Cable force data filling method and storage medium for cable force sensor failure

Abstract

This invention relates to a method and storage medium for filling in cable force data caused by cable force sensor failure. The data filling method includes: selecting the cable where the sensor failed as the target cable and neighboring cables that did not fail as input cables; extracting all cable force data before and after the failure and using them as training and testing sets respectively; preprocessing the cable force data, including removing intermittent breakpoints, downsampling, and reconstructing the data; establishing observation functions for the input and target cable forces based on the principle of extended dynamic mode decomposition and calculating prediction operators; using the prediction operators to predict the target cable force in the testing set and restoring the sampling frequency to fill in the missing data. Compared with the prior art, this invention can effectively and quickly fill in the missing cable force data caused by sensor failure, improving the integrity and reliability of the data.

Description

Technical Field

[0001] This invention relates to the field of bridge cable detection and management technology, and in particular to a method and storage medium for filling cable force data in response to cable force sensor failure. Background Technology

[0002] Cable stays are critical force-transmitting components of cable-stayed bridges. They endure repeated or extreme loads over long periods, leading to complex variations in cable forces and a tendency to accumulate fatigue damage, thus affecting the overall performance of the bridge structure. Cable force monitoring helps detect cable deterioration or damage early, preventing potential safety issues in the bridge structure and is of great significance for assessing the cable condition of cable-stayed bridges. Cable-stayed bridges typically contain dozens or even hundreds of cables or suspenders. Taking a cable-stayed bridge as an example, full-bridge cable force monitoring relies on cable force sensors distributed on each cable. However, due to the durability of cable sensors and the harsh working environment, cable force sensors are prone to malfunction or failure, resulting in data loss. In particular, cable force sensors such as anchor gauges must be installed during bridge construction and cannot be replaced later, making data loss unavoidable.

[0003] Currently, preliminary research has been conducted on prediction and imputation techniques for missing cable force monitoring data, such as polynomial interpolation and machine learning for cable force prediction. However, these methods rely on meticulous data screening and model training, and suffer from high errors and limited applicability. On the other hand, the sheer volume of high-frequency cable force data increases the computational complexity of cable force imputation. Therefore, establishing accurate and efficient cable force data imputation techniques for stay cables with sensor malfunctions is crucial for assessing cable condition and ensuring bridge safety. Summary of the Invention

[0004] The purpose of this invention is to address the above-mentioned problems by proposing a cable force data filling method based on extended dynamic mode decomposition, which provides long-term data support for the practical application of bridge cable force monitoring.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] This invention provides a method for filling in cable force data in the event of a cable force sensor malfunction, comprising the following steps:

[0007] S1: Select the faulty cable as the target cable and the neighboring non-faulty cables as the input cables. Divide the cable force dataset before and after the fault into training set and test set.

[0008] S2: Preprocess the cable force data in the cable force dataset before and after the fault, including deleting breakpoints, downsampling, and reconstructing the data;

[0009] S3: Based on the cable force dataset before and after the fault processed by S2, the input cable force observation function and the target cable force observation function are established using the principle of extended dynamic mode decomposition, and the prediction operator K is calculated through the relationship between the two.

[0010] S4: Based on the prediction operator K obtained in S3, calculate the target cable force observation function, and combine the predicted cable forces in the target cable force observation function according to the original sampling order to restore the high-frequency cable forces and complete the data filling.

[0011] Furthermore, S1 specifically includes the following steps:

[0012] S1-1: Select the cable that has lost its cable force data due to sensor failure as the target cable, and at the same time determine the multiple adjacent cables that have not experienced sensor failure as input cables;

[0013] S1-2: Extract the cable force data of all cables before and after the target cable sensor malfunctions. Record all cable force data before the sensor malfunction as the training set and all cable force data after the sensor malfunction as the test set.

[0014] Furthermore, S2 specifically includes the following steps:

[0015] Remove intermittent data breakpoints: Process the cable tension data of each cable, remove intermittent data breakpoints to ensure that there are no null values ​​in the data, and ensure the integrity and availability of the data;

[0016] Downsampling operation: Downsampling is performed on the training and test sets of the Soli data to divide the original high-frequency data into multiple low-frequency training and test sets;

[0017] Data Reorganization: Select any low-frequency training set and recombine the one-dimensional data of the cable force of all input cables into two-dimensional data. Specify that each column is the cable force data corresponding to each cable. At the same time, clarify the representation of the target cable force data, as well as the representation of the input and target data in the training and test sets.

[0018] Furthermore, after the downsampling operation, the cable frequency does not exceed 0.1Hz;

[0019] The data reconstruction process specifically includes the following steps:

[0020] Select any low-frequency training set and reorganize the one-dimensional data of cable force of all input cables into two-dimensional data, where each column represents the cable force data of each cable, denoted as X = [x1, x2, x3, ..., x...]. n ], where n is the number of input cables, the target cable force data is denoted as Y = y, and the input and target data in the training set are X and X, respectively. train and Y trainThe input and target data in the test set are X and X, respectively. test and Y test .

[0021] Furthermore, S3 specifically includes the following steps:

[0022] S3-1: Based on the principle of extended dynamic mode decomposition, using the training set data processed in S2, establish the observation function Ψ(X) of the input cable force. train )=[x1,x2,...,x n ;x1′,x2′,...,x′ n ;x1″,x′2′,...,x n "] and the observation function Ψ(Y) of the target cable force train = [y; y′; y″];

[0023] S3-2: Based on the observation functions of the input cable force and the target cable force, establish the mapping relationship between them using the formula K = Ψ(X train ) -1 Ψ(Y train ) Calculate the prediction operator K.

[0024] Furthermore, in S3-2, the observation function is defined by a combination of cable force value, first derivative of cable force, and second derivative of cable force, and is used to extract the spatiotemporal correlation characteristics of the cable force system.

[0025] Furthermore, in S3-2, Ψ(X) train ) -1 For Ψ(X) train The pseudo-inverse matrix of ).

[0026] Furthermore, S4 specifically includes the following steps:

[0027] S4-1: Calculate the target cable force observation function: Based on the prediction operator K obtained in S3, calculate the target cable force observation function for each test set using the formula;

[0028] S4-2: High-frequency cable force restoration to complete the filling: In the target cable force observation function, for the predicted cable force of each test set, all test set cable forces are combined in the original sampling order to restore the high-frequency cable force, thereby realizing the filling of the target cable force data.

[0029] Further, in S4-1, the target cable force observation function is a combination of the corresponding cable force, the first derivative of the cable force, and the second derivative of the cable force, expressed as follows:

[0030] In S4-1, based on the prediction operator K, the observation function of the target cable force for each test set is calculated, and the calculation formula is as follows:

[0031] Ψ(Y test )=KΨ(X test );

[0032] Observation function of target cable force The first item The predicted cable force for each test set is used to combine the cable forces of all test sets in the original sampling order to reconstruct the high-frequency cable force.

[0033] A second aspect of the present invention provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a cable force data filling method for cable force sensor failure as described above.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] 1) This invention utilizes the spatiotemporal correlation of cable force to accurately fill in missing data in the event of partial sensor failure, thereby improving the continuity of cable force data monitoring;

[0036] 2) This invention reduces the computational burden and improves the efficiency of cable force monitoring in large-scale cable networks by downsampling and dynamic mode decomposition;

[0037] 3) The method of the present invention has high accuracy in filling cable force data under sensor failure conditions, and can significantly improve the reliability of bridge cable monitoring data. Attached Figure Description

[0038] Figure 1 This is a flowchart of filling in the cable force data for a cable-stayed bridge in one embodiment.

[0039] Figure 2 This is a time history diagram of the original cable force in one embodiment.

[0040] Figure 3 This is a diagram illustrating the effect of intermittent breakpoint handling on the Soli dataset in one embodiment.

[0041] Figure 4 This is a diagram illustrating the effect of filling the cable tension in a faulty cable in one embodiment. Detailed Implementation

[0042] Overall, this invention discloses a method for filling in cable force data after a cable-stayed bridge cable force sensor failure. Specifically, it includes: selecting the cable where the sensor failed as the target cable and adjacent unfailed cables as input cables; extracting all cable force data before and after the failure, and using them as training and testing sets respectively; preprocessing the cable force data, including removing intermittent breakpoints, downsampling, and data reconstruction; establishing observation functions for the input and target cable forces based on the extended dynamic mode decomposition principle, and calculating prediction operators; using the prediction operators to predict the target cable force in the testing set, restoring the sampling frequency, and filling in missing data. This method can effectively and quickly fill in the missing cable force data caused by sensor failure, improving the integrity and reliability of the data.

[0043] The method for filling in cable tension data in response to cable tension sensor failure includes the following steps:

[0044] Step 1: Select the cable whose force data was lost due to sensor failure as the target cable, and the multiple adjacent cables that did not experience sensor failure as input cables. Extract the force data of all cables before and after the sensor failure of the target cable. Record all force data before the sensor failure as the training set, and record all force data after the sensor failure as the test set.

[0045] Step 2: Preprocess the cable tension data for each cable, including:

[0046] Remove intermittent data breaks in the soliput data to ensure that there are no null values ​​in the data;

[0047] The training and test sets of Soli data are downsampled, and the original high-frequency data is divided into multiple low-frequency training and test sets to reduce the amount of computation and improve the efficiency of the algorithm.

[0048] Select any low-frequency training set and reorganize the one-dimensional data of cable force of all input cables into two-dimensional data, where each column represents the cable force data of each cable, which can be represented as X = [x1, x2, x3, ..., x n ], where n is the number of input cables; the cable force data of the target cable is denoted as Y = y; the input data and target data in the training set are X and X, respectively. train and Y train The input and target data in the test set are X and X, respectively. test and Y test ;

[0049] Step 3: Based on the principle of extended dynamic mode decomposition, calculate the prediction operator K of the input cable data acting on the target cable data, including:

[0050] Establish observation functions for the input cable force and the target cable force. The observation functions are combinations of the corresponding cable force, the first derivative of the cable force, and the second derivative of the cable force, denoted as Ψ(X). train )=[x1,x2,...,x n ;x1′,x2′,...,x′ n ;x1″,x′2′,...,x n "] and Ψ(Y train = [y; y′; y″];

[0051] Establish the mapping relationship between the input cable force and the target cable force, and calculate the prediction operator K. The calculation formula is as follows:

[0052] K = Ψ(X) train ) -1 Ψ(Y train )

[0053] Step 4: Based on the prediction operator K, predict and fill in the target cable force in the test set, including:

[0054] Establish an observation function for the input cable force in each low-frequency test set. The observation function is a combination of the corresponding cable force, the first derivative of the cable force, and the second derivative of the cable force, expressed as follows:

[0055] Based on the prediction operator K, the observation function of the target cable force for each test set is calculated using the following formula:

[0056] Ψ(Y test )=KΨ(X test )

[0057] Calculated The first item For the predicted cable force of each test set, the predicted cable forces of all test sets are combined in the original sampling order to restore the high-frequency cable force.

[0058] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Component models, material names, connection structures, control methods, algorithms, and other features not explicitly described in this technical solution are considered common technical features disclosed in the prior art.

[0059] Example 1

[0060] The technical approach of this embodiment is as follows: Figure 1As shown, a data-driven approach treats the cable forces of multiple cables as a system with spatiotemporal correlation. First, a cable with a sensor malfunction is selected as the target cable, and multiple cables on the same cable plane and connected to the same tower as the target cable are used as inputs. The cable force data before and after the sensor malfunction are divided into training and test sets. Then, all data are downsampled and divided into multiple training and test sets. Second, an observation function is constructed based on the input and target cable force training sets, including the cable force and its first and second derivatives. Using the extended dynamic mode decomposition method, the linear mapping relationship between the input and target cables is calculated to obtain the prediction operator. Finally, the prediction operator is applied to the input cable force of each test set to obtain the target cable force for each test set, and the cable force data of the target cable with the sensor malfunction at the original frequency is reconstructed in its original order.

[0061] In this embodiment, simulating a sensor failure in the cable at a certain time, the method for filling in the cable force data of the cable-stayed bridge includes the following steps:

[0062] Step 1: Select the cable whose sensor malfunction caused the loss of cable force data as the target cable, and its eight neighboring cables that did not experience sensor malfunction as input cables. Extract the cable force data of all cables 2 hours before and 1 hour after the target cable's sensor malfunction, such as... Figure 2 As shown;

[0063] Step 2: Preprocess the cable tension data for each cable, including:

[0064] Delete intermittent data breakpoints in the cable force data, such as... Figure 3 As shown;

[0065] The cable force data with an original sampling frequency of 1Hz was downsampled to 0.1Hz. The high-frequency data was divided into 10 low-frequency datasets, of which the cable force data before the failure was the training set and the cable force data after the failure was the test set.

[0066] Select the first low-frequency training set and reorganize the one-dimensional data of the cable force of all input cables into two-dimensional data. Each column represents the cable force data of each cable, which can be represented as X = [x1, x2, x3, ..., x8]; the cable force data of the target cable is denoted as Y = y; the input data and target data in the training set are X and Y, respectively. train and Y train The input and target data in the test set are X and X, respectively. test and Y test ;

[0067] Step 3: Based on the principle of extended dynamic mode decomposition, calculate the prediction operator K of the input cable data acting on the target cable data, including:

[0068] Establish observation functions for the input cable force and the target cable force. The observation functions are combinations of the corresponding cable force, the first derivative of the cable force, and the second derivative of the cable force, denoted as Ψ(X). train )=[x1,x2,...,x 10 ;x1′,x′2,...,x1′0;x1″,x′2′,...,x1″0] and Ψ(Y train = [y; y′; y″];

[0069] Establish the mapping relationship between the input cable force and the target cable force, and calculate the prediction operator K. The calculation formula is as follows:

[0070] K = Ψ(X) train ) -1 Ψ(Y train )

[0071] Step 4: Based on the prediction operator K, predict and fill in the target cable force in the test set, including:

[0072] Establish an observation function for the input cable force in each low-frequency test set. The observation function is a combination of the corresponding cable force, the first derivative of the cable force, and the second derivative of the cable force, expressed as follows:

[0073] Based on the prediction operator K, the observation function of the target cable force for each test set is calculated using the following formula:

[0074] Ψ(Y test )=KΨ(X test )

[0075] Calculated The first item For each test set, the predicted target cable force is calculated by combining the predicted values ​​of all test sets in the original sampling order to reconstruct the target cable force y% with a sampling frequency of 1Hz, expressed as:

[0076]

[0077] Figure 4 The results show the comparison between the cable tension filling value and the actual value when simulating sensor failure. The average error is less than 0.1%, and the time required to fill one hour of data is less than 0.001 seconds.

[0078] This embodiment also provides a computer-readable storage medium storing computer instructions for instructing a computer to execute the aforementioned method for filling in cable force data in response to cable force sensor malfunctions. The storage medium can be an electronic medium, magnetic medium, optical medium, electromagnetic medium, infrared medium, or a semiconductor system or propagation medium. The storage medium may also include semiconductor or solid-state memory, magnetic tape, removable computer disk, random access memory (RAM), read-only memory (ROM), hard disk, and optical disc. Optical discs may include optical disc-read-only memory (CD-ROM), optical disc-read-write (CD-RW), and DVD.

[0079] The technical means disclosed in this invention are not limited to those disclosed in the above embodiments, but also include technical solutions composed of any combination of the above technical features.

[0080] The above description of the embodiments is provided to enable those skilled in the art to understand and use the invention. It will be apparent to those skilled in the art that various modifications can be made to these embodiments, and the general principles described herein can be applied to other embodiments without inventive effort. Therefore, the present invention is not limited to the above embodiments, and any improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the invention should be within the protection scope of the present invention.

Claims

1. A method for filling in cable force data in response to cable force sensor failure, characterized in that, Includes the following steps: S1: Select the faulty cable as the target cable and the neighboring non-faulty cables as the input cables. Divide the cable force dataset before and after the fault into training set and test set. S2: Preprocess the cable force data in the cable force dataset before and after the fault, including deleting breakpoints, downsampling, and reconstructing the data; S3: Based on the cable force dataset before and after the fault processed by S2, the input cable force observation function and the target cable force observation function are established using the principle of extended dynamic mode decomposition. The prediction operator is calculated through the relationship between the two. K Specifically, it includes: Establish observation functions for the input cable force and the target cable force. These observation functions are combinations of the corresponding cable force, the first derivative of the cable force, and the second derivative of the cable force, expressed as follows: and ; Establish the mapping relationship between input cable force and target cable force, and calculate the prediction operator. K The calculation formula is: ； S4: Prediction operator obtained based on S3 K The target cable force observation function is calculated. Based on the predicted cable forces in the target cable force observation function, the high-frequency cable forces are reconstructed according to the original sampling order to complete data filling. Specifically, this includes: Establish an observation function for the input cable force in each low-frequency test set. The observation function is a combination of the corresponding cable force, the first derivative of the cable force, and the second derivative of the cable force, expressed as follows: ; Based on the prediction operator K, the observation function of the target cable force for each test set is calculated using the following formula: Calculated The first item For the predicted cable force of each test set, the predicted cable forces of all test sets are combined in the original sampling order to restore the high-frequency cable force.

2. The method for filling in cable force data in case of cable force sensor failure according to claim 1, characterized in that, S1 specifically includes the following steps: S1-1: Select the cable that has lost its cable force data due to sensor failure as the target cable, and at the same time determine the multiple adjacent cables that have not experienced sensor failure as input cables; S1-2: Extract the cable force data of all cables before and after the target cable sensor malfunctions. Record all cable force data before the sensor malfunction as the training set and all cable force data after the sensor malfunction as the test set.

3. The method for filling in cable force data in case of cable force sensor failure according to claim 1, characterized in that, S2 specifically includes the following steps: Remove intermittent data breakpoints: Process the cable tension data of each cable, remove intermittent data breakpoints to ensure that there are no null values ​​in the data, and ensure the integrity and availability of the data; Downsampling operation: Downsampling is performed on the training and test sets of the Soli data to divide the original high-frequency data into multiple low-frequency training and test sets; Data Reorganization: Select any low-frequency training set and recombine the one-dimensional data of the cable force of all input cables into two-dimensional data. Specify that each column is the cable force data corresponding to each cable. At the same time, clarify the representation of the target cable force data, as well as the representation of the input and target data in the training and test sets.

4. The method for filling in cable force data in case of cable force sensor failure according to claim 3, characterized in that, After the downsampling operation, the cable frequency does not exceed 0.1Hz; The data reconstruction process specifically includes the following steps: Select any low-frequency training set and reorganize the one-dimensional data of cable force of all input cables into two-dimensional data, where each column represents the cable force data of each cable, as shown in Figure 1. , n To input the number of cables, the cable tension data of the target cables is denoted as... Y=y The input data and target data in the training set are respectively X train and Y train The input data and target data in the test set are respectively X test and Y test .

5. A storage medium containing computer-executable instructions, characterized in that, When executed by a computer processor, the storage medium of the computer-executable instructions is used to perform a cable force data filling method for cable force sensor failure as described in any one of claims 1 to 4.

Images