A method for wireless intelligent monitoring of a crossing structure based on the internet of things

Abstract

The application discloses a kind of based on Internet of Things spanning frame state wireless intelligent monitoring method, specifically related to electric power construction safety monitoring technical field, first in spanning frame monitoring site deployment sensor node, spanning frame is collected multidimensional data under normal operating condition by sensor node;Then multidimensional feature extraction is carried out, and uneven settlement feature set, frame body vibration feature set, wire impact feature set and frame body connector feature set are obtained;Then spanning frame normal state feature model is constructed by machine learning algorithm;Finally, the spanning frame multidimensional data collected is input into spanning frame normal state feature model, and response mechanism is triggered according to abnormal output result, and feedback is carried out.The application is based on intelligent analysis and early warning of machine learning, establishes spanning frame normal state model, can evaluate the state of spanning frame in real time, accurately, and constructs multi-modal fusion early warning, improves the accuracy and reliability of early warning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power construction safety monitoring, in particular to a wireless intelligent monitoring method for the state of a crossing frame based on the Internet of Things. BACKGROUND

[0002] In the construction process of power line construction, reconstruction and the like, a crossing frame is an important temporary facility for supporting and protecting a crossed object (such as a highway, a railway, a communication line and the like); the state of the crossing frame is directly related to construction safety, and if the crossing frame is inclined, displaced, structurally damaged and the like, the crossed object can be damaged, and even a serious safety accident can be caused.

[0003] Traditional crossing frame state monitoring mainly relies on artificial periodic inspection, a monitoring method based on the Internet of Things applied to the field of crossing frame monitoring, and a structure monitoring method based on machine learning; among them, the running state data of the crossing frame is recorded by artificial periodic inspection; the monitoring method based on the Internet of Things mostly only monitors a single parameter (such as an inclination angle), and simply threshold alarms according to the monitoring result; the structure monitoring method based on machine learning can improve the monitoring accuracy, so as to achieve the purpose of crossing frame state monitoring, but lacks the fusion analysis ability of multi-source heterogeneous data of the crossing frame.

[0004] However, there are still some deficiencies in actual application, for example, artificial inspection has poor real-time performance, low monitoring accuracy, high labor intensity and the like, which can cause the crossing frame to collapse; the monitoring method based on the Internet of Things cannot comprehensively reflect the overall state of the crossing frame; the structure monitoring method based on machine learning can improve the monitoring accuracy, but lacks the fusion analysis ability of multi-source heterogeneous data of the crossing frame, and has insufficient identification ability for nonlinear abnormal early warning mode; therefore, an efficient, real-time and intelligent crossing frame state monitoring method is urgently needed. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a wireless intelligent monitoring method for the state of a crossing frame based on the Internet of Things to solve the problems proposed in the above background.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a wireless intelligent monitoring method for the state of a crossing frame based on the Internet of Things, comprising:

[0007] S1: deploying a crossing frame sensor node: deploying a sensor node at a crossing frame monitoring site, and each sensor node is equipped with an Internet of Things wireless communication module;

[0008] S2: multi-dimensional data acquisition: collecting multi-dimensional data of the crossing frame under normal running state by the sensor node according to a preset sampling frequency, and transmitting the collected data to a remote monitoring center through the Internet of Things wireless communication module;

[0009] S3: Multi-dimensional feature extraction: The remote monitoring center extracts multi-dimensional features from the multi-dimensional data of the crossing frame under normal operation state through big data analysis technology, respectively obtaining a frame uneven settlement feature set, a frame vibration feature set, a conductor impact feature set, and a frame connector feature set;

[0010] S4: Training the multi-dimensional feature set extracted under the normal operation state of the crossing frame through a machine learning algorithm to generate a feature judgment boundary of the normal state and construct a crossing frame normal state feature model;

[0011] S5: Through the Internet of Things wireless communication module, the multi-dimensional data of the crossing frame collected by the sensor node in real time is subjected to the same feature extraction, the extracted multi-dimensional features are input into the crossing frame normal state feature model, and the response mechanism triggered according to the model abnormal output result;

[0012] S6: The response mechanism triggered according to the model abnormal output result is transmitted to the mobile terminal of the construction management personnel for human-computer interaction.

[0013] Technical effects and advantages of the present application:

[0014] 1. The present application realizes real-time collection and transmission of sensor data through Internet of Things technology, including frame uneven settlement data set, frame vibration data set, conductor impact data set, and frame connector data set, which can timely find the subtle changes of the crossing frame state, greatly improving the timeliness and accuracy of monitoring compared with manual inspection, and providing accurate data support for crossing frame state monitoring and analysis;

[0015] 2. The present application extracts features from the normal state data of the crossing frame through multi-dimensional analysis to obtain a multi-dimensional feature set, improving the fusion analysis capability of multi-source heterogeneous data of the crossing frame; each parameter reflects the specific state information of the crossing frame from different angles, and the combination of these parameters can more accurately evaluate the safety and stability of the crossing frame;

[0016] 3. The present application establishes a crossing frame normal state model based on intelligent analysis and early warning of machine learning, which can evaluate the state of the crossing frame in real time and accurately, and constructs a multi-modal fusion early warning. Compared with the traditional monitoring method based on threshold judgment, the machine learning algorithm can better adapt to the state changes of the crossing frame under different working conditions, improving the accuracy and reliability of the early warning. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The present application is a whole process schematic diagram.

[0018] Figure 2 The present application is a method flowchart. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0020] Please refer to Figure 1 As shown in the figure, the present application provides a wireless intelligent monitoring system for the state of a crossing frame based on the Internet of Things, which comprises a crossing frame sensor node deployment module, a crossing frame state data acquisition and transmission module, a crossing frame normal state feature extraction module, a crossing frame normal state feature model construction module, a crossing frame state intelligent monitoring module and a crossing frame state intelligent monitoring feedback module.

[0021] The crossing frame sensor node deployment module is connected with the crossing frame state data acquisition and transmission module, the crossing frame normal state feature extraction module is connected with the crossing frame state data acquisition and transmission module and the crossing frame normal state feature model construction module respectively, and the crossing frame state intelligent monitoring module is connected with the crossing frame normal state feature model construction module and the crossing frame state intelligent monitoring feedback module respectively.

[0022] The crossing frame sensor node deployment module: deploy sensor nodes at the monitoring part of the crossing frame, and each sensor node is equipped with an Internet of Things wireless communication module;

[0023] The crossing frame state data acquisition and transmission module: collect multi-dimensional data of the crossing frame under normal operation state through the sensor nodes according to a preset sampling frequency, and transmit the collected data to the remote monitoring center through the Internet of Things wireless communication module;

[0024] The crossing frame normal state feature extraction module: the remote monitoring center extracts multi-dimensional features from the multi-dimensional data of the crossing frame under normal operation state through big data analysis technology, respectively obtains a frame uneven settlement feature set, a frame vibration feature set, a conductor impact feature set and a frame connector feature set, and transmits the extracted feature sets to the crossing frame normal state feature model construction module;

[0025] The crossing frame normal state feature model construction module: trains the multi-dimensional feature sets extracted from the crossing frame under normal operation state through a machine learning algorithm, generates a feature judgment boundary of the normal state, constructs a crossing frame normal state feature model, and transmits it to the crossing frame state intelligent monitoring module;

[0026] The cross-pole state intelligent monitoring module: through the Internet of Things wireless communication module, the multi-dimensional data of the cross-pole collected by the sensor node in real time is processed for the same feature extraction, the extracted multi-dimensional features are input into the cross-pole normal state feature model, the response mechanism triggered according to the model abnormal output result is transmitted to the cross-pole state intelligent monitoring feedback module;

[0027] The cross-pole state intelligent monitoring feedback module: the response mechanism triggered according to the model abnormal output result is transmitted to the mobile terminal of the construction management personnel for human-computer interaction.

[0028] Referring to Figure 2 The cross-pole state intelligent monitoring feedback module: the response mechanism triggered according to the model abnormal output result is transmitted to the mobile terminal of the construction management personnel for human-computer interaction.

[0029] S1: Deploy the cross-pole sensor node: deploy the sensor node at the cross-pole monitoring part, and each sensor node is equipped with an Internet of Things wireless communication module;

[0030] The embodiment needs to be specifically explained that the sensor nodes are deployed at the monitoring parts of the crossing frame, including but not limited to installing the inclination sensors at the key connection parts of the vertical rods and the horizontal rods for monitoring the inclination angle of the crossing frame in real time; installing the pressure sensors at the bottom of the frame body for monitoring the pressure changes borne by the key parts of the frame body; installing the displacement sensors at the parts of the frame body where displacement is prone to occur for monitoring whether displacement occurs in each part of the frame body, such as installing the displacement sensors at the connection parts of the horizontal rods and the vertical rods for monitoring the relative displacement between the horizontal rods and the vertical rods; installing the acceleration sensors at the top of the crossing frame for measuring the vibration acceleration of the crossing frame structure; during the installation of the sensors, it is ensured that the sensor nodes are in close contact with the monitored parts and are fixed firmly to avoid affecting the accuracy of the monitoring data due to loosening or falling off.

[0031] S2: Multi-dimensional data acquisition: The sensor nodes collect multi-dimensional data of the crossing frame in the normal running state according to the preset sampling frequency, and transmit the collected data to the remote monitoring center through the Internet of Things wireless communication module;

[0032] The embodiment needs to be specifically explained that the multi-dimensional data of the crossing frame in the normal running state includes a frame body uneven settlement data set, a frame body vibration data set, a conductor impact data set and a frame body connector data set; the frame body uneven settlement data set includes east side leg pressure, west side leg pressure, distance difference of the foundation at the four corners of the crossing frame and soil humidity; the frame body vibration data set includes horizontal acceleration at the top of the frame body, longitudinal acceleration, frame body structure vibration main frequency and average wind speed of the environment; the conductor impact data set includes the lowest point height of the conductor, the top height of the crossing frame, the displacement of the conductor, the monitoring position of the conductor and the crossing frame in the horizontal direction, the acoustic emission energy of the conductor contacting the frame body at the moment, the vibration main frequency when the conductor is impacted, the strain value of the crossing frame in the stable state and the peak value of the strain value in the process of conductor impact; the frame body connector data set includes group bolt pretightening force, frame body connection temperature, relative displacement between the horizontal rods and the vertical rods of the frame body and corrosion sensor resistance value.

[0033] The embodiment needs to be specifically explained that the sensor nodes have preprocessed the collected multi-dimensional data through data preprocessing technology, for example, using Kalman filtering algorithm to filter the inclination angle, vibration and other data to remove random noise interference; then, by setting a reasonable threshold range, the data is subjected to outlier rejection, for example, for the inclination angle data, if the collected value exceeds the normal inclination angle range (such as ±5°) by a certain percentage (such as 20%), it is determined as an outlier and is rejected; finally, the preprocessed multi-dimensional data is transmitted to the remote monitoring center through the Internet of Things wireless communication module.

[0034] S3: Multi-dimensional feature extraction: The remote monitoring center extracts multi-dimensional features from the multi-dimensional data of the normal operation state of the crossing frame through big data analysis technology, respectively obtaining a frame uneven settlement feature set, a frame vibration feature set, a conductor impact feature set, and a frame connector feature set, including the following steps:

[0035] S3.1: Uneven settlement feature set: During the monitoring period, according to the preset collection frequency t1, first measure the distance difference of the four corners of the foundation of the crossing frame through data measurement technology (such as millimeter wave radar technology), take the maximum value of the distance difference of the four corners in n1 groups of data, respectively ΔH1, ΔH2, ΔH3 and ΔH4, obtain the average foundation settlement speed vp, , Δt represents the monitoring period time interval; then obtain the maximum value Wmax of soil humidity in n1 groups of data; secondly, obtain n1 groups of pressure values, obtain the average value μ(P_e) of the east leg pressure of the crossing frame body, the average value μ(P_w) of the west leg pressure and the average value μ(P_ew) of the combined pressure of the east and west legs, μ(P_ew)=(P_e+P_w) / 2, obtain the frame leg pressure fluctuation coefficient η(P_ew), η(P_ew)=σ(P_ew) / μ(P_ew), , obtain the frame uneven settlement risk index TDI, , add 0.01 to avoid the denominator being 0; finally obtain the frame uneven settlement feature set, including the average foundation settlement speed vp and the frame uneven settlement risk index TDI;

[0036] It needs to be specifically pointed out in this embodiment that the average foundation settlement speed reflects the overall trend and speed of foundation settlement, and the uneven settlement risk index focuses on the uniformity of foundation settlement; only when the average settlement speed is within a reasonable range and the uneven settlement risk index is low, can it be considered that the foundation is in a relatively stable state.

[0037] S3.2: Frame vibration feature set: During the monitoring period, first collect n2 groups of acceleration through an acceleration sensor according to a preset collection frequency t2, obtain the average lateral acceleration A x and the average longitudinal acceleration A y of the top of the frame, obtain the root mean square RMS of the frame vibration acceleration, ; then perform frequency spectrum analysis on the frame combined acceleration A tot (t) through frequency spectrum analysis technology (such as fast Fourier transform technology FFT), obtain n2 frame structure vibration main frequencies f, and then take the average value to obtain the average vibration main frequency μ(f), the frame combined acceleration , A x (t) and A y(t) respectively represent the top transverse acceleration and the top longitudinal acceleration of each acquisition; the vibration energy accumulation risk index VEI is obtained, , V w represent the environmental average wind speed; finally, the vibration feature set of the frame body is obtained, including the frame body vibration acceleration root mean square RMS and the frame body vibration energy accumulation risk index VEI;

[0038] It is particularly pointed out in this embodiment that when the RMS value is large, it indicates that the crossing frame is experiencing relatively strong vibration and the vibration energy is high. This may mean that the crossing frame is subjected to a larger external force, such as strong wind, construction load, etc., and its structure may be in a relatively unstable state; when the VEI value exceeds a certain range, reinforcement measures, adjustment of the construction scheme or strengthening of the monitoring frequency, etc. may be taken to ensure the safe and stable operation of the crossing frame.

[0039] S3.3: Conductor impact feature set:

[0040] S3.3.1: In the monitoring period, first, the height values of the lowest points of the conductor are collected in n3 groups according to the preset collection frequency t3 through the ranging technology, the average height He of the lowest points of the conductor is obtained, and the height Hf of the top of the crossing frame is obtained. Then, through the positioning technology (such as GPS positioning technology), n3 difference values of the conductor and the crossing frame in the horizontal direction of the monitoring position are obtained, and the sum of the absolute values of the difference values is obtained to obtain the horizontal offset distance Dm; secondly, the displacement of the conductor is measured in n3 groups through the displacement sensor installed on the conductor, and the average displacement Sm of the conductor swing is obtained by taking the average value, and the dynamic approach risk index DAI of the conductor is obtained, ;

[0041] S3.3.2: In the monitoring period, first, the acoustic emission energy Ea of the conductor contacting the frame body is monitored through the piezoelectric sensor; then the frame body synthesis acceleration A tot c (t) vibration main frequency f c , the frame body vibration main frequency offset Δf is obtained, Δf = |f-f c |, f represents the frame body structure vibration main frequency; secondly, the strain value Esta of the crossing frame in the stable state and the peak value Emax of the strain value in the process of conductor impact are obtained through the strain gauge installed at the monitoring position of the crossing frame (such as the strain gauge installed on the diagonal line of the frame body), the conductor impact strain coefficient KE is obtained, KE = Emax / Esta, and the dynamic impact risk index ICI of the conductor is obtained, , tc represents the conductor impact time, which is determined by the strain signal envelope integral method, , E(t) represents the strain value changing over time, t_1 represents the impact start time, t_2 represents the impact end time, that is, the strain signal returns to the static value or approaches the static value, the integration process stops, at this time the result tc of integration is the total duration of the impact on the conductor, which can be obtained by recording the sensor output signal during the impact of the conductor, determining the time points of the impact start and end through time domain analysis of the signal, and the unit is ms; finally, the conductor impact feature set is obtained, including the conductor dynamic approach risk indicator DAI and the conductor dynamic impact risk indicator ICI;

[0042] It needs to be specifically explained in this embodiment that in the use scenario of the crossing frame, various dynamic loads will act on it, such as the conductor swinging and impacting the crossing frame under the influence of wind force, self-gravitational change and other factors; when these dynamic loads act on the components of the crossing frame, the components will deform, and the dynamic strain is a physical quantity used to describe the deformation degree of the components under dynamic changes.

[0043] S3.4: Frame connector feature set: In the monitoring period, n4 groups of bolt pre-tightening force F, frame connection temperature T, relative displacement xd between frame crossbars and vertical rods, and corrosion sensor resistance value R are obtained respectively; then the maximum value of the bolt pre-tightening force Fmax, the maximum value of the frame connection temperature Tmax, the maximum value of the relative displacement between the frame crossbars and the vertical rods xdmax, and the maximum value of the corrosion sensor resistance value Rmax are taken respectively to obtain the frame connector health degree indicator CHI, , T0 represents the reference temperature, and sd0 represents the rated relative displacement threshold value; the bolt pre-tightening force is obtained by ultrasonic measurement technology, and the frame connection temperature is obtained by a temperature sensor; finally, the frame connector feature set is obtained, including the frame connector health degree indicator CHI;

[0044] It needs to be specifically explained in this embodiment that when the connector is corroded, the corrosion products will affect the resistance value of the sensor. By measuring the resistance change of the sensor, the corrosion degree of the connector can be indirectly evaluated; Tmax+273 is the conversion of Celsius temperature into thermodynamic temperature (Kelvin temperature).

[0045] S4: Through a machine learning algorithm (for example, a support vector machine SVM), the multi-dimensional feature set extracted under the normal operating state of the crossing frame is trained to generate a feature judgment boundary of the normal state, and a crossing frame normal state feature model is constructed, including the following steps:

[0046] S4.1: Model training: a one-class support vector machine is used in combination with a radial basis function RBF kernel function, the multi-dimensional feature set extracted under the normal state of the crossing frame is taken as a sample x, and n groups are taken for model training, and the RBF kernel function is defined as: , x i and xj respectively represent the i-th and j-th samples, i < j, i e n, j e n, ||x i -x j || represents the Euclidean distance, and γ is a kernel parameter, , n_fea represents the feature dimension, and σ 2 represents the variance of the feature, , x k represents the k-th feature, and Var() represents the variance function;

[0047] S4.2: Model output: based on the model training of S4.1, the model outputs the decision function f(x), generates the feature judgment boundary across the normal state of the span, , ρ represents the bias term, and the decision threshold is used to adjust the position of the classification boundary, and α J represents the Lagrange multiplier of the J-th support vector, which can be obtained by model training, N represents the number of support vectors, and sgn[] is a sign function, outputting values of +1 and -1, +1 representing a normal class, and -1 representing an abnormal class; finally, the span normal state feature model of the model training is constructed;

[0048] The embodiment needs to be specifically described that One-Class SVM is more suitable for small sample normal data training, and the target is to find a hyperplane to enclose the single-class data points inside it, while maximizing the distance between the hyperplane boundary and the data points; the feature dimension of the present application is 7, and the multi-dimensional feature set sample x includes the average foundation settlement velocity vp and the span uneven settlement risk index TDI, the span vibration acceleration root mean square RMS and the span vibration energy aggregation risk index VEI, the wire dynamic approach risk index DAI and the wire dynamic impact risk index ICI, and the span connector health index CHI.

[0049] S5: through the Internet of Things wireless communication module, the multi-dimensional data of the span collected by the sensor node in real time is processed, the same feature extraction is performed, the extracted multi-dimensional features are input into the span normal state feature model, and a response mechanism is triggered according to the model abnormal output result, including the following steps:

[0050] S5.1: through the Internet of Things wireless communication module, the multi-dimensional data of the span collected by the sensor node in real time is processed, the same feature extraction is performed, the extracted multi-dimensional features sample x rt is input into the span normal state feature model, and the decision function value f(x rt ) is calculated, ; if f(x rt ) > 0, it indicates that the sample x rt is located on the normal state side of the feature boundary, it is judged that the span is in a normal state, and if f(x rt) < 0, indicating that the sample x rt If the abnormal state is located on one side of the feature boundary, it is determined that the abnormal state exists across the frame state, and an abnormal diagnosis response mechanism is triggered;

[0051] S5.2: Based on the abnormal diagnosis response mechanism triggered by S4.1, through big data analysis technology, the frame uneven settlement risk index TDI, the frame vibration energy aggregation risk index VEI, the dynamic approach risk index DAI of the conductor, the dynamic impact risk index ICI of the conductor, and the frame connector health index CHI are compared with the corresponding threshold values TDI th , VEI th , DAI th , ICI th , and CHI th , if any of them does not meet the threshold range, a yellow warning response mechanism is triggered, and if at least two of them do not meet the threshold range, a red warning response mechanism is triggered, and the response mechanism includes the specific location of the crossing frame, the abnormal type (such as frame uneven settlement, frame vibration, conductor impact, and connector abnormality), the abnormal value, and the occurrence time;

[0052] It is necessary to specifically explain that any item not meeting the threshold range includes that the frame uneven settlement risk index TDI is greater than the threshold value TDI th , the frame vibration energy aggregation risk index VEI is greater than the threshold value VEI th , the dynamic approach risk index DAI of the conductor is greater than the threshold value DAI th , the dynamic impact risk index ICI of the conductor is greater than the threshold value ICI th , and the frame connector health index CHI is less than the threshold value CHI th .

[0053] S6: The response mechanism triggered according to the model abnormal output result is transmitted to the mobile terminal of the construction management personnel for human-computer interaction, and the construction management personnel takes corresponding response measures according to the response mechanism content. For example, if the inclination degree is slight, temporary reinforcement measures such as increasing support rods, adjusting stay wires, etc. can be taken; vibration reduction devices such as shock absorbers, dampers, etc. are installed on the frame; the position of the conductor is adjusted or an isolation device is installed; if the connector is severely damaged, a new connector is replaced in time, so as to realize wireless intelligent monitoring of the crossing frame state.

[0054] Secondly, in the drawings of the disclosed embodiments, only the structures related to the disclosed embodiments are involved, other structures can be referred to the usual design, and in the case of no conflict, the same embodiment and different embodiments of the present application can be combined with each other;

[0055] Finally: the above only for the preferred embodiments of the present application, and not for limiting the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application, should be included in the scope of protection of the present application.

Claims

1. A wireless intelligent monitoring method for spanning rack status based on the Internet of Things, characterized by: include: S1: Deploy sensor nodes across the rack: Deploy sensor nodes at the monitoring location of the rack, and each sensor node is equipped with an IoT wireless communication module; S2: Multi-dimensional data collection: The sensor nodes collect multi-dimensional data of the cross-rack under normal operation according to the preset sampling frequency, and transmit the collected data to the remote monitoring center through the Internet of Things wireless communication module; S3: Multidimensional feature extraction: The remote monitoring center uses big data analysis technology to extract multidimensional features from the multidimensional data of the normal operation of the span frame, and obtains the frame uneven settlement feature set, frame vibration feature set, conductor impact feature set, and frame connector feature set. S4: Using machine learning algorithms, the multidimensional feature set extracted from the normal operation of the span is trained to generate a feature determination boundary for the normal state and to construct a span normal state feature model. S5: Through the IoT wireless communication module, the multi-dimensional data of the span rack collected in real time by the sensor node is subjected to the same feature extraction. The extracted multi-dimensional features are input into the span rack normal state feature model, and the response mechanism is triggered according to the abnormal output results of the model. The S5 implementation includes: S5.1: Through the Internet of Things wireless communication module, the multi-dimensional data collected by the sensor node in real time across the rack is extracted and the same feature extraction is performed. The extracted multi-dimensional feature samples x rt Input the normal state characteristic model of the spanning frame and calculate the decision function value f(x rt ), ; If f(x rt )>0, indicating that sample x rt Located on the normal state side of the feature boundary, it is judged that the spanning frame is in a normal state. If f(x rt )<0, indicating that sample x rt Located on the abnormal state side of the feature boundary, it is determined that there is an abnormality in the spanning rack state and the abnormal diagnosis response mechanism is triggered; S5.2: Based on the abnormal diagnosis response mechanism triggered by S4.1, through big data analysis technology, the frame uneven settlement risk index TDI, the frame vibration energy accumulation risk index VEI, the conductor dynamic approach risk index DAI, the conductor dynamic impact risk index ICI and the frame connector health index CHI are respectively compared with the corresponding threshold TDI. th VEI th 、DAI th 、ICI th and CHI th Comparison is performed. If any one of the items does not meet the threshold range, a yellow warning response mechanism is triggered. If at least two items do not meet the threshold range, a red warning response mechanism is triggered. The response mechanism includes the specific location of the span, the abnormality type, the abnormal value, and the time of occurrence. S6: The response mechanism triggered by the abnormal output result of the model is transmitted to the mobile terminal of the construction manager for human-computer interaction.

2. The method for wireless intelligent monitoring of spanning rack status based on the Internet of Things according to claim 1, characterized in that: The uneven settlement feature set in S3: During the monitoring period, according to the preset acquisition frequency t1, firstly, the distance difference of the four corners of the spanning frame is measured by data measurement technology, and the maximum value of the four corner distance difference in the n1 group of data is taken as ΔH1, ΔH2, ΔH3 and ΔH4 respectively, to obtain the average foundation settlement velocity vp, , Δt represents the monitoring cycle time interval; then obtain the maximum value Wmax of soil moisture in the n1 group of data; secondly, obtain the n1 group of pressure values, and obtain the average pressure μ(P_e) of the east leg of the span frame, the average pressure μ(P_w) of the west leg, and the average pressure μ(P_ew) of the east and west legs, μ(P_ew)=(P_e+P_w) / 2, and obtain the pressure fluctuation coefficient η(P_ew) of the span frame leg, η(P_ew)=σ(P_ew) / μ(P_ew), , we get the risk index of uneven settlement of the frame TDI, , add 0.01 to avoid the denominator being 0; finally, the frame uneven settlement feature set is obtained, including the average foundation settlement velocity and the frame uneven settlement risk index TDI.

3. The method for wireless intelligent monitoring of spanning rack status based on the Internet of Things according to claim 1, characterized in that: The frame vibration feature set in S3: During the monitoring period, first, the acceleration sensor is used to collect n2 groups of accelerations according to the preset collection frequency t2 to obtain the average lateral acceleration A of the top of the frame. x and the top longitudinal average acceleration A y , get the frame vibration acceleration RMS, ; Then the frame's synthetic acceleration A is analyzed by spectrum analysis technology tot (t) Perform spectrum analysis to obtain n2 main vibration frequencies f of the frame structure, and then take the average value to obtain the average main vibration frequency μ(f). The composite acceleration of the frame , A x (t) and A y (t) represents the top lateral acceleration and top longitudinal acceleration collected each time; the frame vibration energy accumulation risk index VEI is obtained, , V w represents the average wind speed of the environment; finally, the frame vibration feature set is obtained, including the frame vibration acceleration root mean square RMS and the frame vibration energy accumulation risk index VEI.

4. The method for wireless intelligent monitoring of spanning rack status based on the Internet of Things according to claim 1, characterized in that: The conductor impact feature set in S3 includes: first, during the monitoring period, using ranging technology, according to the preset collection frequency t3, the height values ​​of the lowest points of n3 groups of conductors are collected to obtain the average height H of the lowest points of the conductors e , get the top height H of the spanning frame f Then, through positioning technology, we obtain n3 differences in the horizontal direction between the conductor and the span at the monitoring position, and then take the absolute value of the difference and sum it to get the horizontal offset distance Dm; secondly, we use the displacement sensor installed on the conductor to measure the displacement of n3 groups of conductors, and then take the average value to get the conductor swing average displacement Sm, and get the conductor dynamic proximity risk index DAI. .

5. The method for wireless intelligent monitoring of spanning rack status based on the Internet of Things according to claim 1, characterized in that: The wire impact feature set in S3 also includes: during the monitoring period, firstly monitoring the acoustic emission energy Ea at the moment when the wire contacts the frame through the piezoelectric sensor; then obtaining the frame composite acceleration A when the frame is impacted by the wire tot c (t) the main frequency of vibration f c , we can get the main frequency offset of the frame vibration Δf, Δf=|f-fc| , f represents the main frequency of the frame structure vibration; secondly, the strain gauges installed at the monitoring position of the crossing frame are used to obtain the strain value Esta of the crossing frame in a stable state and the peak value Emax of the strain value during the conductor impact process, and the conductor impact strain coefficient KE is obtained, KE=Emax / Esta, and the conductor dynamic impact risk index ICI is obtained. , tc represents the duration of conductor impact, , E(t) represents the strain value changing with time, t_1 represents the start time of the impact, and t_2 represents the end time of the impact; finally, the conductor impact feature set is obtained, including the conductor dynamic approach risk index DAI and the conductor dynamic impact risk index ICI.

6. The method for wireless intelligent monitoring of spanning rack status based on the Internet of Things according to claim 1, characterized in that: The frame connector feature set in S3 is as follows: During the monitoring period, n4 sets of bolt preload F, frame connection temperature T, relative displacement xd between the frame crossbar and the vertical pole, and corrosion sensor resistance R are obtained respectively; then, the maximum bolt preload Fmax, the maximum frame connection temperature Tmax, the maximum relative displacement xdmax between the frame crossbar and the vertical pole, and the maximum corrosion sensor resistance Rmax are taken respectively to obtain the frame connector health index CHI. , T0 represents the reference temperature, xd0 represents the rated relative displacement threshold; The bolt preload is obtained by ultrasonic measurement technology, and the temperature of the frame connection is obtained by a temperature sensor; finally, a frame connection feature set is obtained, including a frame connection health index CHI.

7. The method for wireless intelligent monitoring of spanning rack status based on the Internet of Things according to claim 1, characterized in that: The S4 implementation includes: S4.1: Model training: A first-class support vector machine is used in combination with a radial basis function (RBF) kernel function. The multidimensional feature set extracted from the normal state of the spanning frame is used as sample x, and n groups are selected for model training. The RBF kernel function is defined as: ,K(x i ,x j ) represents the RBF kernel function, x i and x j Represent the i-th and j-th samples respectively, i<j, i∈n, j∈n, γ is the kernel parameter, , n_fea represents the feature dimension, σ 2 represents the variance of the feature, , x k represents the kth feature, Var() represents the variance function; S4.2: Model output: Based on the model training in S4.1, the model outputs the decision function f(x) to generate the feature decision boundary of the normal state of the span frame. ,ρ represents the bias term, α J represents the Lagrange multiplier of the J-th support vector, N represents the number of support vectors, sgn[] is the sign function, and the output values ​​are +1 and -1, +1 represents the normal category, and -1 represents the abnormal category; finally, the normal state feature model of the spanning rack after model training is constructed.

Images