of a bridge health monitoring data processing system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are only for illustration and are not intended to limit the present invention. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is obvious to those of ordinary skill in the art that these specific details do not need to be adopted to implement the present invention. In other examples, in order to avoid confusing the present invention, known circuits, software or methods are not specifically described.

[0050] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "in one embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or subcombination. In addition, it should be understood by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale.

[0051] It should be noted in advance that, in an optional embodiment, except for independent explanations, the same symbols or letters appearing in all formulas have the same meanings and values.

[0052] In an alternative embodiment, see Figure 1 The present invention provides a bridge health monitoring data processing method, the method comprising the following steps:

[0053] S1. Collect bridge health monitoring template data and bridge health status to form a bridge health monitoring template data set, and collect bridge health monitoring data to be processed.

[0054] Wherein, step S1 specifically includes the following steps:

[0055] S11. Determine bridge health monitoring indicators, and then collect bridge health conditions within multiple data collection cycles and bridge health monitoring template data of different bridge health monitoring indicators.

[0056] Specifically, in this embodiment, the bridge health monitoring indicators include displacement, deflection, stress, strain, cable force, acceleration, temperature and wind speed. Therefore, the bridge health monitoring template data within a data collection cycle actually contains the bridge health monitoring template data of 8 bridge health monitoring indicators. In addition, the data collection cycle is specifically 24 hours, and the bridge health monitoring template data is the historical bridge health monitoring data without data anomalies and data missing. The bridge health monitoring indicators listed in this step can all be obtained using corresponding sensors, and other bridge health monitoring indicators can also be selected in other optional embodiments.

[0057] More specifically, since the operation of measuring bridge health monitoring indicators using sensors is obtained at equal time intervals. If 24 hours is used as a data collection cycle, and each bridge health monitoring indicator is measured every 2 hours, then in one data collection cycle, one bridge health monitoring indicator will have 12 bridge health monitoring template data, that is, in one data collection cycle, the bridge health monitoring template data corresponding to one bridge health monitoring indicator is a numerical sequence arranged in chronological order of collection time. Similarly, the bridge health monitoring data to be processed corresponding to a subsequent bridge health monitoring indicator is also a numerical sequence.

[0058] Furthermore, when collecting the bridge health monitoring template data of different bridge health monitoring indicators in multiple data collection cycles, the bridge health status in each data collection cycle is recorded at the same time. In addition, in order to improve the integrity of the data and establish an accurate and reliable bridge health monitoring template data set to provide data support for the subsequent missing data supplementation, thereby improving the accuracy of missing data supplementation, it is best to obtain the bridge health status and bridge health monitoring template data of different bridge health monitoring indicators in four data collection cycles in the first and second half of each month of the year.

[0059] S12. Taking the bridge health monitoring template data of all the bridge health monitoring indicators in a data collection cycle as a group of bridge health monitoring template data, and establishing a bridge health monitoring template data set in combination with the corresponding bridge health conditions.

[0060] Specifically, in this embodiment, a set of bridge health monitoring template data includes a numerical sequence corresponding to each bridge health monitoring indicator within a data collection cycle.

[0061] S13, collecting bridge health monitoring data of different bridge health monitoring indicators in a most recent data collection cycle as the bridge health monitoring data to be processed.

[0062] Specifically, in this embodiment, the most recent data collection cycle is also the current data collection cycle.

[0063] S2. Using a feature detection method to identify and remove abnormal data in the bridge health monitoring data to be processed.

[0064] Wherein, step S2 specifically includes the following steps:

[0065] S21. Using the acquisition time of the bridge health monitoring data to be processed as the horizontal coordinate, mark the bridge health monitoring data to be processed of each bridge health monitoring indicator in different plane rectangular coordinate systems, and record the marked points as feature nodes.

[0066] Specifically, in this embodiment, as described in S11, the bridge health monitoring template data corresponding to each bridge health monitoring indicator is a data sequence. Therefore, the collection time of the bridge health monitoring data to be processed can be used as the horizontal coordinate, and the bridge health monitoring data to be processed can be used as the vertical coordinate. The bridge health monitoring data to be processed of the bridge health monitoring indicator can be marked in a plane rectangular coordinate system to obtain a series of feature nodes.

[0067] S22: extract key feature nodes from the feature nodes.

[0068] Wherein, step S22 specifically includes the following steps:

[0069] S221. Obtain a fitting curve of the bridge health monitoring data to be processed according to the characteristic nodes marked in the plane rectangular coordinate system.

[0070] Specifically, in this embodiment, for different bridge health monitoring indicators, Matlab is used to fit the characteristic nodes in the plane rectangular coordinate system to obtain the bridge health monitoring data to be processed fitting curves of 8 bridge health monitoring indicators.

[0071] S222, calculating the characteristic values ​​of each characteristic node except the first and the last characteristic node in the plane rectangular coordinate system according to the fitting curve of the bridge health monitoring data to be processed.

[0072] Specifically, in this embodiment, the characteristic value of the characteristic node satisfies the following relationship:

[0073]

[0074] Among them, T ij is the characteristic value of the jth characteristic node of the i-th bridge health monitoring index along the horizontal axis of the plane rectangular coordinate system, t ij is the time for collecting the bridge health monitoring data to be processed for the i-th bridge health monitoring indicator for the jth time, t i,j-1 is the time for collecting the bridge health monitoring data to be processed for the i-th bridge health monitoring indicator for the j-1th time, t i,j+1 is the time for collecting the bridge health monitoring data to be processed for the i-th bridge health monitoring indicator for the j+1th time, a ij is the bridge health monitoring data to be processed for the i-th bridge health monitoring indicator collected for the j-th time, a i,j-1 is the bridge health monitoring data to be processed for the i-th bridge health monitoring indicator collected for the j-1th time, a i,j+1 is the bridge health monitoring data to be processed of the i-th bridge health monitoring indicator collected for the j+1th time. In this step, j=2, 3, ..., M, and M is the number of times each bridge health monitoring indicator is collected in one data collection cycle.

[0075] S223, setting a feature threshold, when the feature value is greater than the feature threshold, recording the feature node corresponding to the feature value as a key feature node, and at the same time directly recording the first and last feature nodes in the plane rectangular coordinate system as key feature nodes.

[0076] Specifically, in this embodiment, different feature thresholds need to be set for bridges in different locations, different bridge health monitoring indicators, and different seasons. The set feature threshold should be greater than 0.5, and its specific value is given by experts after field investigation, which can improve the accuracy of abnormal data identification.

[0077] For a certain bridge health monitoring indicator, after setting the characteristic threshold, the first and last characteristic nodes in the plane rectangular coordinate system are directly recorded as the key characteristic nodes of the bridge health monitoring data to be processed corresponding to the bridge health monitoring indicator. If the characteristic value of a characteristic node among the remaining characteristic nodes is greater than the characteristic threshold corresponding to the bridge health monitoring indicator, the characteristic node is recorded as the key characteristic node of the bridge health monitoring data to be processed corresponding to the bridge health monitoring indicator. By extracting the key characteristic nodes, a data basis is provided for the subsequent acquisition of the key characteristic node line segment set, and then, without the need to define a training validation set, the abnormal data can be efficiently identified by calculating the abnormal degree index value.

[0078] S23. In the same plane rectangular coordinate system, two adjacent key feature nodes are connected by a straight line to obtain a key feature node line segment, and a key feature node line segment set of each bridge health monitoring indicator is obtained.

[0079] Specifically, in this embodiment, only one characteristic node of a bridge health monitoring indicator is marked in a plane rectangular coordinate system, and the obtained key characteristic node segment set is a broken line composed of each key characteristic node segment, which can reflect the main change trend of the bridge health monitoring indicator, effectively eliminate the interference of noise data, and facilitate the subsequent identification of abnormal data.

[0080] S24. Calculate the abnormality index value of the key characteristic node line segment according to the key characteristic node line segment set, and then identify and eliminate abnormal data in the bridge health monitoring data to be processed.

[0081] S241. Calculate the abnormality index value of the key feature node line segment according to the key feature node line segment set.

[0082] Specifically, in this embodiment, the abnormality index value satisfies the following relationship:

[0083]

[0084] Among them, EX a is the abnormality index value of the ath key feature node segment, B is the number of key feature node segments adjacent to a key feature node segment, l a is the length of the ath key feature node segment, l b is the length of the bth key feature node segment adjacent to the ath key feature node segment, l c is the length of the cth key feature node segment adjacent to the bth key feature node segment, k a is the slope of the ath key feature node line segment, k b is the slope of the bth key feature node segment adjacent to the ath key feature node segment, k c is the slope of the cth key feature node segment adjacent to the bth key feature node segment.

[0085] Furthermore, by extracting a set of key feature node segments to reflect the main change trends of various bridge health monitoring indicators, the interference of noise data can be effectively eliminated, and then without defining a training verification set, abnormal data can be efficiently identified by calculating the abnormality degree index value. Moreover, the calculation of the abnormality degree index value of this embodiment is simple, which can further improve the efficiency of abnormal data identification.

[0086] S242, setting an abnormality index threshold value, when the abnormality index value is greater than the abnormality index threshold value, taking the bridge health monitoring data to be processed within the time period where the key feature node segment is located as the abnormal data and deleting it.

[0087] Specifically, in this embodiment, the abnormality index threshold is set to 0.75.

[0088] S3. After removing the abnormal data, the missing data are filled in based on the bridge health monitoring template data using the maximum similarity method that considers the correlation of indicators to obtain accurate bridge health monitoring data.

[0089] Wherein, step S3 specifically includes the following steps:

[0090] S31. After removing the abnormal data, mark the bridge health monitoring indicators with missing data in the bridge health monitoring data to be processed as indicators to be filled, and calculate the standard deviation of the indicators to be filled.

[0091] S32, calculating the comprehensive similarity between the bridge health monitoring data to be processed and each group of bridge health monitoring template data in the bridge health monitoring template data set, and recording a group of bridge health monitoring template data corresponding to the largest comprehensive similarity as comparative monitoring data.

[0092] Specifically, in this embodiment, the comprehensive similarity satisfies the following relationship:

[0093]

[0094] Among them, S is the comprehensive similarity, N is the number of bridge health monitoring indicators, b ij is the bridge health monitoring template data of the i-th bridge health monitoring indicator collected for the j-th time in a data collection cycle, and in this step, j=1, 2, 3, ..., M.

[0095] By calculating the comprehensive similarity between the bridge health monitoring data to be processed and each group of bridge health monitoring template data in the bridge health monitoring template data set, a group of bridge health monitoring template data that is most similar to the bridge health monitoring data to be processed is found, and then the missing data is supplemented while considering the mutual influence of various bridge health monitoring indicators, so that the filled data can reflect the characteristics of the original data to the greatest extent.

[0096] S33, using the data in the comparative monitoring data corresponding to the missing data as basic filling data.

[0097] Specifically, in this embodiment, for ease of understanding, it is assumed that problems occurred in the third and fourth collections of the bridge health monitoring data to be processed of stress in the bridge health monitoring indicators, resulting in the missing of the stress data collected in these two times. In this case, the stress data collected for the third and fourth times in the comparison monitoring data will be used as the basic filling data.

[0098] S34, using the basic filling data and the standard deviation of the indicator to be filled, determine the filling value range of the missing data of each indicator to be filled.

[0099] Specifically, in this embodiment, the missing data at different positions of an indicator to be filled have different filling value ranges. Assuming that the bridge health monitoring data to be processed collected for the jth time for the dth indicator to be filled is missing, then the basic filling data corresponding to the missing data should be the comparative monitoring data v collected for the jth time for the dth indicator to be filled. dj , then the filling value range of the missing data of the indicator to be filled is [v dj -2σ d ,v dj +2σ d ],σ d is the standard deviation of the to-be-filled indicator of the dth to-be-filled indicator.

[0100] S35, continuously randomly selecting values ​​within the filling value range to fill the corresponding missing data and performing abnormal data identification until the abnormal data cannot be identified.

[0101] Specifically, in this embodiment, if the bridge health monitoring data collected for the jth time for the dth indicator to be filled is missing, then in [v dj -2σ d ,v dj +2σ d ] randomly selects a value as the index to be filled, and the j-th collected bridge health monitoring data to be processed is used to supplement the value sequence of the index to be filled, and then the abnormal data is identified according to the method of step S2. If the abnormal data is identified, then in [v dj -2σ d ,v dj +2σ d ] again randomly selects a value as the j-th collected bridge health monitoring data to be processed for the indicator to be filled, supplements the value sequence of the indicator to be filled and identifies abnormal data until no abnormal data can be identified. The filling of missing data at different positions can be analogized, thereby realizing the filling of missing data at different positions of different indicators to be filled, and finally obtaining accurate bridge health monitoring data.

[0102] S4. Construct a bridge health prediction model to predict the health status of the bridge based on the precise data of bridge health monitoring.

[0103] Wherein, step S4 specifically includes the following steps:

[0104] S41, digitize the health status of the bridge, and then use the bridge health monitoring index as input and the bridge health status level as output, and use the BP neural network to build a bridge health prediction primitive model.

[0105] Specifically, in this embodiment, different positive integers are used to represent different bridge health conditions to achieve digitization of bridge health conditions, and then the bridge health monitoring index is used as input and the bridge health status level is used as output, and the BP neural network is used to build a bridge health prediction original model.

[0106] S42: Use the bridge health monitoring template data set to train and verify the bridge health prediction original model to obtain a bridge health prediction model.

[0107] S43: Based on the accurate bridge health monitoring data, use the bridge health prediction model to predict the health status of the bridge.

[0108] Specifically, in this embodiment, accurate data from bridge health monitoring is input into a bridge health prediction model, so that the bridge health prediction model outputs the bridge health status, thereby realizing the prediction of the bridge health status, so that relevant personnel can maintain the bridge in a timely manner and improve the safety of the bridge.

[0109] It should be noted that, in some cases, the actions described in the specification can be performed in a different order and still achieve the desired results. In this embodiment, the order of steps given is only to make the embodiment appear clearer and easier to explain, rather than to limit it.

[0110] In an alternative embodiment, see Figure 2 The present invention also provides a bridge health monitoring data processing system, which uses a bridge health monitoring data processing method provided by the present invention. The system includes a data acquisition module A1, an abnormality identification and elimination module A2, a missing supplement module A3, a health prediction module A4, a data storage module A5 and a data output module A6.

[0111] The data acquisition module A1 is used to acquire bridge health monitoring template data and bridge health conditions to form a bridge health monitoring template data set, and at the same time acquire bridge health monitoring data to be processed.

[0112] The abnormality identification and elimination module A2 is used to identify and eliminate abnormal data in the bridge health monitoring data to be processed by using a feature detection method.

[0113] The missing data supplement module A3 is used to remove the abnormal data and fill in the missing data based on the bridge health monitoring template data using the maximum similarity method considering the correlation of indicators to obtain accurate bridge health monitoring data.

[0114] The health prediction module A4 is used to construct a bridge health prediction model, and then predict the health status of the bridge based on the precise data of bridge health monitoring.

[0115] The data storage module A5 is used to store the bridge health monitoring template data set, the bridge health monitoring data to be processed, the bridge health monitoring precise data and the predicted bridge health status.

[0116] The data output module A6 is used to output the data stored in the data storage module A5.

[0117] Specifically, in this embodiment, each module of the system realizes information transmission through a data line. The data acquisition module A1 includes a sensor and a module body. The sensor and the module body use the Internet of Things for digital transmission. The module body includes a digital display screen and a USB interface. The bridge health monitoring data to be processed can be collected in real time by the sensor. The bridge health monitoring template data and the bridge health status can be manually input into the data acquisition module A1 through the digital display screen, or can be input into the data acquisition module A1 from other data storage devices using a USB interface. The data output module A6 also includes a digital display screen for displaying the bridge health monitoring data to be processed, the bridge health monitoring accurate data and the predicted bridge health status. Relevant personnel can also use the data output module A6 to query other data in the data storage module A5 as needed.

[0118] Further, the data acquisition module A1 specifically performs the content described in step S1, the abnormal identification and elimination module A2 specifically performs the content described in step S2, the missing supplement module A3 specifically performs the content described in step S3, and the health prediction module A4 specifically performs the content described in step S4. The system provided by the present invention can stably execute the method provided by the present invention and improve the practical application ability of the present invention.

[0119] In summary, the present invention reflects the main change trends of various bridge health monitoring indicators by extracting a set of key feature node segments, effectively eliminating the interference of noise data, and then directly based on the original monitoring data without defining a training verification set, the abnormal data can be identified and eliminated in real time and efficiently by calculating the abnormal degree index value. After eliminating the abnormal data, the present invention calculates the comprehensive similarity between the bridge health monitoring data to be processed and each group of bridge health monitoring template data in the bridge health monitoring template data set, so as to supplement the missing data while considering the mutual influence of each bridge health monitoring indicator, so that the obtained bridge health monitoring accurate data reflects the characteristics of the original data to the greatest extent. Therefore, the present method has high accuracy, reliability and real-time performance. In addition, the system provided by the present invention is a system compatible with the method provided by the present invention, which can stably execute the method provided by the present invention and improve the practical application ability of the present invention.

[0120] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A bridge health monitoring data processing method, characterized in that: The steps include: Collect bridge health monitoring template data and bridge health status to form a bridge health monitoring template data set, and collect bridge health monitoring data to be processed; Using a feature detection method to identify and remove abnormal data in the bridge health monitoring data to be processed; After the abnormal data is removed, the missing data is filled in based on the bridge health monitoring template data using the maximum similarity method considering the correlation of indicators to obtain accurate bridge health monitoring data; Constructing a bridge health prediction model to predict the health status of the bridge based on the accurate data of bridge health monitoring; The collecting of the bridge health monitoring template data and the bridge health status to form a bridge health monitoring template data set and the collecting of the bridge health monitoring data to be processed include the following steps: Determine bridge health monitoring indicators, and then collect bridge health conditions within multiple data collection cycles and bridge health monitoring template data of different bridge health monitoring indicators; Taking the bridge health monitoring template data of all the bridge health monitoring indicators in a data collection cycle as a group of bridge health monitoring template data, and establishing a bridge health monitoring template data set in combination with the corresponding bridge health status; Collecting bridge health monitoring data of different bridge health monitoring indicators in a recent data collection cycle as the bridge health monitoring data to be processed; The method of using the feature detection method to identify and remove abnormal data in the bridge health monitoring data to be processed includes the following steps: Taking the acquisition time of the bridge health monitoring data to be processed as the horizontal coordinate, the bridge health monitoring data to be processed of each bridge health monitoring indicator is marked in different plane rectangular coordinate systems, and the marked points are recorded as characteristic nodes; Extracting key feature nodes from the feature nodes; In the same plane rectangular coordinate system, two adjacent key feature nodes are connected by a straight line to obtain a key feature node line segment, and a key feature node line segment set of each bridge health monitoring indicator is obtained; Calculating the abnormality index value of the key characteristic node segment according to the key characteristic node segment set, thereby identifying and eliminating abnormal data in the bridge health monitoring data to be processed; The step of extracting key feature nodes from the feature nodes comprises the following steps: Obtaining a fitting curve of the bridge health monitoring data to be processed according to the characteristic nodes marked in the plane rectangular coordinate system; Calculating characteristic values ​​of each characteristic node except the first and the last characteristic node in a plane rectangular coordinate system according to the fitting curve of the bridge health monitoring data to be processed; A feature threshold is set. When the feature value is greater than the feature threshold, the feature node corresponding to the feature value is recorded as a key feature node, and the first and last feature nodes in the plane rectangular coordinate system are directly recorded as key feature nodes; The step of calculating the abnormality index value of the key characteristic node segment according to the key characteristic node segment set, and then identifying and eliminating abnormal data in the bridge health monitoring data to be processed comprises the following steps: Calculate the abnormality index value of the key feature node line segment according to the key feature node line segment set; Setting an abnormality index threshold, when the abnormality index value is greater than the abnormality index threshold, treating the bridge health monitoring data to be processed within the time period where the key feature node line segment is located as the abnormal data and deleting it; The abnormality index value satisfies the following relationship: in, For the The abnormality index value of a key feature node segment, B is the number of key feature node segments adjacent to a key feature node segment, For the The length of the key feature node line segment, For the The length of the bth key feature node segment adjacent to the bth key feature node segment, is the length of the cth key feature node segment adjacent to the bth key feature node segment, For the The slope of the key feature node line segment, For the The slope of the bth key feature node segment adjacent to the bth key feature node segment, is the slope of the cth key feature node line segment adjacent to the bth key feature node line segment; After the abnormal data is removed, the missing data is filled using the maximum similarity method considering the correlation of indicators based on the bridge health monitoring template data to obtain accurate bridge health monitoring data, which includes the following steps: After removing the abnormal data, marking the bridge health monitoring index with missing data in the bridge health monitoring data to be processed as an index to be filled, and calculating the standard deviation of the index to be filled; Calculating the comprehensive similarity between the bridge health monitoring data to be processed and each group of bridge health monitoring template data in the bridge health monitoring template data set, and recording a group of bridge health monitoring template data corresponding to the largest comprehensive similarity as comparative monitoring data; Using the data corresponding to the missing data in the comparative monitoring data as basic filling data; Determine the filling value range of the missing data of each of the indicators to be filled using the basic filling data and the standard deviation of the indicators to be filled; Continuously randomly selecting values ​​within the filling value range to fill in the corresponding missing data and identify abnormal data until no abnormal data can be identified; The comprehensive similarity satisfies the following relationship: Wherein, S is the comprehensive similarity, N is the number of the bridge health monitoring indicators, and M is the number of times each bridge health monitoring indicator is collected within a data collection cycle. is the bridge health monitoring data to be processed of the i-th bridge health monitoring indicator collected for the j-th time, is the bridge health monitoring template data of the i-th bridge health monitoring indicator collected for the j-th time within the data collection cycle; The construction of the bridge health prediction model and the prediction of the bridge health status based on the bridge health monitoring accurate data include the following steps: The health status of the bridge is digitized, and the bridge health monitoring index is used as input, the bridge health status level is used as output, and a BP neural network is used to construct a bridge health prediction primitive model; Using the bridge health monitoring template data set to train and verify the bridge health prediction original model to obtain a bridge health prediction model; Based on the accurate bridge health monitoring data, the bridge health prediction model is used to predict the health status of the bridge.

2. A bridge health monitoring data processing system, the system using the bridge health monitoring data processing method according to claim 1, characterized in that: include: A data acquisition module, wherein the data acquisition module is used to collect bridge health monitoring template data and bridge health conditions to form a bridge health monitoring template data set, and to collect bridge health monitoring data to be processed; The collecting of the bridge health monitoring template data and the bridge health status to form a bridge health monitoring template data set and the collecting of the bridge health monitoring data to be processed include the following steps: Determine bridge health monitoring indicators, and then collect bridge health conditions within multiple data collection cycles and bridge health monitoring template data of different bridge health monitoring indicators; Taking the bridge health monitoring template data of all the bridge health monitoring indicators in a data collection cycle as a group of bridge health monitoring template data, and establishing a bridge health monitoring template data set in combination with the corresponding bridge health status; Collecting bridge health monitoring data of different bridge health monitoring indicators in a recent data collection cycle as the bridge health monitoring data to be processed; An abnormality identification and elimination module, the abnormality identification and elimination module is used to identify and eliminate abnormal data in the bridge health monitoring data to be processed by using a feature detection method; The method of using the feature detection method to identify and remove abnormal data in the bridge health monitoring data to be processed includes the following steps: Taking the acquisition time of the bridge health monitoring data to be processed as the horizontal coordinate, the bridge health monitoring data to be processed of each bridge health monitoring indicator is marked in different plane rectangular coordinate systems, and the marked points are recorded as characteristic nodes; Extracting key feature nodes from the feature nodes; In the same plane rectangular coordinate system, two adjacent key feature nodes are connected by a straight line to obtain a key feature node line segment, and a key feature node line segment set of each bridge health monitoring indicator is obtained; Calculating the abnormality index value of the key characteristic node segment according to the key characteristic node segment set, thereby identifying and eliminating abnormal data in the bridge health monitoring data to be processed; The step of extracting key feature nodes from the feature nodes comprises the following steps: Obtaining a fitting curve of the bridge health monitoring data to be processed according to the characteristic nodes marked in the plane rectangular coordinate system; Calculating characteristic values ​​of each characteristic node except the first and the last characteristic node in a plane rectangular coordinate system according to the fitting curve of the bridge health monitoring data to be processed; A feature threshold is set. When the feature value is greater than the feature threshold, the feature node corresponding to the feature value is recorded as a key feature node, and the first and last feature nodes in the plane rectangular coordinate system are directly recorded as key feature nodes; The step of calculating the abnormality index value of the key characteristic node segment according to the key characteristic node segment set, and then identifying and eliminating abnormal data in the bridge health monitoring data to be processed comprises the following steps: Calculate the abnormality index value of the key feature node line segment according to the key feature node line segment set; Setting an abnormality index threshold, when the abnormality index value is greater than the abnormality index threshold, treating the bridge health monitoring data to be processed within the time period where the key feature node line segment is located as the abnormal data and deleting it; The abnormality index value satisfies the following relationship: in, For the The abnormality index value of a key feature node segment, B is the number of key feature node segments adjacent to a key feature node segment, For the The length of the key feature node line segment, For the The length of the bth key feature node segment adjacent to the bth key feature node segment, is the length of the cth key feature node segment adjacent to the bth key feature node segment, For the The slope of the key feature node line segment, For the The slope of the bth key feature node segment adjacent to the bth key feature node segment, is the slope of the cth key feature node line segment adjacent to the bth key feature node line segment; A missing data supplementation module is used to remove the abnormal data and fill in the missing data based on the bridge health monitoring template data by using a maximum similarity method that considers the association of indicators to obtain accurate bridge health monitoring data; After the abnormal data is removed, the missing data is filled using the maximum similarity method considering the correlation of indicators based on the bridge health monitoring template data to obtain accurate bridge health monitoring data, which includes the following steps: After removing the abnormal data, marking the bridge health monitoring index with missing data in the bridge health monitoring data to be processed as an index to be filled, and calculating the standard deviation of the index to be filled; Calculating the comprehensive similarity between the bridge health monitoring data to be processed and each group of bridge health monitoring template data in the bridge health monitoring template data set, and recording a group of bridge health monitoring template data corresponding to the largest comprehensive similarity as comparative monitoring data; Using the data corresponding to the missing data in the comparative monitoring data as basic filling data; Determine the filling value range of the missing data of each of the indicators to be filled using the basic filling data and the standard deviation of the indicators to be filled; Continuously randomly selecting values ​​within the filling value range to fill in the corresponding missing data and identify abnormal data until no abnormal data can be identified; The comprehensive similarity satisfies the following relationship: Wherein, S is the comprehensive similarity, N is the number of the bridge health monitoring indicators, and M is the number of times each bridge health monitoring indicator is collected within a data collection cycle. is the bridge health monitoring data to be processed of the i-th bridge health monitoring indicator collected for the j-th time, is the bridge health monitoring template data of the i-th bridge health monitoring indicator collected for the j-th time within the data collection cycle; A health prediction module, which is used to construct a bridge health prediction model and further predict the health status of the bridge based on the accurate data of bridge health monitoring; The construction of the bridge health prediction model and the prediction of the bridge health status based on the bridge health monitoring accurate data include the following steps: The health status of the bridge is digitized, and the bridge health monitoring index is used as input, the bridge health status level is used as output, and a BP neural network is used to construct a bridge health prediction primitive model; Using the bridge health monitoring template data set to train and verify the bridge health prediction original model to obtain a bridge health prediction model; Based on the accurate bridge health monitoring data, using the bridge health prediction model to predict the health status of the bridge; A data storage module, the data storage module is used to store the bridge health monitoring template data set, the bridge health monitoring data to be processed, the bridge health monitoring accurate data and the predicted bridge health status; A data output module is used to output the data stored in the data storage module.

Images