Internet of Things-based online health monitoring system for cement poles

By constructing bending strain vectors and analyzing strain transmission differences through an Internet of Things (IoT) system, the problem of separating load changes from state changes in online monitoring of cement poles was solved, and a stable health status assessment was achieved.

CN123065698APending Publication Date: 2026-11-10CHANGZHOU INST OF TECH +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611585984.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-14
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing online monitoring systems for cement poles struggle to accurately distinguish between load changes and changes in the pole's own condition. Abnormal responses caused by local stiffness changes are difficult to separate from overall load responses, leading to unstable health status assessments.

Method used

By using an IoT-based pole strain acquisition module, strain transmission analysis module, load influence elimination module, and stiffness degradation analysis module, a bending strain vector is constructed, a load variation basis is extracted, strain transmission difference characteristics are analyzed, pole stiffness degradation paths are generated, and a healthy state distribution of cement poles is formed.

Benefits of technology

It can accurately identify changes in the transmission relationship between adjacent monitoring locations, reduce the impact of common load fluctuations, distinguish between local and overall stiffness changes, and provide a stable health status assessment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123065698A_ABST
    Figure CN123065698A_ABST
Patent Text Reader

Abstract

This invention belongs to the field of structural performance testing technology. It discloses an online monitoring system for the health status of cement poles based on the Internet of Things (IoT). The system includes: collecting surface longitudinal strain values ​​at corresponding monitoring locations and transmitting them to an online analysis terminal; sequentially arranging and associating the surface longitudinal strain values ​​within the same load cycle with adjacent locations to obtain a pole height strain transmission sequence; generating strain transmission difference characteristics by analyzing the correlation between the bending strain characteristics in the pole height strain transmission sequence and the transmission relationship between common load variations and adjacent monitoring locations; obtaining the pole stiffness degradation path by analyzing the correlation between the strain transmission difference characteristics and the transmission process along the pole height direction during loading and unloading; and continuously matching and fitting the pole stiffness degradation paths within adjacent load cycles to obtain the health status distribution of the cement pole. This invention can identify continuous stiffness degradation across locations and quantify local health status.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of structural performance testing technology, and more specifically, to an online health status monitoring system for cement poles based on the Internet of Things. Background Technology

[0002] Concrete poles are commonly used supporting components in power distribution and communication lines. Operating outdoors for extended periods, they are susceptible to wind loads, conductor pull, temperature variations, and external impacts. With the development of IoT communication and online sensing technologies, distributed monitoring devices can continuously acquire the structural responses of concrete poles and transmit the data to a remote platform. This data helps determine the operational status of the concrete poles and provides a reference for inspection and maintenance.

[0003] In the existing online monitoring of cement poles, the structural response at different monitoring locations usually fluctuates synchronously with changes in external loads. Abnormal responses caused by local stiffness changes are easily mixed with the overall load response, making it difficult to accurately distinguish between load changes and changes in the pole's own state. Furthermore, local response anomalies may also be caused by factors such as load direction changes, short-term vibrations, or the temporary opening and closing of cracks. Based solely on changes at a single location or in a single monitoring cycle, it is difficult to determine whether anomalies exist continuously in adjacent areas of the pole, thus affecting the stability of stiffness degradation area identification and health status assessment. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: an online health status monitoring system for cement poles based on the Internet of Things, comprising: The rod strain acquisition module acquires the surface longitudinal strain value at the corresponding monitoring position through the monitoring node, and transmits the surface longitudinal strain value to the online analysis terminal through the Internet of Things communication link; The strain transfer analysis module arranges the longitudinal strain values ​​of the surface within the same load cycle according to the installation height of each monitoring node in a time sequence and associates adjacent positions to obtain the strain transfer sequence of the pole height. The load influence elimination module generates strain transmission difference characteristics by analyzing the correlation between the bending strain characteristics in the strain transmission sequence of the pole height and the transmission relationship between the common load variation law and adjacent monitoring positions. The stiffness degradation analysis module obtains the stiffness degradation path of the rod by analyzing the correlation between the strain transfer difference characteristics during the loading and unloading process and the transfer process in the height direction of the rod. The health status generation module continuously matches and fits the stiffness degradation path of the pole within adjacent load cycles to obtain the health status distribution of the cement pole.

[0005] Preferably, the processing flow of the rod strain acquisition module includes: Obtain the longitudinal strain values ​​of the cement pole along the circumferential direction and construct a circumferential longitudinal strain sequence according to the circumferential distribution. A first-order circular harmonic fitting is performed on the longitudinal strain sequence of the circumference, and the cosine and sine coefficients obtained from the fitting are combined to form the bending strain vector. By arranging the bending strain vectors sequentially according to the sampling time, the bending strain change sequence at the corresponding monitoring location is obtained; Each monitoring node combines the bending strain change sequence with the node location identifier and sampling period identifier into a strain data packet, which is then sent to the online analysis terminal via the Internet of Things communication link.

[0006] Preferably, the processing flow of the strain transfer analysis module includes: The monitoring nodes installed along the height of the concrete poles are sorted in descending order of installation height, and the monitoring node with the largest installation height is marked as the load reference node. For the two bending strain vectors formed at adjacent sampling times of the load reference node, calculate the angle between the two bending strain vectors, arrange the angles according to the sampling times, and obtain the bending direction change sequence. Calculate the angle difference between adjacent bending direction angles in the bending direction angle sequence in turn. When the product of two adjacent angle differences is not greater than zero and the two adjacent angle differences are not simultaneously zero, mark the sampling time corresponding to the two adjacent angle differences as the direction conversion time. Arrange the corresponding direction conversion times according to the sampling order, determine the sampling interval between two adjacent direction conversion times as the load application period, and extract the bending strain change sequence of the remaining monitoring nodes according to the same period boundary. The bending strain vectors within each load cycle are arranged according to the sampling time, and the sampling interval between adjacent sampling times is used as the time delay interval. According to the preset embedding order, starting from the first bending strain vector, the same number of bending strain vectors as the preset embedding order are sequentially extracted. The bending strain vectors obtained from each truncation are arranged into matrix columns according to the sampling order. Different matrix columns are arranged according to the truncation order to obtain the strain time series matrix of the corresponding monitoring node. Based on the installation height, the strain time series matrices of different monitoring nodes are connected in descending order to obtain the pole height strain transfer sequence.

[0007] Preferably, the process for analyzing the variation law of the common load includes: Singular subspace decomposition was performed on each strain time series matrix in the pole height strain transfer sequence to extract the time variation subspace of the corresponding monitoring node. Multiply each time-varying subspace by its transpose matrix to obtain an orthogonal projection matrix. Normalize and superimpose different orthogonal projection matrices to obtain a common-varying projection matrix. Perform eigenvalue decomposition on the common transformation projection matrix, sort the different eigenvalues ​​in descending order, and calculate the difference between two adjacent eigenvalues ​​after sorting. Mark the location where the maximum difference first appears as the target location, mark the previous feature value corresponding to the target location as the target feature value, and mark the descending order of the target feature values ​​as the common load dimension. Extract the eigenvectors corresponding to the eigenvalues ​​that are sorted first and have the same number as the dimension of the common load, and form the common load variation basis according to the order of the corresponding eigenvalues. By projecting the strain time series matrices of different monitoring nodes onto a common load variation basis, the strain coordinate matrices of the corresponding monitoring nodes under the same common load coordinates are obtained.

[0008] Preferably, the process of analyzing the transmission relationship between adjacent monitoring locations includes: Select two adjacent strain time series matrices in the strain transfer sequence of pole height in turn, and read the monitoring nodes corresponding to the two strain time series matrices respectively; The strain time series matrix with the monitoring position at the top is used as the input strain matrix, and the strain time series matrix with the monitoring position at the bottom is used as the output strain matrix. A joint covariance matrix is ​​constructed based on the input strain matrix and the output strain matrix. Based on the joint covariance matrix, the input strain matrix and the output strain matrix are respectively converted into a metric transformation input matrix and a metric transformation output matrix. For the transformed metric transformation input matrix and metric transformation output matrix, the minimum residual mapping is solved by orthogonal operators to obtain the current strain transfer matrix of the corresponding two monitoring nodes; The strain transfer matrices are arranged in descending order according to the different monitoring nodes to obtain the current strain transfer matrix sequence.

[0009] Preferably, the process for generating strain transfer difference characteristics includes: Obtain the reference strain transfer matrix sequence of the cement pole, and extract the reference strain transfer matrix corresponding to each current strain transfer matrix position from the reference strain transfer matrix sequence; The current strain transfer matrix and the reference strain transfer matrix at the same position are combined into a transfer matrix pair. The transfer matrix pair is then generalizedly decomposed to extract the current upper triangular transfer matrix and the reference upper triangular transfer matrix. Calculate the inverse of the reference upper triangular transfer matrix, and multiply the current upper triangular transfer matrix with the inverse of the reference upper triangular transfer matrix to obtain the relative strain transfer matrix at the corresponding monitoring location. Perform a matrix logarithmic transformation on the relative strain transfer matrix, and use the resulting logarithmic matrix as the strain transfer difference characteristics of the corresponding monitoring node at the corresponding monitoring location.

[0010] Preferably, the process of analyzing the loading and unloading process includes: Arrange the bending strain vectors of the load reference nodes according to the sampling time within one load cycle, calculate the modulus of each bending strain vector, and mark the bending strain vector with the largest modulus as the period peak vector. Extract the sampling interval between the start time of the load application period and the sampling time corresponding to the peak vector of the period, as the loading stage; Extract the sampling interval between the sampling time corresponding to the peak vector of the period and the end time of the load application period, and use it as the unloading phase; The bending strain change sequences of different monitoring nodes during the loading and unloading stages are extracted respectively. According to the processing flow of the strain transfer analysis module and the load influence elimination module, the strain transfer difference characteristics obtained by each monitoring node are marked as loading strain transfer matrix and unloading strain transfer matrix respectively. Calculate the inverse matrix of the unloading strain transfer matrix, and multiply the inverse matrix of the loading strain transfer matrix with the inverse matrix of the unloading strain transfer matrix to obtain the loading and unloading difference matrix at the corresponding monitoring location. Perform a matrix logarithmic transformation on the loading and unloading difference matrix, and use the result of the logarithmic transformation as the loading and unloading inconsistency feature of the corresponding monitoring location.

[0011] Preferably, the process of transferring the analysis rod height includes: Based on the descending order of the monitoring nodes in terms of installation height, a pair of location nodes is constructed according to the monitoring positions corresponding to two adjacent monitoring nodes, and the reference strain transfer matrix of the location nodes is obtained. For each pair of location nodes, the loading / unloading inconsistency feature corresponding to the monitoring node at the upper position is marked as the upstream difference matrix; Multiply the reference strain transfer matrix, the upstream difference matrix, and the inverse of the reference strain transfer matrix in sequence to obtain the prediction difference matrix corresponding to the monitoring node located at the bottom. Calculate the inverse of the prediction difference matrix, multiply the loading and unloading inconsistency feature corresponding to the monitoring position below by the inverse of the prediction difference matrix, perform a matrix logarithmic transformation on the multiplication result, and calculate the norm of the resulting matrix to obtain the transmission difference value of the corresponding node pair. Sort the monitoring nodes in descending order of installation height, accumulate the transmission difference values ​​of different monitoring node pairs, and divide by the number of location node pairs to obtain the average cumulative transmission difference value. When the average cumulative transmission difference value is not greater than the preset path difference threshold, the corresponding monitoring location sequence is constructed based on the accumulated monitoring node pairs; Based on the inconsistent loading and unloading characteristics of different monitoring locations according to the sequence of monitoring locations, a stiffness degradation path of the rod is generated according to the arrangement result.

[0012] Preferably, the process of continuously matching and fitting the stiffness degradation path of the bar within adjacent load cycles includes: Read the previous load cycle and the next load cycle, extract the stiffness degradation path of the rod respectively, and mark them as the previous cycle path and the next cycle path respectively; The path transmission cost matrix is ​​constructed by using the monitoring positions in the previous cycle path as the matrix rows and the monitoring positions in the subsequent cycle path as the matrix columns. For each element in the path transmission cost matrix, calculate the normalized installation height difference between the two corresponding monitoring locations, and the normalized logarithmic distance between the two loading / unloading inconsistencies. Add the normalized installation height difference to the logarithmic distance of the normalized matrix, and use the sum as the path transmission cost corresponding to the matrix element to solve for the minimum total transmission cost of the path transmission cost matrix. The minimum total transmission cost is mapped inversely, and the mapping result is used as the path matching quality of the corresponding monitoring location between two adjacent load cycles. Select consecutive load cycles according to the preset number of tracking cycles, accumulate the path matching quality of the same monitoring location between adjacent load cycles, and obtain the continuous stiffness degradation value of the corresponding monitoring location. The continuous stiffness degradation values ​​of the cement poles are arranged in descending order according to the installation height of different monitoring locations to obtain the continuous stiffness degradation distribution.

[0013] Preferably, the process by which the health status generation module fits the continuous stiffness degradation value and local stiffness retention rate at different monitoring locations includes: Several preset load levels were applied to different monitoring locations of the cement poles, and the surface longitudinal strain values ​​at the corresponding monitoring locations under the corresponding load levels were collected to obtain the corresponding bending strain change sequence. The continuous matching process of the strain transfer analysis module, load influence elimination module, stiffness degradation analysis module and health state generation module is invoked to obtain the continuous stiffness degradation value corresponding to each transverse load. Calculate the bending moment at the corresponding monitoring location based on the load value and monitoring location corresponding to different load levels; The bending curvature at the corresponding monitoring location is calculated based on the bending strain change sequence, and the ratio of bending moment to bending curvature is used as the measured value of local bending stiffness at the corresponding monitoring location. Divide the measured value of bending stiffness at the same monitoring location by the initial measured value of bending stiffness to obtain the local stiffness retention rate at the corresponding monitoring location. By combining the local stiffness retention rate with the corresponding continuous stiffness degradation value, stiffness calibration data pairs are obtained. Piecewise interpolation is used to fit the stiffness calibration data pairs to obtain the stiffness state mapping curve. The continuous stiffness degradation value of each monitoring location obtained from online monitoring is input into the stiffness state mapping curve to obtain the local stiffness retention rate of the corresponding monitoring location. By integrating the local stiffness retention rates at different monitoring locations, the health status distribution of cement poles is obtained, and the health status distribution of cement poles is sent to the monitoring platform via an IoT communication link.

[0014] The technical effects and advantages of the Internet of Things-based online health status monitoring system for cement poles of this invention are as follows: (1) The present invention constructs a bending strain vector by circular harmonic fitting, extracts the load change basis common to each monitoring position, and compares the current strain transfer matrix with the reference strain transfer matrix under a unified coordinate system to obtain the strain transfer difference characteristics; it can highlight the change in the transfer relationship between adjacent monitoring positions and reduce the influence of common load fluctuation on local state identification.

[0015] (2) The present invention constructs loading strain transfer matrix and unloading strain transfer matrix respectively, extracts loading and unloading inconsistency features, and then analyzes the continuity of the feature between adjacent monitoring positions according to the reference transfer relationship to form the stiffness degradation path of the rod; it can distinguish isolated, short-term local changes from stiffness degradation that is continuously distributed along the rod height.

[0016] (3) The present invention performs sequential matching of stiffness degradation paths in adjacent cycles, accumulates the path matching quality of each monitoring position, and obtains the continuous stiffness degradation value; then uses the local stiffness retention rate obtained by lateral loading calibration to establish a stiffness state mapping curve; it can reduce the influence of occasional changes on the judgment result and convert the continuous degradation degree into a local stiffness retention rate with mechanical significance. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall method flow of the Internet of Things-based online health status monitoring system for cement poles according to the present invention.

[0018] Figure 2 This is a flowchart illustrating the process of analyzing the transmission of pole height in the online health status monitoring system for cement poles based on the Internet of Things of the present invention.

[0019] Figure 3 This is a schematic diagram illustrating the process of continuously matching and fitting the pole stiffness degradation path within adjacent load cycles in the IoT-based online health status monitoring system for cement poles of the present invention.

[0020] Figure 4 This is a schematic diagram of the module structure of the Internet of Things-based online health status monitoring system for cement poles according to the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] This application provides an online health status monitoring system for cement poles based on the Internet of Things (IoT). By deploying monitoring nodes along the height of the cement pole, longitudinal strain is collected around the circumference, and a bending strain sequence is constructed. Combining common load separation, adjacent location transmission analysis, loading / unloading difference identification, and cross-cycle path matching, continuous stiffness degradation values ​​are obtained, which are then mapped to local stiffness retention rates, forming a distribution of the cement pole's health status and uploading it to the monitoring platform. This invention separates common load interference, identifies continuous stiffness degradation across locations, and quantifies local health status.

[0023] Please see Figure 1 , Figure 2 , Figure 3 and Figure 4 In this embodiment of the invention, the IoT-based online health status monitoring system for cement poles is implemented in detail through the following modules: The online health status monitoring system for cement poles includes: multiple monitoring nodes deployed along the height of the cement pole, a pole strain acquisition module, a strain transfer analysis module, a load influence elimination module, a stiffness degradation analysis module, and a health status generation module; In this embodiment, a tapered cement pole with a total length of 12m and a burial depth of 2m is selected, with the pole length above ground being 10m. Starting from the ground, monitoring nodes are set at 1.5m, 3.0m, 4.5m, 6.0m, 7.5m and 9.0m respectively, and are recorded as the first monitoring node to the sixth monitoring node in descending order of installation height. Four strain sensors are installed at equal angles along the circumference of the concrete pole at each monitoring node. The installation angles of the four strain sensors are 0°, 90°, 180° and 270° respectively. The sensitive direction of each strain sensor is consistent with the axis of the concrete pole, and the longitudinal strain value of the concrete pole surface is collected. Each monitoring node collects data synchronously at a sampling frequency of 50Hz, and the collected results are sent to the online analysis terminal through the Internet of Things communication link.

[0024] The rod strain acquisition module collects the surface longitudinal strain value at the corresponding monitoring location through the monitoring node, and transmits the surface longitudinal strain value to the online analysis terminal through the Internet of Things communication link.

[0025] In one embodiment of the present invention, the processing flow of the rod strain acquisition module includes: The longitudinal strain values ​​of the cement pole's surface along the circumference are acquired, and a circumferential longitudinal strain sequence is constructed according to the circumferential distribution. A first-order circular harmonic fitting is performed on the circumferential longitudinal strain sequence, and the cosine and sine coefficients obtained from the fitting are combined to form the bending strain vector. The principle of the first-order circular harmonic fitting is to decompose the strain values ​​distributed along the circumference into an average component and first-order sine and cosine components. The longitudinal strain of the cement pole's cross-section under planar bending state is approximately sinusoidally distributed along the circumference, and the first-order circular harmonic component can represent the magnitude and direction of the bending strain. When the cement pole is subjected to a transverse load, the circumferential longitudinal strain at the same monitoring location can be represented by a first-order circular harmonic model as follows: ; In the formula, Indicates the installation angle is The strain sensor collects the longitudinal strain value of the surface; c represents the axial average strain at the monitoring location; a and b represent the components of the bending strain in two mutually perpendicular directions, respectively; the modulus of the bending strain vector represents the bending strain amplitude at the monitoring location, and the direction of the bending strain vector represents the bending direction of the cement pole.

[0026] For four strain sensors arranged at equal angles, the cosine and sine coefficients can be obtained according to the following formula: The cosine and sine coefficients are combined to form the bending strain vector: The bending strain vectors are arranged sequentially according to the sampling time to obtain the bending strain change sequence at the corresponding monitoring location. Each monitoring node combines the bending strain change sequence with the node location identifier and the sampling period identifier into a strain data packet, which is then sent to the online analysis terminal through the Internet of Things communication link.

[0027] The strain transfer analysis module arranges the longitudinal strain values ​​of the surface within the same load cycle according to the installation height of each monitoring node and correlates adjacent positions to obtain the strain transfer sequence of the pole height.

[0028] The processing flow of the strain transfer analysis module includes: In one embodiment of the present invention, monitoring nodes arranged along the height of the cement pole are sorted in descending order of installation height, and the monitoring node with the largest installation height is marked as the load reference node; for two bending strain vectors formed at adjacent sampling times of the load reference node, the angle between the two bending strain vectors is calculated, and the angles are arranged according to the sampling times to obtain the bending direction change sequence; wherein, the angle is calculated by the difference between the bending strain vector at the later sampling time and the bending strain vector at the migration sampling time; for the bending strain vector formed at each sampling time, the corresponding bending direction angle is calculated: In the formula, the atan2 function determines the bending direction angle based on the quadrants in which the cosine and sine coefficients are located.

[0029] The angle difference between adjacent bending direction angles in the bending direction angle sequence is calculated sequentially. When the product of two adjacent angle differences is not greater than zero and the two adjacent angle differences are not simultaneously zero, the sampling time corresponding to the two adjacent angle differences is marked as the direction conversion time. The corresponding direction conversion times are arranged according to the sampling order, and the sampling interval between two adjacent direction conversion times is determined as the load application period. The bending strain change sequence of the remaining monitoring nodes is extracted according to the same period boundary.

[0030] For example, in a continuous monitoring data, if the 38th, 139th, and 240th sampling times satisfy the above-mentioned direction conversion condition, then the interval between the 38th and 139th sampling times is determined as one load application cycle, and the interval between the 139th and 240th sampling times is determined as the next load application cycle. According to the same cycle boundary, the bending strain change sequence of the remaining five monitoring nodes is extracted respectively, so that the six monitoring positions have the same start and end times in the same load application cycle.

[0031] The bending strain vectors within each load cycle are arranged according to the sampling time, with the sampling interval between adjacent sampling times as the time delay interval. The preset embedding order is an integer greater than 1 and less than the number of sampling times contained in the corresponding load cycle. Taking a load cycle containing 101 sampling times as an example, the preset embedding order is set to 4, and one sampling interval between adjacent sampling times is determined as the time delay interval. According to the preset embedding order, starting from the first bending strain vector, the same number of bending strain vectors as the preset embedding order are sequentially extracted. The bending strain vectors obtained from each extraction are arranged into matrix columns according to the sampling order. Different matrix columns are arranged according to the extraction order to obtain the strain time series matrix of the corresponding monitoring node.

[0032] Based on the installation height, the strain time series matrices of different monitoring nodes are connected in descending order to obtain the pole height strain transfer sequence. Specifically, the above processing is performed on six monitoring positions to obtain six strain time series matrices, which are then arranged in descending order of installation height to form the pole height strain transfer sequence.

[0033] The load influence elimination module generates strain transmission difference characteristics by analyzing the correlation between the bending strain characteristics in the pole height strain transmission sequence and the common load variation law and the transmission relationship between adjacent monitoring locations.

[0034] In one embodiment of the present invention, the process of analyzing the common load variation law by the load influence elimination module includes: Singular value decomposition (SVD) was performed on each strain time series matrix in the pole height strain transfer sequence to extract the time variation subspace of the corresponding monitoring node. The principle of SVD is to decompose the matrix into three parts: spatial direction, singular values, and temporal direction. The right singular vector in the matrix represents the main temporal pattern of strain data change with sampling time, and the singular values ​​represent the contribution of the corresponding temporal pattern to the original data. After SVD, the singular values ​​are arranged from largest to smallest as follows: ;in, Let represent the i-th singular value; n is the total number of singular values; in this embodiment, the preset energy ratio is set to 0.95, and the temporal subspace dimension is determined according to the following formula:

[0035] In the formula, r represents the number of singular values ​​selected when accumulating from the largest singular value, which is also the time subspace dimension to be determined; m represents the total number of singular values ​​contained in the strain time series matrix; for example, the cumulative squared value of the first six singular values ​​at the 6.0m monitoring position accounts for 96.2% of the sum of squared singular values ​​of all singular values, while the first five account for only 93.8%, so the time subspace dimension of the monitoring position is determined to be 6; select several right singular vectors that determine the time subspace dimension to form the time variation subspace.

[0036] Multiplying each time-varying subspace by its transpose matrix yields an orthogonal projection matrix. Normalizing and superimposing these different orthogonal projection matrices results in a common variation projection matrix. The orthogonal projection matrix represents the time variation range spanned by the main time pattern at the monitoring location. Normalization and superposition involve first dividing the orthogonal projection matrix P by its trace value, using the following formula: ; The normalized orthogonal projection matrix is ​​obtained by averaging and superimposing the six normalized orthogonal projection matrices. .

[0037] Eigenvalue decomposition is performed on the common transformation projection matrix. The resulting eigenvalues ​​are sorted in descending order, and the differences between adjacent eigenvalues ​​are calculated sequentially. The position where the maximum difference first appears is marked as the target position, the preceding eigenvalue at the target position is marked as the target eigenvalue, and the descending order of the target eigenvalues ​​is marked as the common load dimension. For example, the top ten eigenvalues, sorted from largest to smallest, are: 0.98, 0.95, 0.91, 0.88, 0.84, 0.80, 0.77, 0.73, 0.31, 0.27. Adjacent eigenvalues... The differences are calculated sequentially. The difference between the eighth and ninth eigenvalues ​​is 0.42, which is the first occurrence of the maximum difference. Therefore, the common load dimension is determined to be 8. The eigenvectors corresponding to the first eight eigenvalues ​​are extracted to form the common load variation basis. The eigenvectors corresponding to the eigenvalues ​​that are ranked first and have the same number as the common load dimension are extracted and the common load variation basis is formed according to the order of the corresponding eigenvalues. The strain time series matrix of different monitoring nodes is projected onto the common load variation basis to obtain the strain coordinate matrix of the corresponding monitoring nodes under the same common load coordinate.

[0038] In one embodiment of the present invention, the process of the load influence elimination module analyzing the transmission relationship between adjacent monitoring locations includes: Two adjacent strain time series matrices in the strain transfer sequence of pole height are selected sequentially, and the monitoring nodes corresponding to the two strain time series matrices are read respectively. The strain time series matrix with the monitoring position above is used as the input strain matrix, and the strain time series matrix with the monitoring position below is used as the output strain matrix. A joint covariance matrix is ​​constructed based on the input strain matrix and the output strain matrix. Based on the joint covariance matrix, the input strain matrix and the output strain matrix are converted into metric transformation input matrix and metric transformation output matrix respectively. Specifically, this includes calculating the input covariance matrix and the output covariance matrix, adding the two covariance matrices to obtain the joint covariance matrix, performing eigenvalue decomposition on the joint covariance matrix, and constructing the inverse square root matrix based on the eigenvalues, as shown in the following formula:

[0039] In the formula, C represents the joint covariance matrix; Let E represent the inverse square root matrix of the joint covariance matrix; E represents the orthogonal eigenvector matrix composed of the eigenvectors of the joint covariance matrix. Represents an eigenvalue diagonal matrix; This represents the inverse square root matrix of the eigenvalue diagonal matrix; represents the eigenvalues, f represents the total number of eigenvalues; diag() means to arrange the values ​​in parentheses into a diagonal matrix; The following formula is used to convert the input strain matrix and the output strain matrix:

[0040] In the formula, This represents the strain time series matrix at the monitoring location; This represents the strain time series matrix at the monitoring location; Represents the input matrix for metric transformation; This represents the output matrix of the metric transformation.

[0041] For the transformed metric transformation input matrix and metric transformation output matrix, the minimum residual mapping is solved using orthogonal operators to obtain the current strain transfer matrix corresponding to the two monitoring nodes; the formula is as follows:

[0042] In the formula, O represents the orthogonal mapping matrix to be solved; represents the optimal orthogonal mapping matrix that minimizes the objective function; argmin represents the variable values ​​that minimize the subsequent objective function. This represents the result of transforming the metric transformation input matrix according to the orthogonal mapping matrix; This represents the residual matrix between the input and output matrices after mapping. The Frobenius norm is represented; the square symbol indicates that the squares of each element of the residual matrix are summed. Therefore, the optimization objective, under orthogonal constraints, is to minimize the total squared error between the mapped upstream and downstream strain change modes. The orthogonal constraints ensure that the mapping process does not arbitrarily change the length and angle relationship of the vectors, and the resulting variable values... It mainly represents the rotation and recombination relationship of strain change direction between adjacent monitoring locations; according to the descending order of different monitoring nodes, the corresponding strain transfer matrices are arranged to obtain the current strain transfer matrix sequence.

[0043] In one embodiment of the present invention, the process of generating strain transfer difference characteristics by the load influence elimination module includes: A sequence of reference strain transfer matrices for cement poles is obtained, and the reference strain transfer matrix corresponding to each current strain transfer matrix position is extracted from the sequence. In this embodiment, after the cement pole is installed and passes the initial mechanical performance test, a sequence of reference strain transfer matrices is established according to the above process, under the condition of no obvious cracks and stiffness degradation. The current strain transfer matrix and the reference strain transfer matrix at the same position are combined into a transfer matrix pair, and the transfer matrix pair is generalizedly decomposed to extract the current upper triangular transfer matrix and the reference upper triangular transfer matrix.

[0044] Taking the position node pair from 7.5m to 6.0m as an example, let the current strain transfer matrix be A and the corresponding reference strain transfer matrix be B. Perform generalized Schur decomposition on the matrix pair (A, B). The basic principle of generalized Schur decomposition is to use the same set of left and right orthogonal matrices to transform the two matrices simultaneously, so that the two matrices are transformed to the same orthogonal coordinate system and form an upper triangular structure. Since the current strain transfer matrix and the reference strain transfer matrix use the same orthogonal transformation, their corresponding elements can be compared under the same coordinate relationship.

[0045] Calculate the inverse of the reference upper triangular transfer matrix, and multiply the current upper triangular transfer matrix by the inverse of the reference upper triangular transfer matrix to obtain the relative strain transfer matrix at the corresponding monitoring location. Perform a matrix logarithmic transformation on the relative strain transfer matrix, and use the resulting logarithmic matrix as the strain transfer difference characteristic of the corresponding monitoring node at the corresponding monitoring location. The logarithmic transformation uses the following formula: In the formula, R is the relative strain transfer matrix; It is a logarithmic matrix; The above processing is performed on each of the five position node pairs to obtain the corresponding strain transmission difference characteristics. The strain transmission difference characteristics describe the change of the current rod state relative to the initial mechanical state and are the basis for subsequent judgments on whether the change has loading and unloading inconsistencies and rod height continuity.

[0046] In existing online monitoring, wind loads and conductor traction cause synchronous strain changes at different locations on cement poles. The transmission differences caused by local stiffness changes are easily mixed with the overall load response. To address this, this invention constructs a bending strain vector through circular harmonic fitting, extracts the load change basis common to each monitoring location, and compares the current strain transmission matrix with the reference strain transmission matrix under a unified coordinate system to obtain strain transmission difference characteristics. This can identify changes in the transmission relationship between adjacent monitoring locations and reduce the impact of common load fluctuations on local state identification.

[0047] The stiffness degradation analysis module obtains the stiffness degradation path of the rod by analyzing the correlation between the strain transfer difference characteristics during the loading and unloading process and the transfer process in the rod height direction.

[0048] In one embodiment of the present invention, the process of analyzing the loading and unloading process by the stiffness degradation analysis module includes: A load cycle obtained from the strain transfer analysis module is selected, which includes 101 sampling times. The bending strain vectors of the load reference nodes within a load cycle are arranged according to the sampling times. The modulus of each bending strain vector is calculated, and the bending strain vector with the largest modulus is marked as the period peak vector. For example, the largest modulus in this load cycle first appears at the 52nd sampling time. Therefore, the bending strain vector corresponding to the 52nd sampling time is determined as the period peak vector.

[0049] The sampling interval between the start time of the load application period and the sampling time corresponding to the peak vector of the period is extracted as the loading stage; the sampling interval between the sampling time corresponding to the peak vector of the period and the end time of the load application period is extracted as the unloading stage; the sampling interval between the 1st sampling time and the 52nd sampling time is determined as the loading stage, and the sampling interval between the 53rd sampling time and the 101st sampling time is determined as the unloading stage.

[0050] The bending strain change sequences of different monitoring nodes during the loading and unloading phases are extracted respectively. According to the processing flow of the strain transfer analysis module and the load influence elimination module, the strain transfer difference characteristics obtained for each monitoring node are marked as loading strain transfer matrix and unloading strain transfer matrix respectively. The inverse matrix of the unloading strain transfer matrix is ​​calculated, and the loading strain transfer matrix and the inverse matrix of the unloading strain transfer matrix are multiplied to obtain the loading and unloading difference matrix of the corresponding monitoring location.

[0051] A matrix logarithmic transformation was performed on the loading-unloading difference matrix, and the result was used as the loading-unloading inconsistency feature for the corresponding monitoring location. When the cement pole has no cracks or the cracks remain closed, the strain transmission relationship between the loading and unloading stages is similar, the loading-unloading difference matrix is ​​close to the identity matrix, and the loading-unloading inconsistency feature is close to the zero matrix. When a local crack opens during the loading stage and closes during the unloading stage, the local stiffness and strain transmission relationship between the loading and unloading stages differs, and the matrix elements and norm of the loading-unloading inconsistency feature increase. For example, the Frobenius norm of the loading-unloading inconsistency feature at the 4.5m monitoring location in the initial state is 0.026, increasing to 0.184 during the online monitoring stage; the corresponding value at the 1.5m monitoring location in the same period is 0.031. This result indicates that the loading-unloading inconsistency change mainly occurs in the middle of the pole, rather than changing synchronously throughout the entire cement pole. The loading-unloading inconsistency features at each monitoring location are used as a factor in analyzing whether the inconsistency change is transmitted along the direction of increasing pole height.

[0052] In one embodiment of the present invention, the process of analyzing the rod height transfer process by the stiffness degradation analysis module includes: Based on the descending order of the monitoring nodes in terms of installation height, position node pairs are constructed according to the monitoring positions corresponding to two adjacent monitoring nodes, and the baseline strain transfer matrix of the position nodes is obtained. For each position node pair, the loading and unloading inconsistency feature corresponding to the monitoring node at the upper position is marked as the upstream difference matrix. The baseline strain transfer matrix, the upstream difference matrix, and the inverse matrix of the baseline strain transfer matrix are multiplied sequentially to obtain the predicted difference matrix corresponding to the monitoring node at the lower position. This process belongs to matrix conjugate transfer. Its principle is that the baseline strain transfer matrix describes the coordinate transformation relationship between two adjacent monitoring positions under the initial health state. By using the baseline strain transfer matrix and its inverse matrix to perform a conjugate transformation on the upstream difference matrix, the loading and unloading inconsistency feature of the upstream monitoring position can be transformed into the baseline coordinate relationship of the downstream monitoring position.

[0053] The inverse of the prediction difference matrix is ​​calculated. The loading / unloading inconsistency feature corresponding to the monitoring position at the lower position is multiplied by the inverse of the prediction difference matrix. The result of the multiplication is subjected to a matrix logarithmic transformation, and the norm of the resulting matrix is ​​calculated to obtain the transmission difference value of the corresponding position node pair. Here, the norm specifically refers to the same Frobenius norm mentioned above. The monitoring nodes are sorted in descending order of installation height, and the transmission difference values ​​of different monitoring node pairs are accumulated and divided by the number of position node pairs to obtain the average cumulative transmission difference value. When the average cumulative transmission difference value is not greater than the preset path difference threshold, the corresponding monitoring position sequence is constructed based on the accumulated monitoring node pairs. When multiple monitoring position sequences contain the same number of monitoring positions, the monitoring position sequence with the smallest average cumulative transmission difference value is retained.

[0054] The transmission difference values ​​of each node pair obtained during the initial mechanical performance testing phase are acquired, and the maximum value of the transmission difference value is used as the preset path difference threshold. Based on the sequence of monitoring positions, the loading and unloading inconsistency features corresponding to different monitoring positions are arranged, and the stiffness degradation path of the pole is generated according to the arrangement results. The stiffness degradation path does not only indicate that a certain monitoring position is abnormal, but also indicates that the loading and unloading inconsistency features can continuously correspond between multiple adjacent monitoring positions according to the strain transmission relationship of the cement pole in the initial healthy state.

[0055] Even after eliminating the influence of common loads, the transmission differences at individual monitoring locations may still be caused by load reversal, short-term vibration, or temporary opening and closing of cracks, making it difficult to determine whether continuous degradation has occurred. To address this, this invention constructs loading strain transfer matrices and unloading strain transfer matrices separately, extracts loading-unloading inconsistency features, and then analyzes the continuity of these features between adjacent monitoring locations according to the baseline transmission relationship, forming a stiffness degradation path of the pole. This allows for the differentiation between isolated, short-term local changes and stiffness degradation that is continuously distributed along the pole height.

[0056] The health status generation module continuously matches and fits the stiffness degradation path of the pole within adjacent load cycles to obtain the health status distribution of the cement pole.

[0057] In one embodiment of the present invention, the process of continuously matching and fitting the stiffness degradation path of the bar within adjacent load cycles includes: Read the previous and subsequent load cycles, extract the stiffness degradation path of the rod, and label it as the previous cycle path and the subsequent cycle path, respectively. Construct a path transmission cost matrix using the monitoring positions in the previous cycle path as matrix rows and the monitoring positions in the subsequent cycle path as matrix columns. For each element in the path transmission cost matrix, calculate the normalized installation height difference between the corresponding two monitoring positions, and the normalized logarithmic distance between the two loading / unloading inconsistencies. In this embodiment, the logarithmic distance is calculated using the following formula:

[0058] In the formula, This represents the matrix logarithmic distance between the i-th monitoring position in the previous cycle path and the j-th monitoring position in the subsequent cycle path. This represents the loading and unloading inconsistency characteristic corresponding to the i-th monitoring position in the stiffness degradation path of the bar in the previous load cycle; i represents the sequence number of the monitoring positions in the previous cycle path. Let represent the inverse matrix of the loading-unloading inconsistency characteristic corresponding to the j-th monitoring location in the subsequent load cycle; j represents the sequence number of the monitoring location in the subsequent cycle path; and the matrix logarithmic distance is... The smaller the value, the closer the loading and unloading inconsistencies between the two monitoring locations are; conversely, the greater the difference in the loading and unloading transmission states between the two monitoring locations during adjacent load cycles.

[0059] Add the normalized installation height difference to the logarithmic distance of the normalized matrix, and use the sum as the path transmission cost corresponding to the matrix element; denoted as... The number of monitoring locations in the previous and subsequent cycle paths is used to set normalized transmission amounts for different matrix rows and columns. Under the condition that the row number relationship and column number relationship of any two non-zero transmission elements are the same, the minimum total transmission cost of the path transmission cost matrix is ​​solved to obtain the order-preserving transmission matrix.

[0060] In this embodiment, for example, both the pre-cycle path and the post-cycle path contain five monitoring positions, so the normalized transmission amount of each matrix row and the normalized transmission amount of each matrix column are both 0.2. The basic principle of the optimal transmission algorithm is to find the transmission matrix with the minimum total transmission cost given the matrix row transmission amount and matrix column reception amount. Traditional optimal transmission allows matching between arbitrary rows and columns. This embodiment further restricts the order relationship of non-zero transmission elements, and does not allow the pre-cycle node at a higher position to match with the post-cycle node at a lower position, and then the pre-cycle node at a lower position to cross-match with the post-cycle node at a higher position. The specific process includes: Let X be the matrix corresponding to the minimum total transmission cost, and the objective is to solve for: and satisfy And the condition under which non-zero transmission elements do not undergo positional order crossover.

[0061] For each monitoring location in the preceding cycle path, the transmission quantities of each corresponding row in the sequentially maintained transmission matrix are multiplied by the corresponding path transmission cost, summed, and then divided by the sum of the transmission quantities of the corresponding matrix rows to obtain the minimum total transmission cost for the corresponding monitoring location; denoted as . The minimum total transmission cost is inversely proportionally mapped, and the mapping result is used as the path matching quality of the corresponding monitoring location between two adjacent load cycles. Through formula Calculated; Select consecutive load cycles according to the preset tracking cycle number, accumulate the path matching quality of the same monitoring position between adjacent load cycles to obtain the continuous stiffness degradation value of the corresponding monitoring position; when the monitoring position does not appear in the pole stiffness degradation path of any load cycle, set the path matching quality corresponding to the load cycle to zero; sort the corresponding continuous stiffness degradation values ​​according to the descending order of the installation height of different monitoring positions to obtain the continuous stiffness degradation distribution of the cement pole.

[0062] In one embodiment of the present invention, the process by which the health status generation module fits the continuous stiffness degradation value and local stiffness retention rate at different monitoring locations includes: Several preset load levels of transverse load were applied to different monitoring locations on the cement pole. The surface longitudinal strain values ​​of the corresponding monitoring locations under the corresponding load levels were collected to obtain the corresponding bending strain change sequence. Under each bending stiffness state, the transverse load loading point corresponding to each monitoring location was selected. The transverse load loading point was located 1.0m above the corresponding monitoring location. For example, for the 4.5m monitoring location, the 5.5m location was determined as the transverse load loading point. Transverse loads of three preset load levels of 0.5kN, 1.0kN and 1.5kN were applied in sequence. Under each load level, the surface longitudinal strain values ​​of six monitoring locations were collected, and the processing procedure of the pole strain acquisition module was called to obtain the bending strain change sequence of each monitoring location.

[0063] The continuous matching process of the strain transfer analysis module, load effect elimination module, stiffness degradation analysis module, and health state generation module is invoked to obtain the continuous stiffness degradation value corresponding to each transverse load; based on the load values ​​corresponding to different load levels and monitoring locations, the bending moment at the corresponding monitoring location is calculated; the calculation formula is: In the formula, M represents the bending moment at the monitoring location; Indicates the lateral load value; The height of the lateral load application point is represented by h; the height of the monitoring position is represented by h. The bending curvature of the corresponding monitoring position is calculated based on the bending strain change sequence, and the ratio of bending moment to bending curvature is used as the measured value of local bending stiffness at the corresponding monitoring position. Specifically, this includes: first, calculating the vector magnitude of the bending strain vector at each sampling time, obtaining the outer radius of the monitoring position, dividing the vector magnitude by h to obtain the bending curvature; then dividing the bending moment by the bending curvature to obtain the measured value of local bending stiffness.

[0064] Dividing the measured bending stiffness value at the same monitoring location by the initial measured bending stiffness value yields the local stiffness retention rate for that location. The initial measured bending stiffness value is the baseline local bending stiffness value obtained from a lateral loading test at each monitoring location when the concrete pole is in its initial intact state. Combining the local stiffness retention rate with the corresponding sustained stiffness degradation value yields a stiffness calibration data pair. A piecewise interpolation method is then used to fit the stiffness calibration data pair to obtain the stiffness state mapping curve. The specific process is as follows: The stiffness calibration data pairs are arranged in ascending order of continuous stiffness degradation value, and the piecewise cubic Hermite interpolation method is used for fitting. The principle of piecewise cubic Hermite interpolation is to construct a cubic polynomial between every two adjacent calibration data pairs, so that the polynomial passes through both calibration data pairs at the same time, and the slope at the interpolation node is limited according to the direction of change of the adjacent calibration data pairs. Let the kth calibration data be denoted as , Indicates the sustained stiffness degradation value; The sustained stiffness degradation value indicates The measured local stiffness retention rate is the local stiffness retention rate corresponding to the left endpoint of the current interpolation interval; for two adjacent calibration points, let... The corresponding piecewise cubic interpolation function is expressed as: ; ; ; ; ; ; In the formula, This represents the local stiffness retention rate obtained through the stiffness state mapping curve when the continuous stiffness degradation value is W. The basis function for the function value at the left endpoint controls the local stiffness retention rate at the left endpoint. Contribution to the interpolation results; The slope basis function is the slope at the left endpoint; The basis function for the function value at the right endpoint controls the local stiffness retention rate at the right endpoint. Contribution to the interpolation results; The slope basis function is the slope at the right endpoint; This represents the slope of the stiffness state mapping curve at the kth calibration point, i.e., the rate of change of the local stiffness retention rate relative to the sustained stiffness degradation value.

[0065] The continuous stiffness degradation value of each monitoring location obtained from online monitoring is input into the stiffness state mapping curve to obtain the local stiffness retention rate of the corresponding monitoring location. The local stiffness retention rates of different monitoring locations are integrated to obtain the health status distribution of the cement poles. The health status distribution of the cement poles is sent to the monitoring platform through the Internet of Things communication link.

[0066] Even if a stiffness degradation path is identified within a single load cycle, its location and extent may still fluctuate with the cycle, and the matrix characteristics are difficult to directly represent the actual mechanical performance. To address this, this invention performs sequential matching of stiffness degradation paths in adjacent cycles, accumulates the path matching quality at each monitoring location, and obtains the continuous stiffness degradation value. Then, it uses the local stiffness retention rate obtained from the lateral loading calibration to establish a stiffness state mapping curve. This reduces the impact of occasional changes on the judgment results and converts the continuous degradation degree into a local stiffness retention rate with mechanical significance.

[0067] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0068] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0069] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0070] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. An online health status monitoring system for cement poles based on the Internet of Things, characterized in that, include: The rod strain acquisition module acquires the surface longitudinal strain value at the corresponding monitoring position through the monitoring node, and transmits the surface longitudinal strain value to the online analysis terminal through the Internet of Things communication link; The strain transfer analysis module arranges the longitudinal strain values ​​of the surface within the same load cycle according to the installation height of each monitoring node in a time sequence and associates adjacent positions to obtain the strain transfer sequence of the pole height. The load influence elimination module generates strain transmission difference characteristics by analyzing the correlation between the bending strain characteristics in the strain transmission sequence of the pole height and the transmission relationship between the common load variation law and adjacent monitoring positions. The stiffness degradation analysis module obtains the stiffness degradation path of the rod by analyzing the correlation between the strain transfer difference characteristics during the loading and unloading process and the transfer process in the height direction of the rod. The health status generation module continuously matches and fits the stiffness degradation path of the pole within adjacent load cycles to obtain the health status distribution of the cement pole.

2. The online health status monitoring system for cement poles based on the Internet of Things according to claim 1, characterized in that, The processing flow of the rod strain acquisition module includes: Obtain the longitudinal strain values ​​of the cement pole along the circumferential direction and construct a circumferential longitudinal strain sequence according to the circumferential distribution. A first-order circular harmonic fitting is performed on the longitudinal strain sequence of the circumference, and the cosine and sine coefficients obtained from the fitting are combined to form the bending strain vector. By arranging the bending strain vectors sequentially according to the sampling time, the bending strain change sequence at the corresponding monitoring location is obtained; Each monitoring node combines the bending strain change sequence with the node location identifier and sampling period identifier into a strain data packet, which is then sent to the online analysis terminal via the Internet of Things communication link.

3. The online health status monitoring system for cement poles based on the Internet of Things according to claim 2, characterized in that, The processing flow of the strain transfer analysis module includes: The monitoring nodes installed along the height of the concrete poles are sorted in descending order of installation height, and the monitoring node with the largest installation height is marked as the load reference node. For the two bending strain vectors formed at adjacent sampling times of the load reference node, calculate the angle between the two bending strain vectors, arrange the angles according to the sampling times, and obtain the bending direction change sequence. Calculate the angle difference between adjacent bending direction angles in the bending direction angle sequence in turn. When the product of two adjacent angle differences is not greater than zero and the two adjacent angle differences are not simultaneously zero, mark the sampling time corresponding to the two adjacent angle differences as the direction conversion time. Arrange the corresponding direction conversion times according to the sampling order, determine the sampling interval between two adjacent direction conversion times as the load application period, and extract the bending strain change sequence of the remaining monitoring nodes according to the same period boundary. The bending strain vectors within each load cycle are arranged according to the sampling time, and the sampling interval between adjacent sampling times is used as the time delay interval. According to the preset embedding order, starting from the first bending strain vector, the same number of bending strain vectors as the preset embedding order are sequentially extracted. The bending strain vectors obtained from each truncation are arranged into matrix columns according to the sampling order. Different matrix columns are arranged according to the truncation order to obtain the strain time series matrix of the corresponding monitoring node. Based on the installation height, the strain time series matrices of different monitoring nodes are connected in descending order to obtain the pole height strain transfer sequence.

4. The online health status monitoring system for cement poles based on the Internet of Things according to claim 3, characterized in that, The process for analyzing the variation law of the common load includes: Singular subspace decomposition was performed on each strain time series matrix in the pole height strain transfer sequence to extract the time variation subspace of the corresponding monitoring node. Multiply each time-varying subspace by its transpose matrix to obtain an orthogonal projection matrix. Normalize and superimpose different orthogonal projection matrices to obtain a common-varying projection matrix. Perform eigenvalue decomposition on the common transformation projection matrix, sort the different eigenvalues ​​in descending order, and calculate the difference between two adjacent eigenvalues ​​after sorting. Mark the location where the maximum difference first appears as the target location, mark the previous feature value corresponding to the target location as the target feature value, and mark the descending order of the target feature values ​​as the common load dimension. Extract the eigenvectors corresponding to the eigenvalues ​​that are sorted first and have the same number as the dimension of the common load, and form the common load variation basis according to the order of the corresponding eigenvalues. By projecting the strain time series matrices of different monitoring nodes onto a common load variation basis, the strain coordinate matrices of the corresponding monitoring nodes under the same common load coordinates are obtained.

5. The online health status monitoring system for cement poles based on the Internet of Things according to claim 4, characterized in that, The process for analyzing the transmission relationship between adjacent monitoring locations includes: Select two adjacent strain time series matrices in the strain transfer sequence of pole height in turn, and read the monitoring nodes corresponding to the two strain time series matrices respectively; The strain time series matrix with the monitoring position at the top is used as the input strain matrix, and the strain time series matrix with the monitoring position at the bottom is used as the output strain matrix. A joint covariance matrix is ​​constructed based on the input strain matrix and the output strain matrix. Based on the joint covariance matrix, the input strain matrix and the output strain matrix are respectively converted into a metric transformation input matrix and a metric transformation output matrix. For the transformed metric transformation input matrix and metric transformation output matrix, the minimum residual mapping is solved by orthogonal operators to obtain the current strain transfer matrix of the corresponding two monitoring nodes; The strain transfer matrices are arranged in descending order according to the different monitoring nodes to obtain the current strain transfer matrix sequence.

6. The online health status monitoring system for cement poles based on the Internet of Things according to claim 5, characterized in that, The process for generating strain transfer difference characteristics includes: Obtain the reference strain transfer matrix sequence of the cement pole, and extract the reference strain transfer matrix corresponding to each current strain transfer matrix position from the reference strain transfer matrix sequence; The current strain transfer matrix and the reference strain transfer matrix at the same position are combined into a transfer matrix pair. The transfer matrix pair is then generalizedly decomposed to extract the current upper triangular transfer matrix and the reference upper triangular transfer matrix. Calculate the inverse of the reference upper triangular transfer matrix, and multiply the current upper triangular transfer matrix with the inverse of the reference upper triangular transfer matrix to obtain the relative strain transfer matrix at the corresponding monitoring location. Perform a matrix logarithmic transformation on the relative strain transfer matrix, and use the resulting logarithmic matrix as the strain transfer difference characteristics of the corresponding monitoring node at the corresponding monitoring location.

7. The online health status monitoring system for cement poles based on the Internet of Things according to claim 6, characterized in that, The analysis of the loading and unloading process includes the following steps: Arrange the bending strain vectors of the load reference nodes according to the sampling time within one load cycle, calculate the modulus of each bending strain vector, and mark the bending strain vector with the largest modulus as the period peak vector. Extract the sampling interval between the start time of the load application period and the sampling time corresponding to the peak vector of the period, as the loading stage; Extract the sampling interval between the sampling time corresponding to the peak vector of the period and the end time of the load application period, and use it as the unloading phase; The bending strain change sequences of different monitoring nodes during the loading and unloading stages are extracted respectively. According to the processing flow of the strain transfer analysis module and the load influence elimination module, the strain transfer difference characteristics obtained by each monitoring node are marked as loading strain transfer matrix and unloading strain transfer matrix respectively. Calculate the inverse matrix of the unloading strain transfer matrix, and multiply the inverse matrix of the loading strain transfer matrix with the inverse matrix of the unloading strain transfer matrix to obtain the loading and unloading difference matrix at the corresponding monitoring location. Perform a matrix logarithmic transformation on the loading and unloading difference matrix, and use the result of the logarithmic transformation as the loading and unloading inconsistency feature of the corresponding monitoring location.

8. The online health status monitoring system for cement poles based on the Internet of Things according to claim 7, characterized in that, The process of transferring the height of the analysis rod includes: Based on the descending order of the monitoring nodes in terms of installation height, a pair of location nodes is constructed according to the monitoring positions corresponding to two adjacent monitoring nodes, and the reference strain transfer matrix of the location nodes is obtained. For each pair of location nodes, the loading / unloading inconsistency feature corresponding to the monitoring node at the upper position is marked as the upstream difference matrix; Multiply the reference strain transfer matrix, the upstream difference matrix, and the inverse of the reference strain transfer matrix in sequence to obtain the prediction difference matrix corresponding to the monitoring node located at the bottom. Calculate the inverse of the prediction difference matrix, multiply the loading and unloading inconsistency feature corresponding to the monitoring position below by the inverse of the prediction difference matrix, perform a matrix logarithmic transformation on the multiplication result, and calculate the norm of the resulting matrix to obtain the transmission difference value of the corresponding node pair. Sort the monitoring nodes in descending order of installation height, accumulate the transmission difference values ​​of different monitoring node pairs, and divide by the number of location node pairs to obtain the average cumulative transmission difference value. When the average cumulative transmission difference value is not greater than the preset path difference threshold, the corresponding monitoring location sequence is constructed based on the accumulated monitoring node pairs; Based on the inconsistent loading and unloading characteristics of different monitoring locations according to the sequence of monitoring locations, a stiffness degradation path of the rod is generated according to the arrangement result.

9. The online health status monitoring system for cement poles based on the Internet of Things according to claim 8, characterized in that, The process of continuously matching and fitting the stiffness degradation path of the bar within adjacent load cycles includes: Read the previous load cycle and the next load cycle, extract the stiffness degradation path of the rod respectively, and mark them as the previous cycle path and the next cycle path respectively; The path transmission cost matrix is ​​constructed by using the monitoring positions in the previous cycle path as the matrix rows and the monitoring positions in the subsequent cycle path as the matrix columns. For each element in the path transmission cost matrix, calculate the normalized installation height difference between the two corresponding monitoring locations, and the normalized logarithmic distance between the two loading / unloading inconsistencies. Add the normalized installation height difference to the logarithmic distance of the normalized matrix, and use the sum as the path transmission cost corresponding to the matrix element to solve for the minimum total transmission cost of the path transmission cost matrix. The minimum total transmission cost is mapped inversely, and the mapping result is used as the path matching quality of the corresponding monitoring location between two adjacent load cycles. Select consecutive load cycles according to the preset number of tracking cycles, accumulate the path matching quality of the same monitoring location between adjacent load cycles, and obtain the continuous stiffness degradation value of the corresponding monitoring location. The continuous stiffness degradation values ​​of the cement poles are arranged in descending order according to the installation height of different monitoring locations to obtain the continuous stiffness degradation distribution.

10. The online health status monitoring system for cement poles based on the Internet of Things according to claim 9, characterized in that, The process by which the health status generation module fits the continuous stiffness degradation value and local stiffness retention rate at different monitoring locations includes: Several preset load levels were applied to different monitoring locations of the cement poles, and the surface longitudinal strain values ​​at the corresponding monitoring locations under the corresponding load levels were collected to obtain the corresponding bending strain change sequence. The continuous matching process of the strain transfer analysis module, load influence elimination module, stiffness degradation analysis module and health state generation module is invoked to obtain the continuous stiffness degradation value corresponding to each transverse load. Calculate the bending moment at the corresponding monitoring location based on the load value and monitoring location corresponding to different load levels; The bending curvature at the corresponding monitoring location is calculated based on the bending strain change sequence, and the ratio of bending moment to bending curvature is used as the measured value of local bending stiffness at the corresponding monitoring location. Divide the measured value of bending stiffness at the same monitoring location by the initial measured value of bending stiffness to obtain the local stiffness retention rate at the corresponding monitoring location. By combining the local stiffness retention rate with the corresponding continuous stiffness degradation value, stiffness calibration data pairs are obtained. Piecewise interpolation is used to fit the stiffness calibration data pairs to obtain the stiffness state mapping curve. The continuous stiffness degradation value of each monitoring location obtained from online monitoring is input into the stiffness state mapping curve to obtain the local stiffness retention rate of the corresponding monitoring location. By integrating the local stiffness retention rates at different monitoring locations, the health status distribution of cement poles is obtained, and the health status distribution of cement poles is sent to the monitoring platform via an IoT communication link.