An online diagnosis method for bridge structure abnormalities based on beam longitudinal displacement monitoring

Abstract

The present invention relates to the technical field of bridge structure monitoring data processing and abnormal state identification, and specifically to an online bridge structure abnormality diagnosis method based on beam longitudinal displacement monitoring, comprising: obtaining a plurality of beam longitudinal displacement monitoring data; preprocessing the plurality of beam longitudinal displacement monitoring data to obtain beam longitudinal displacement sampling data; obtaining a bridge longitudinal displacement feature vector based on the beam longitudinal displacement sampling data; and inputting the bridge longitudinal displacement feature vector into a bridge structure abnormality diagnosis network model to obtain the bridge structure state. The plurality of beam longitudinal displacement monitoring data are used to construct a feature vector reflecting the abnormal state of the beam structure, and the abnormal state of the bridge structure is diagnosed online using a trained bridge structure diagnosis network model, providing bridge maintenance personnel with a method for quickly determining the abnormal state of the beam structure, thereby improving diagnostic efficiency and enabling timely risk warning.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge structure monitoring data processing and abnormal state identification, and in particular to an online diagnosis method for bridge structure abnormalities based on beam longitudinal displacement monitoring. Background Art

[0002] Longitudinal displacement of beams is a key monitoring item in bridge structural health monitoring systems. Bridge maintenance personnel typically deploy sensors at pre-set locations on the bridge for monitoring. Based on the monitoring data, they determine whether the bridge structure is abnormal. However, existing techniques rely on combining monitoring data with expert experience and structural calculations to determine whether a structural anomaly exists. Furthermore, in the case of structural anomalies, manual on-site inspections are required to determine the type of structural anomaly. This method of determining bridge structural anomalies is inefficient, relies heavily on professional experience, and delays risk warnings. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the present invention proposes an online diagnosis method for bridge structure anomalies based on beam longitudinal displacement monitoring to improve diagnosis efficiency.

[0004] The technical solution adopted by the present invention is an online diagnosis method for bridge structure abnormalities based on beam longitudinal displacement monitoring.

[0005] In a first possible implementation, an online diagnosis method for bridge structural abnormalities based on beam longitudinal displacement monitoring includes:

[0006] Obtaining multiple beam longitudinal displacement monitoring data;

[0007] Preprocessing the monitoring data of multiple beam longitudinal displacements to obtain beam longitudinal displacement sampling data;

[0008] Obtaining the longitudinal displacement characteristic vector of the bridge according to the longitudinal displacement sampling data of the beam body;

[0009] The bridge longitudinal displacement feature vector is input into the bridge structure abnormality diagnosis network model to obtain the bridge structure status.

[0010] In combination with the first possible implementation, in the second possible implementation, multiple beam longitudinal displacement monitoring data are obtained, including:

[0011] At least one temperature monitoring point and at least two displacement monitoring points are respectively provided at the left and right beam ends of the bridge; temperature monitoring data is obtained in real time through the temperature sensors arranged at each temperature monitoring point, and displacement monitoring data is obtained through the displacement sensors arranged at each displacement monitoring point.

[0012] In combination with the first implementable manner, in a third implementable manner, preprocessing is performed on a plurality of beam longitudinal displacement monitoring data to obtain beam longitudinal displacement sampling data, including:

[0013] A pulse coupled neural network is used to perform adaptive noise reduction on multiple beam longitudinal displacement monitoring data to obtain alternative beam longitudinal displacement data.

[0014] The alternative data of the longitudinal displacement of the beam are processed simultaneously and at the same frequency to obtain the sampling data of the longitudinal displacement of the beam.

[0015] In combination with the third implementable method, in the fourth implementable method, the candidate data of the longitudinal displacement of the beam body are processed simultaneously and at the same frequency to obtain the sampling data of the longitudinal displacement of the beam body, including:

[0016] The candidate data of longitudinal displacement of the beam body in the same acquisition period are sampled at the same frequency to obtain two sets of sampling data; each set of sampling data includes multiple beam body displacement sampling values ​​or multiple temperature sampling values;

[0017] Obtain the effective monitoring value of each group of sampling data;

[0018] The beam displacement sampling data and temperature sampling data are determined according to each effective monitoring value, and the beam displacement sampling data and temperature sampling data are used as the beam longitudinal displacement sampling data, and the collection time is used as the label of the beam longitudinal displacement sampling data.

[0019] In combination with the first implementable method, in the fifth implementable method, obtaining the longitudinal displacement characteristic vector of the bridge according to the beam longitudinal displacement sampling data includes:

[0020] Obtain the displacement change rate of the beam monitoring point based on the beam longitudinal displacement sampling data;

[0021] Obtain the displacement change acceleration of the beam monitoring point according to the longitudinal displacement change rate of the beam;

[0022] Obtain multiple longitudinal slip characteristics of the beam body based on the beam body longitudinal displacement sampling data, the displacement change rate of the beam body monitoring point, and the displacement change acceleration of the beam body monitoring point;

[0023] Acquire multiple beam rotation angle features based on beam longitudinal displacement sampling data, displacement change rate of beam monitoring points, and displacement change acceleration of beam monitoring points;

[0024] The expansion joint stuck characteristics are obtained based on the longitudinal displacement sampling data of the beam body;

[0025] The longitudinal displacement feature vector of the bridge is composed of multiple beam longitudinal sliding features, multiple beam rotation features, and expansion joint stuck features.

[0026] In combination with the fifth implementable method, in a sixth implementable method, multiple longitudinal slip characteristics of the beam are obtained based on the beam longitudinal displacement sampling data, the displacement change rate of the beam monitoring point, and the displacement change acceleration of the beam monitoring point, including:

[0027] According to the longitudinal displacement sampling data of the beam body, the first displacement average value of the left end of the bridge and the second displacement average value of the right end of the bridge are respectively obtained, and the difference between the first displacement average value and the second displacement average value is used as the longitudinal slip feature of the first beam body;

[0028] According to the displacement change rate of the beam monitoring point, the first rate average value of the left end of the bridge and the second rate average value of the right end of the bridge are obtained respectively, and the difference between the first rate average value and the second rate average value is used as the longitudinal slip feature of the second beam;

[0029] According to the acceleration of the displacement change of the beam monitoring point, the first acceleration average value of the left end of the bridge and the second acceleration average value of the right end of the bridge are obtained respectively, and the difference between the first acceleration average value and the second acceleration average value is used as the third beam longitudinal slip feature.

[0030] In combination with the fifth implementable method, in the seventh implementable method, multiple beam rotation angle features are obtained based on the beam longitudinal displacement sampling data, the displacement change rate of the beam monitoring point, and the displacement change acceleration of the beam monitoring point, including:

[0031] Obtain the first displacement absolute difference of the left end of the bridge and the second displacement absolute difference of the right end of the bridge based on the longitudinal displacement sampling data of the beam body, and use the average value of the first displacement absolute difference and the second displacement absolute difference as the first beam body rotation angle feature;

[0032] According to the displacement change rate of the beam monitoring point, the first rate absolute difference at the left end of the bridge and the second rate absolute difference at the right end of the bridge are obtained respectively, and the average value of the first rate absolute difference and the second rate absolute difference is used as the second beam rotation angle feature;

[0033] According to the acceleration of the displacement change of the beam monitoring point, the first absolute acceleration difference of the left end of the bridge and the second absolute acceleration difference of the right end of the bridge are obtained respectively, and the average value of the first absolute acceleration difference and the second absolute acceleration difference is used as the third beam angle feature.

[0034] In combination with the fifth implementable method, in the eighth implementable method, the expansion joint stuck feature is obtained based on the beam longitudinal displacement sampling data, including:

[0035] The second-order derivative of the displacement sampling data and temperature sampling data of each monitoring point is performed to obtain multiple alternative expansion joint stuck features;

[0036] The sum of the absolute values ​​of multiple candidate expansion joint stuck features is used as the expansion joint stuck feature.

[0037] Combined with the first implementable method, in the ninth implementable method, the bridge structure abnormality diagnosis network model includes an input layer, a membership function layer, a fuzzy reasoning layer, a normalization layer, and an output layer:

[0038] Input layer, inputs the bridge longitudinal displacement feature vector into the network model. The number of input layer nodes corresponds to the number of feature vectors.

[0039] The membership function layer maps multiple feature vectors to corresponding lexical variables through the cloud model normal distribution membership function to obtain multiple membership values. The number of lexical variables corresponds to the number of bridge structure states.

[0040] The fuzzy reasoning layer includes multiple neurons, each of which represents a fuzzy rule. It is used to match fuzzy rules to all membership values ​​and calculate the fitness value of each rule.

[0041] Normalization layer, normalizes the membership values ​​of all the previous fuzzy rules. The number of nodes in the normalization layer is the same as the number of nodes in the fuzzy reasoning layer.

[0042] The output layer uses the normalized probability values ​​to obtain the probability of occurrence of different bridge structure status types, and outputs the bridge structure status type corresponding to the maximum probability.

[0043] In combination with the ninth implementable method, in the tenth implementable method, the membership degree is calculated by the following method:

[0044]

[0045] in, represents the number of neurons, Expressed as The variable Membership function (MF), represents the height of the membership function, represents the width of the membership function, Expressed as feature vectors, All are positive integers.

[0046] It can be seen from the above technical solution that the beneficial technical effects of the present invention are as follows:

[0047] 1. The longitudinal displacement monitoring data of multiple beams are used to construct a feature vector reflecting the abnormal state of the beam structure. The abnormal state of the bridge structure is diagnosed online through the trained bridge structure diagnosis network model. This provides bridge maintenance personnel with a method to quickly determine the abnormal state of the beam structure, thereby improving the diagnosis efficiency and enabling timely risk warnings. It can also diagnose the bridge structure status online and realize real-time monitoring of abnormal bridge structures.

[0048] 2. Compared with the existing technology that directly uses monitoring data to obtain the bridge structure status, this solution preprocesses the beam longitudinal displacement monitoring data. After obtaining the beam longitudinal displacement sampling data, the bridge longitudinal displacement feature vector is obtained based on the beam longitudinal displacement sampling data. In this way, after preprocessing the monitoring data, the obtained feature vector can better represent the structural characteristics of the bridge. Then, the bridge structure status can be obtained based on the bridge longitudinal displacement feature vector, with higher accuracy, reduced reliance on manual labor, and improved intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0050] Figure 1 A schematic diagram of an online diagnosis method for bridge structure abnormalities based on beam longitudinal displacement monitoring provided by the present invention;

[0051] Figure 2 This is a schematic diagram of the layout of bridge monitoring points provided by the present invention. DETAILED DESCRIPTION

[0052] The following embodiments of the technical solution of the present invention will be described in detail with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention and are therefore only examples and are not intended to limit the scope of protection of the present invention.

[0053] It should be noted that, unless otherwise specified, the technical or scientific terms used in this application should have the common meanings understood by those skilled in the art to which the present invention belongs.

[0054] Combine Figure 1 As shown, this embodiment provides an online diagnosis method for bridge structure abnormalities based on beam longitudinal displacement monitoring, including:

[0055] Step S01, obtaining a plurality of beam longitudinal displacement monitoring data;

[0056] Step S02: pre-processing the plurality of beam longitudinal displacement monitoring data to obtain beam longitudinal displacement sampling data;

[0057] Step S03: Obtaining a bridge longitudinal displacement characteristic vector based on the beam longitudinal displacement sampling data;

[0058] Step S04: input the bridge longitudinal displacement feature vector into the bridge structure abnormality diagnosis network model to obtain the bridge structure status.

[0059] By using multiple beam longitudinal displacement monitoring data and temperature monitoring data, a feature vector reflecting the abnormal state of the beam structure is constructed. The abnormal state of the bridge structure is diagnosed online through the trained bridge structure diagnosis network model. The bridge structure diagnosis model can diagnose several types of conditions, including longitudinal slippage of the beam, beam rotation, and stuck expansion joints of the beam. It provides bridge maintenance personnel with a method to quickly determine the abnormal state of the beam structure, thereby improving diagnostic efficiency.

[0060] Optionally, multiple beam longitudinal displacement monitoring data are obtained, including: at least one temperature monitoring point and at least two displacement monitoring points are respectively set at the left and right beam ends of the bridge; temperature monitoring data is obtained in real time through temperature sensors arranged at each temperature monitoring point, and displacement monitoring data is obtained through displacement sensors arranged at each displacement monitoring point.

[0061] In some embodiments, temperature and humidity sensors are deployed at each temperature monitoring point to monitor the temperature and obtain temperature monitoring data.

[0062] In some embodiments, combined Figure 2 As shown, the circular points at the two ends of the bridge are monitoring points. Each end of the bridge is equipped with a temperature measurement point and two end displacement measurement points. The four end displacement measurement points are symmetrically distributed on both sides of the bridge deck: the first left displacement monitoring point s1, the second left displacement monitoring point s2, the first right displacement monitoring point s3, and the second right displacement monitoring point s4. Two temperature measurement points are distributed at the two ends of the bridge: the left temperature monitoring point THM1 and the right temperature monitoring point THM1. The temperature of the left temperature monitoring point THM1 is used as the common temperature for the first left displacement monitoring point s1 and the second left displacement monitoring point s2. The temperature of the right temperature monitoring point THM2 is used as the common temperature for the first right displacement monitoring point s3 and the second right displacement monitoring point s4.

[0063] Optionally, multiple beam longitudinal displacement monitoring data are preprocessed to obtain beam longitudinal displacement sampling data, including: using a pulse coupled neural network to adaptively reduce the noise of multiple beam longitudinal displacement monitoring data to obtain beam longitudinal displacement alternative data; and performing simultaneous and frequency processing on the beam longitudinal displacement alternative data to obtain beam longitudinal displacement sampling data.

[0064] Optionally, the longitudinal displacement monitoring data of the beam body is obtained by the formula Indicates; among them, is the monitoring data of the longitudinal displacement of the beam, is the temperature effect value, expressed as Gaussian noise, is the variation value of the bridge structure, Compared with the monitoring data, it is the high-frequency noise that affects the The random effect.

[0065] In some embodiments, since the longitudinal displacement monitoring data of the bridge beam is mainly affected by the environmental temperature and humidity, traffic load, random effects, etc., there are burrs and mutation points in the data performance. The pulse coupled neural network can effectively filter out such superimposed mixed noise. After the pulse coupled neural network (PCNN) is used to perform adaptive noise reduction processing on the longitudinal displacement and temperature monitoring data of the beam end, the influence of Gaussian noise and high-frequency noise is reduced, making the ;in, is the residual high-frequency noise. The residual high-frequency noise influence is the residual Gaussian noise and the residual high-frequency noise influence have little effect on the monitoring results and can be directly ignored. Therefore, the monitoring data only has the temperature influence and the change value of the bridge structure, that is, , greatly reducing the interference data in the monitoring data, thereby improving the accuracy of subsequent analysis.

[0066] Optionally, the candidate data of the longitudinal displacement of the beam body are processed simultaneously and at the same frequency to obtain the sampled data of the longitudinal displacement of the beam body, including: sampling the candidate data of the longitudinal displacement of the beam body in the same collection period at the same frequency to obtain two groups of sampled data; each group of sampled data includes multiple beam displacement sampled values ​​or multiple temperature sampled values;

[0067] Obtain the effective monitoring value of each group of sampling data; determine the beam displacement sampling data and temperature sampling data respectively according to each effective monitoring value, and use the beam displacement sampling data and temperature sampling data as the beam longitudinal displacement sampling data, and use the acquisition time as the label of the beam longitudinal displacement sampling data.

[0068] In some embodiments, since the bridge structure changes slowly, it can be assumed that the displacement data does not change much in a short period of time; according to the requirements of the "Technical Specifications for Highway Bridge Structure Monitoring JT / T 1037-2022", the structural response monitoring content sampling frequency table stipulates that the dynamic sampling frequency of the displacement sampling frequency is 20Hz and the static sampling frequency is 1Hz. The beam end displacement belongs to the displacement monitoring item, and static sampling is usually adopted with a sampling frequency of 1Hz. The temperature sampling frequency is ≤1 / 600Hz. Therefore, it is necessary to process the beam displacement monitoring data and the temperature monitoring data in the alternative data of the longitudinal displacement of the beam at the same time and frequency, so as to exclude other interfering data, obtain the beam displacement sampling data and temperature sampling data at the same acquisition time, and construct the correlation characteristics between the beam displacement sampling data and the temperature sampling data.

[0069] In some embodiments, within a preset collection time period, the beam displacement monitoring data is sampled at a frequency of 1 / 60 Hz, and the temperature monitoring data is sampled at a frequency of 1 / 60 Hz. The unified sampling time scale of the monitoring data is 10 minutes per time, and it can be approximately considered that the monitoring results have not changed.

[0070] Optionally, obtaining the effective monitoring value of each group of sampling data includes: sorting the sampling values ​​in each group of sampling data in descending order; removing the maximum sampling value and the minimum sampling value in each group of sampling data, leaving several sampling values; obtaining the average value of the remaining sampling values, and using the average value as the effective monitoring value.

[0071] In some embodiments, the preset acquisition time period is 10 minutes, and the beam displacement monitoring data is sampled strictly at 1 / 60 Hz. Due to the limitation of hardware and transmission conditions, it is not realistic to sample 10 times in 10 minutes. The final sampling data is mostly 8 to 9 times in 10 minutes. <10. The sample values ​​are sorted in descending order to obtain a sequence . Calculate using the following formula to obtain the effective monitoring value.

[0072] Optionally,

[0073] In the above formula, is the effective monitoring data of the first collection period, is the minimum sampling value, is the maximum sampling value, For the A sampling value.

[0074] Optionally, the minutes of the acquisition time are forward aligned, and the processed acquisition time is used as a label for the beam displacement sampling data and the temperature sampling data. In some embodiments, forward aligning the minutes includes processing numbers between 0 and 10 as 0, numbers between 10 and 20 as 10, numbers between 20 and 30 as 20, numbers between 30 and 40 as 30, numbers between 40 and 50 as 40, and numbers between 50 and 60 as 50. That is, if the acquisition time data is between 4:00 and 4:10, the acquisition time after forward alignment is 4:00; if the acquisition time data is between 4:10 and 4:20, the acquisition time after forward alignment is 4:10.

[0075] Optionally, the longitudinal displacement characteristic vector of the bridge is obtained according to the longitudinal displacement sampling data of the beam body, including: obtaining the displacement change rate of the beam monitoring point according to the longitudinal displacement sampling data of the beam body; obtaining the displacement change acceleration of the beam monitoring point according to the longitudinal displacement change rate of the beam body; obtaining multiple longitudinal slip characteristics of the beam body according to the longitudinal displacement sampling data of the beam body, the displacement change rate of the beam monitoring point, and the displacement change acceleration of the beam monitoring point; obtaining multiple beam rotation angle characteristics according to the longitudinal displacement sampling data of the beam body, the displacement change rate of the beam monitoring point, and the displacement change acceleration of the beam monitoring point; obtaining the expansion joint jamming feature according to the longitudinal displacement sampling data of the beam body; and combining the multiple longitudinal slip characteristics of the beam body, the multiple beam rotation angle characteristics, and the expansion joint jamming feature to form a bridge longitudinal displacement characteristic vector.

[0076] Optionally, obtaining the displacement change rate of the beam monitoring point according to the beam longitudinal displacement sampling data includes: calculating Obtain the displacement change rate of each monitoring point; where, is the displacement change rate of the qth monitoring point, is the displacement sampling data of the qth monitoring point, q={1, 2, 3, 4}, Represents derivation.

[0077] Optionally, obtaining the displacement change acceleration of the beam monitoring point according to the longitudinal displacement change rate of the beam includes: calculating Obtain the displacement change rate of each monitoring point; where, is the displacement change rate of the qth monitoring point, is the acceleration of the qth monitoring point.

[0078] Optionally, multiple longitudinal slip characteristics of the beam body are obtained based on the longitudinal displacement sampling data of the beam body, the displacement change rate of the beam body monitoring point, and the displacement change acceleration of the beam body monitoring point, including: obtaining the first displacement average value of the left end of the bridge and the second displacement average value of the right end of the bridge according to the longitudinal displacement sampling data of the beam body, and taking the difference between the first displacement average value and the second displacement average value as the first longitudinal slip characteristic of the beam body; obtaining the first velocity average value of the left end of the bridge and the second velocity average value of the right end of the bridge according to the displacement change rate of the beam body monitoring point, and taking the difference between the first velocity average value and the second velocity average value as the second longitudinal slip characteristic of the beam body; obtaining the first acceleration average value of the left end of the bridge and the second acceleration average value of the right end of the bridge according to the displacement change acceleration of the beam body monitoring point, and taking the difference between the first acceleration average value and the second acceleration average value as the third longitudinal slip characteristic of the beam body.

[0079] Optionally, the longitudinal slip characteristic of the first beam is obtained by the following formula:

[0080]

[0081] In the above formula, is the longitudinal sliding characteristic of the first beam, They are the displacement sampling data of the longitudinal displacement measuring points s1, s2, s3, and s4 of the beam body respectively.

[0082] Optionally, the longitudinal slip characteristic of the second beam is obtained by the following formula:

[0083]

[0084] In the above formula, is the longitudinal sliding characteristic of the second beam, They are the displacement change rates of the longitudinal displacement measuring points s1, s2, s3, and s4 of the beam body respectively.

[0085] Optionally, the longitudinal slip characteristics of the third beam are obtained by the following formula:

[0086]

[0087] In the above formula, is the longitudinal sliding characteristic of the third beam, They are the changing accelerations of the longitudinal displacement measuring points s1, s2, s3, and s4 of the beam body respectively.

[0088] Optionally, multiple beam body rotation angle features are obtained based on the beam body longitudinal displacement sampling data, the displacement change rate of the beam body monitoring point, and the displacement change acceleration of the beam body monitoring point, including: obtaining the first displacement absolute difference at the left end of the bridge and the second displacement absolute difference at the right end of the bridge based on the beam body longitudinal displacement sampling data, and taking the average of the first displacement absolute difference and the second displacement absolute difference as the first beam body rotation angle feature; obtaining the first velocity absolute difference at the left end of the bridge and the second velocity absolute difference at the right end of the bridge based on the displacement change rate of the beam body monitoring point, and taking the average of the first velocity absolute difference and the second velocity absolute difference as the second beam body rotation angle feature; obtaining the first acceleration absolute difference at the left end of the bridge and the second acceleration absolute difference at the right end of the bridge based on the displacement change acceleration of the beam body monitoring point, and taking the average of the first acceleration absolute difference and the second acceleration absolute difference as the third beam body rotation angle feature.

[0089] Optionally, the first beam angle feature is obtained by the following formula:

[0090]

[0091] In the above formula, is the first beam corner feature, They are the displacement sampling data of the longitudinal displacement measuring points s1, s2, s3, and s4 of the beam body respectively.

[0092] Optionally, the second beam angle feature is obtained by the following formula:

[0093]

[0094] In the above formula, is the second beam corner feature, They are the displacement change rates of the longitudinal displacement measuring points s1, s2, s3, and s4 of the beam body respectively.

[0095] Optionally, the third beam angle feature is obtained by the following formula:

[0096]

[0097] In the above formula, is the corner feature of the third beam, They are the changing accelerations of the longitudinal displacement measuring points s1, s2, s3, and s4 of the beam body respectively.

[0098] Optionally, the expansion joint stuck feature is obtained based on the longitudinal displacement sampling data of the beam body, including: performing second-order derivative of the displacement sampling data and temperature sampling data of each monitoring point to obtain multiple alternative expansion joint stuck features; and taking the sum of the absolute values ​​of the multiple alternative expansion joint stuck features as the expansion joint stuck feature.

[0099] Optionally, by calculating Obtain the alternative expansion joint stuck characteristics, where is the displacement sampling data, is the temperature sampling data.

[0100] Optionally, the expansion joint stuck characteristic is calculated using the following formula:

[0101]

[0102] In the above formula, This is the stuck feature of the expansion joint. The first alternative expansion joint stuck feature, The second alternative expansion joint stuck feature, The third alternative expansion joint stuck feature, This is the fourth alternative expansion joint stuck feature.

[0103] In some embodiments, the main types of beam structure failures are: longitudinal slippage of the beam, beam rotation, and expansion joint jamming of the beam. The first beam longitudinal slip feature, the second beam longitudinal slip feature, and the third beam longitudinal slip feature are used together to evaluate whether the bridge has longitudinal slippage, the first beam rotation feature, the second beam rotation feature, and the third beam rotation feature are used together to evaluate whether the bridge beam has rotated, and the expansion joint jamming feature is used to evaluate whether the bridge has expansion joint jamming. The first beam longitudinal slip feature, the second beam longitudinal slip feature, the third beam longitudinal slip feature, the first beam rotation feature, the second beam rotation feature, the third beam rotation feature, and the expansion joint jamming feature are combined to form the longitudinal displacement feature vector of the bridge, that is, In this way, since the relationship between features and bridge structural states is not a simple one-to-one relationship, a bridge structural state type may often be reflected in multiple features. Therefore, using multiple features to evaluate the bridge structural state type improves accuracy.

[0104] Optionally, before inputting the bridge longitudinal displacement feature vector into the bridge structure abnormality diagnosis network model, the bridge structure abnormality diagnosis network model is constructed by the following method, including:

[0105] Step S11: Acquire longitudinal displacement monitoring data of the bridge beam under various structural states, where the monitoring data types include displacement monitoring data and temperature monitoring data under structural states such as normal bridge, longitudinal slippage of the bridge, torsion of the bridge beam, and stuck expansion joint of the bridge;

[0106] Step S12: Construct monitoring data under different structural states based on the acquired monitoring data, use the monitoring data as samples, and use the structural states corresponding to the monitoring data as labels for the monitoring data. The number of types under each structural state is ≥10,000, and the monitoring data is divided into an alternative training set and an alternative test set in a ratio of 9:1.

[0107] Step S13: performing adaptive noise reduction and simultaneous frequency co-processing on the sample data of the candidate training set to obtain a training set, and performing adaptive noise reduction and simultaneous frequency co-processing on the sample data of the candidate test set to obtain a test set;

[0108] Step S14: obtaining a bridge longitudinal displacement feature vector for training based on the training set, and obtaining a bridge longitudinal displacement feature vector for testing based on the test set;

[0109] Step S15: sending the bridge longitudinal displacement feature vector used for training into the fuzzy neural network model training to obtain an alternative bridge structure abnormality diagnosis network model;

[0110] Step S16: Input the longitudinal displacement feature vector of the bridge used for testing into the alternative bridge structure abnormality diagnosis network model to obtain the test accuracy of various bridge structure states; if the test accuracy does not reach 99%, continue to add samples corresponding to the bridge structure state, and return to step S15 for training until the model training requirements are met. If the test accuracy reaches 99%, the alternative bridge structure abnormality diagnosis network model is used as the final bridge structure abnormality diagnosis network model.

[0111] Optionally, the bridge structure abnormality diagnosis network model includes an input layer, a membership function layer, a fuzzy reasoning layer, a normalization layer, and an output layer: the input layer inputs the bridge longitudinal displacement feature vector into the network model, and the number of input layer nodes corresponds to the number of feature vectors; the membership function layer maps multiple feature vectors to corresponding lexical variables through the cloud model normal distribution membership function to obtain multiple membership values, and the number of lexical variables corresponds to the number of bridge structure states; the fuzzy reasoning layer includes multiple neurons, each neuron represents a fuzzy rule, which is used to match fuzzy rules to all membership values ​​and calculate the fitness value of each rule; the normalization layer normalizes the membership values ​​of all previous fuzzy rules, and the number of nodes in the normalization layer is the same as the number of nodes in the fuzzy reasoning layer; the output layer uses the probability value obtained by normalization to obtain the probability of occurrence of different bridge structure state types, and outputs the bridge structure state type corresponding to the maximum probability.

[0112] Optionally, membership is calculated as follows:

[0113]

[0114] in, represents the number of neurons, Expressed as The variable Membership function (MF), represents the height of the membership function, represents the width of the membership function, Expressed as feature vectors, All are positive integers.

[0115] Optionally, the fuzzy inference layer is implemented by formula Calculate the fitness value of each rule; where, Representatives The membership degree of the fuzzy rules.

[0116] Optionally, the normalization layer is implemented by the formula Normalize the membership values ​​of all the previous fuzzy rules; among them, Corresponding to each The normalized credibility of the fuzzy rules.

[0117] Optionally, the output layer is given by the formula Obtain the probability of each bridge structure status type; where, is the weight, and the back propagation algorithm is used to update it.

[0118] In some embodiments, the input layer converts the corresponding bridge structure feature vector The network model is passed in, and the number of nodes is 7, which corresponds to the number of eigenvectors. The membership function layer will Features are mapped to On the basis of lexical variables, 4 lexical variables correspond to 4 bridge structural states, and we get Group membership value, each group has Fuzzy subsets, each fuzzy subset corresponds to 1 membership degree. Each neuron in the fuzzy reasoning layer represents a fuzzy rule, which is used to match the fuzzy rules. The fuzzy reasoning layer is based on The fitness value of each rule is calculated based on the group membership value. The normalization layer normalizes the membership values ​​of all the previous fuzzy rules to obtain the normalized credibility of each fuzzy rule, that is, the probability value. The output layer uses the probability value obtained by normalization to obtain the cumulative sum of the probabilities of different fuzzy rules, that is, the results of different structural state types, and outputs the bridge structure state type corresponding to the maximum probability. In this way, the monitoring data of the longitudinal displacement of multiple beams and the temperature monitoring data are used to construct a feature vector reflecting the abnormal state of the beam structure. The abnormal state of the bridge structure is diagnosed online through the trained bridge structure diagnosis network model. The bridge structure diagnosis model can diagnose several types including longitudinal slip of the beam, beam rotation, and stuck expansion joints of the beam, providing bridge maintenance personnel with a method to quickly determine the abnormal state of the beam structure.

[0119] 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 make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions 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 description of the present invention.

Claims

1. An online diagnosis method for bridge structure abnormality based on beam longitudinal displacement monitoring, characterized in that: include: Acquire multiple beam longitudinal displacement monitoring data in real time; Preprocessing the plurality of beam longitudinal displacement monitoring data to obtain beam longitudinal displacement sampling data; Obtaining a bridge longitudinal displacement characteristic vector according to the beam longitudinal displacement sampling data; Inputting the bridge longitudinal displacement characteristic vector into a bridge structure abnormality diagnosis network model to obtain the bridge structure state; Obtaining a bridge longitudinal displacement characteristic vector according to the beam longitudinal displacement sampling data includes: Obtaining the displacement change rate of the beam monitoring point according to the beam longitudinal displacement sampling data; Obtaining the displacement change acceleration of the beam monitoring point according to the longitudinal displacement change rate of the beam; Acquire multiple longitudinal slip characteristics of the beam body according to the beam body longitudinal displacement sampling data, the displacement change rate of the beam body monitoring point, and the displacement change acceleration of the beam body monitoring point; Acquire multiple beam body rotation angle features based on the beam body longitudinal displacement sampling data, the displacement change rate of the beam body monitoring point, and the displacement change acceleration of the beam body monitoring point; Acquiring expansion joint stuck characteristics based on the beam longitudinal displacement sampling data; The longitudinal displacement feature vector of the bridge is composed of multiple beam longitudinal sliding features, multiple beam rotation features, and expansion joint stuck features.

2. The method according to claim 1, characterized in that Acquire multiple beam longitudinal displacement monitoring data, including: At least one temperature monitoring point and at least two displacement monitoring points are respectively provided at the left and right beam ends of the bridge; temperature monitoring data is obtained in real time through the temperature sensors arranged at each temperature monitoring point, and displacement monitoring data is obtained through the displacement sensors arranged at each displacement monitoring point.

3. The method according to claim 2, characterized in that Preprocessing the plurality of beam longitudinal displacement monitoring data to obtain beam longitudinal displacement sampling data includes: Adopting a pulse coupled neural network to perform adaptive noise reduction processing on the plurality of beam longitudinal displacement monitoring data to obtain candidate beam longitudinal displacement data; The candidate data of the longitudinal displacement of the beam body are processed simultaneously and at the same frequency to obtain sampling data of the longitudinal displacement of the beam body.

4. The method according to claim 3, characterized in that The candidate data of the longitudinal displacement of the beam body are processed simultaneously and at the same frequency to obtain the sampling data of the longitudinal displacement of the beam body, including: The candidate data of longitudinal displacement of the beam body in the same acquisition period are sampled at the same frequency to obtain two sets of sampling data; each set of sampling data includes multiple beam body displacement sampling values ​​or multiple temperature sampling values; Obtain the effective monitoring value of each group of sampling data; The beam displacement sampling data and the temperature sampling data are respectively determined according to each effective monitoring value, and the beam displacement sampling data and the temperature sampling data are used as the beam longitudinal displacement sampling data, and the acquisition time is used as the label of the beam longitudinal displacement sampling data.

5. The method according to claim 1, characterized in that A plurality of longitudinal slip characteristics of the beam body are obtained according to the longitudinal displacement sampling data of the beam body, the displacement change rate of the beam body monitoring point, and the displacement change acceleration of the beam body monitoring point, including: According to the longitudinal displacement sampling data of the beam body, the first displacement average value of the left end of the bridge and the second displacement average value of the right end of the bridge are respectively obtained, and the difference between the first displacement average value and the second displacement average value is used as the longitudinal slip feature of the first beam body; According to the displacement change rate of the beam monitoring point, the first rate average value of the left end of the bridge and the second rate average value of the right end of the bridge are obtained respectively, and the difference between the first rate average value and the second rate average value is used as the longitudinal slip feature of the second beam; According to the acceleration of the displacement change of the beam monitoring point, the first acceleration average value of the left end of the bridge and the second acceleration average value of the right end of the bridge are obtained respectively, and the difference between the first acceleration average value and the second acceleration average value is used as the third beam longitudinal slip feature.

6. The method according to claim 1, characterized in that A plurality of beam rotation angle features are obtained based on the beam longitudinal displacement sampling data, the displacement change rate of the beam monitoring point, and the displacement change acceleration of the beam monitoring point, including: Obtaining a first displacement absolute difference at the left end of the bridge and a second displacement absolute difference at the right end of the bridge according to the beam longitudinal displacement sampling data, and taking an average value of the first displacement absolute difference and the second displacement absolute difference as a first beam rotation angle feature; According to the displacement change rate of the beam monitoring point, a first rate absolute difference at the left end of the bridge and a second rate absolute difference at the right end of the bridge are respectively obtained, and an average value of the first rate absolute difference and the second rate absolute difference is used as the second beam rotation angle feature; According to the displacement change acceleration of the beam monitoring point, a first acceleration absolute difference at the left end of the bridge and a second acceleration absolute difference at the right end of the bridge are respectively obtained, and the average value of the first acceleration absolute difference and the second acceleration absolute difference is used as the third beam angle feature.

7. The method according to claim 1, characterized in that Obtaining the expansion joint stuck feature according to the beam longitudinal displacement sampling data includes: The second-order derivative of the displacement sampling data and temperature sampling data of each monitoring point is performed to obtain multiple alternative expansion joint stuck features; The sum of the absolute values ​​of multiple candidate expansion joint stuck features is used as the expansion joint stuck feature.

8. The method according to claim 1, characterized in that The bridge structure abnormality diagnosis network model includes input layer, membership function layer, fuzzy reasoning layer, normalization layer, and output layer: Input layer, inputs the bridge longitudinal displacement feature vector into the network model. The number of input layer nodes corresponds to the number of feature vectors. The membership function layer maps multiple feature vectors to corresponding lexical variables through the cloud model normal distribution membership function to obtain multiple membership values. The number of lexical variables corresponds to the number of bridge structure states. The fuzzy reasoning layer includes multiple neurons, each of which represents a fuzzy rule. It is used to match fuzzy rules to all membership values ​​and calculate the fitness value of each rule. Normalization layer, normalizes the membership values ​​of all the previous fuzzy rules. The number of nodes in the normalization layer is the same as the number of nodes in the fuzzy reasoning layer. The output layer uses the normalized probability values ​​to obtain the probability of occurrence of different bridge structure status types, and outputs the bridge structure status type corresponding to the maximum probability.

9. The method according to claim 8, characterized in that Membership is calculated as follows: in, represents the number of neurons, Expressed as The variable Membership function (MF), represents the height of the membership function, represents the width of the membership function, Expressed as feature vectors, All are positive integers.

Images