Beidou Monitoring Method and System for the Bridge Health Degree of Long- and Short-Term Memory Learning

Abstract

The present invention discloses a Beidou monitoring method and system for bridge health based on long short-term memory learning. The method includes: obtaining bridge information of a bridge through the Beidou system; setting up a bridge health evaluation model, and calculating a first bridge health value according to the bridge information. Among them, the bridge health evaluation model includes: a bridge health evaluation function for environmental wind speed, a bridge health evaluation function for wind direction, a bridge health evaluation function for the inclination degree of the bridge, a bridge health evaluation function for the displacement of the bridge, a bridge health evaluation function for environmental temperature, and a bridge health evaluation function for environmental humidity; inputting the bridge information into a long short-term memory neural network, calculating a second first bridge health value, and completing the evaluation of the bridge health in combination with the first bridge health value.

Description

Technical Field

[0001] The present invention belongs to the technical field of Beidou monitoring of bridge health, and more specifically, relates to a method and system for Beidou monitoring of bridge health based on long short-term memory learning. Background Art

[0002] At present, a relatively large-scale bridge health monitoring system has been established in China, which uses Beidou post-differential positioning technology and multi-sensor fusion monitoring technology to conduct all-weather monitoring of important information such as the structural vibration of the bridge, the vibration of the cables, wind speed and direction, inclination, displacement, temperature and humidity. In recent years, researchers at home and abroad have achieved certain results in aspects such as the optimal layout of sensors, intelligent control of automatic monitoring, network sharing of real-time monitoring information, automatic diagnosis of damage identification, analysis of bridge bearing capacity and structural reliability, and estimation of bridge remaining life.

[0003] However, there is no technical solution in the prior art that can accurately evaluate the bridge health. Summary of the Invention

[0004] To solve the above technical problems, the present invention proposes a method for Beidou monitoring of bridge health based on long short-term memory learning, including:

[0005] Obtaining bridge information of the bridge through the Beidou system, where the bridge information includes: environmental wind speed monitored by Beidou, wind direction monitored by Beidou, inclination degree of the bridge monitored by Beidou, displacement of the bridge monitored by Beidou, environmental temperature monitored by Beidou, and environmental humidity monitored by Beidou;

[0006] Setting up a bridge health evaluation model, and calculating a first bridge health value according to the bridge information, where the bridge health evaluation model includes: a bridge health evaluation function for environmental wind speed, a bridge health evaluation function for wind direction, a bridge health evaluation function for the inclination degree of the bridge, a bridge health evaluation function for the displacement of the bridge, a bridge health evaluation function for environmental temperature, and a bridge health evaluation function for environmental humidity;

[0007] Inputting the bridge information into a long short-term memory neural network, calculating a second first bridge health value, and completing the evaluation of the bridge health in combination with the first bridge health value.

[0008] Further, the bridge health evaluation model includes:

[0009] Health = f WS (WS)·f WD (WD)·f I (I)·f D (D)·f T (T)·f H(H)

[0010] Among them, Health is the first bridge health value, f WS (WS) is the bridge health assessment function of the environmental wind speed. WS is the environmental wind speed monitored by Beidou, f WD (WD) is the bridge health assessment function of the wind direction. WD is the wind direction monitored by Beidou, f I (I) is the bridge health assessment function of the inclination degree of the bridge. I is the inclination degree of the bridge monitored by Beidou, f D (D) is the bridge health assessment function of the displacement of the bridge. D is the displacement of the bridge monitored by Beidou, f T (T) is the bridge health assessment function of the environmental temperature. T is the environmental temperature monitored by Beidou, f H (H) is the bridge health assessment function of the environmental humidity. H is the environmental humidity monitored by Beidou.

[0011] Furthermore, the bridge health assessment function f WS (WS) of the environmental wind speed includes:

[0012]

[0013] Among them, k is the first wind speed adjustment factor, WS max is the maximum wind speed that the bridge design can withstand, and c is the second wind speed adjustment factor.

[0014] Furthermore, the bridge health assessment function f WD (WD) of the wind direction includes:

[0015]

[0016] Among them, WD opt is the optimal wind direction angle.

[0017] Furthermore, the bridge health assessment function f I (I) of the inclination degree of the bridge includes:

[0018]

[0019] Among them, α is the first inclination degree adjustment factor, I max is the maximum allowable inclination degree of the bridge design, and β is the second inclination degree adjustment factor.

[0020] Furthermore, the bridge health assessment function f D (D) of the displacement of the bridge includes:

[0021]

[0022] where γ is the first displacement adjustment factor, D max is the maximum allowable displacement of the bridge design, and δ is the second displacement adjustment factor.

[0023] Furthermore, the bridge health assessment function f T (T) of the environmental temperature includes:

[0024]

[0025] where T ref is the reference temperature, T max is the highest temperature that the bridge can withstand, and T min is the lowest temperature that the bridge can withstand.

[0026] Furthermore, the bridge health assessment function f H (H) of the environmental humidity includes:

[0027]

[0028] where θ is the humidity adjustment factor, H ref is the reference humidity, μ is the first time adjustment factor, ω is the second time adjustment factor, and t is the humidity acquisition time.

[0029] Furthermore, it includes: taking the average value of the first bridge health value and the second bridge health value as the final bridge health value, and conducting the assessment of the bridge health.

[0030] The present invention also proposes a Beidou monitoring system for bridge health with long short-term memory learning, including:

[0031] An information acquisition module, which is used to acquire bridge information through the Beidou system, where the bridge information includes: the environmental wind speed monitored by Beidou, the wind direction monitored by Beidou, the inclination degree of the bridge monitored by Beidou, the displacement of the bridge monitored by Beidou, the environmental temperature monitored by Beidou, and the environmental humidity monitored by Beidou;

[0032] A module for calculating the first bridge health value, which is used to set a bridge health evaluation model and calculate the first bridge health value according to the bridge information, where the bridge health evaluation model includes: the bridge health assessment function of the environmental wind speed, the bridge health assessment function of the wind direction, the bridge health assessment function of the inclination degree of the bridge, the bridge health assessment function of the displacement of the bridge, the bridge health assessment function of the environmental temperature, and the bridge health assessment function of the environmental humidity;

[0033] An evaluation module is used to input the bridge information into a long short-term memory neural network, calculate a second first bridge health degree value, and complete the evaluation of the bridge health degree in combination with the first bridge health degree value.

[0034] Compared with the prior art by the above technical solution conceived by the present invention, the following beneficial effects are achieved:

[0035] The present invention obtains bridge information of a bridge through the Beidou system, where the bridge information includes: environmental wind speed monitored by Beidou, wind direction monitored by Beidou, inclination degree of the bridge monitored by Beidou, displacement of the bridge monitored by Beidou, environmental temperature monitored by Beidou, and environmental humidity monitored by Beidou; a bridge health degree evaluation model is set up to calculate a first bridge health degree value according to the bridge information, where the bridge health degree evaluation model includes: a bridge health degree evaluation function for environmental wind speed, a bridge health degree evaluation function for wind direction, a bridge health degree evaluation function for the inclination degree of the bridge, a bridge health degree evaluation function for the displacement of the bridge, a bridge health degree evaluation function for environmental temperature, and a bridge health degree evaluation function for environmental humidity; the bridge information is input into a long short-term memory neural network to calculate a second first bridge health degree value, and the evaluation of the bridge health degree is completed in combination with the first bridge health degree value. The present invention can accurately evaluate the bridge health degree according to multi-source data through the above technical solution. Description of the Drawings

[0036] Figure 1 is the flowchart of Embodiment 1 of the present invention;

[0037] Figure 2 is the structural diagram of the system of Embodiment 2 of the present invention;

[0038] Figure 3 is the schematic diagram of the bridge health degree intelligent monitoring system of Embodiment 5 of the present invention;

[0039] Figure 4 is the data processing flowchart of Embodiment 5 of the present invention;

[0040] Figure 5 is the internal structural diagram of the LSTM unit of Embodiment 5 of the present invention. Detailed Embodiments

[0041] In order to better understand the above technical solution, the above technical solution will be described in detail below in combination with the accompanying drawings of the specification and specific embodiments.

[0042] The method provided by the present invention can be implemented in the following terminal environment. The terminal may include one or more of the following components: a processor, a storage medium, and a display screen. Among them, at least one instruction is stored in the storage medium, and the instruction is loaded and executed by the processor to implement the method described in the following embodiments.

[0043] The processor may include one or more processing cores. The processor uses various interfaces and lines to connect various parts within the entire terminal, and by running or executing instructions, programs, code sets, or instruction sets stored in the storage medium, and by calling data stored in the storage medium, it executes various functions of the terminal and processes data.

[0044] The storage medium may include a random access memory (RAM), and may also include a read-only memory (ROM). The storage medium can be used to store instructions, programs, code, code sets, or instructions.

[0045] The display screen is used to display the user interfaces of various application programs.

[0046] All subscripts in the formula of the present invention are only for distinguishing parameters and have no actual meaning.

[0047] In addition, those skilled in the art can understand that the structure of the above terminal does not constitute a limitation on the terminal. The terminal may include more or fewer components, or combine certain components, or have different component arrangements. For example, the terminal may further include components such as a radio frequency circuit, an input unit, a sensor, an audio circuit, a power supply, etc., which will not be elaborated here.

[0048] Embodiment 1

[0049] As Figure 1 shown, the embodiment of the present invention provides a Beidou monitoring method for the bridge health of long short-term memory learning, including:

[0050] Step 101, obtain the bridge information of the bridge through the Beidou system, where the bridge information includes: the environmental wind speed monitored by Beidou, the wind direction monitored by Beidou, the inclination degree of the bridge monitored by Beidou, the displacement of the bridge monitored by Beidou, the environmental temperature monitored by Beidou, and the environmental humidity monitored by Beidou;

[0051] Step 102, set up a bridge health evaluation model, and calculate the first bridge health value according to the bridge information, where the bridge health evaluation model includes: the bridge health evaluation function of the environmental wind speed, the bridge health evaluation function of the wind direction, the bridge health evaluation function of the inclination degree of the bridge, the bridge health evaluation function of the displacement of the bridge, the bridge health evaluation function of the environmental temperature, and the bridge health evaluation function of the environmental humidity;

[0052] Specifically, the bridge health evaluation model includes:

[0053] Health = f WS (WS)·f WD (WD)·f I (I)·f D (D)·f T (T)·f H (H)

[0054] Among them, Health is the first bridge health value, f WS (WS) is the bridge health evaluation function of environmental wind speed, WS is the environmental wind speed monitored by Beidou, f WD (WD) is the bridge health evaluation function of wind direction, WD is the wind direction monitored by Beidou, f I (I) is the bridge health evaluation function of the inclination degree of the bridge, I is the inclination degree of the bridge monitored by Beidou, f D (D) is the bridge health evaluation function of the displacement of the bridge, D is the displacement of the bridge monitored by Beidou, f T (T) is the bridge health evaluation function of environmental temperature, T is the environmental temperature monitored by Beidou, f H (H) is the bridge health evaluation function of environmental humidity, H is the environmental humidity monitored by Beidou.

[0055] Specifically, the bridge health evaluation function f WS (WS) of the environmental wind speed includes:

[0056]

[0057] Among them, k is the first wind speed adjustment factor, WS max is the maximum wind speed that the bridge design can withstand, and c is the second wind speed adjustment factor.

[0058] Specifically, the bridge health evaluation function f WD (WD) of the wind direction includes:

[0059]

[0060] Among them, WD opt is the optimal wind direction angle.

[0061] Specifically, the bridge health evaluation function f I (I) of the inclination degree of the bridge includes:

[0062]

[0063] Among them, α is the first inclination degree adjustment factor, and I max is the maximum allowable inclination of the bridge design, and β is the second inclination degree adjustment factor.

[0064] Specifically, the bridge health assessment function f D (D) of the displacement of the bridge includes:

[0065]

[0066] Among them, γ is the first displacement adjustment factor, and D max is the maximum allowable displacement of the bridge design, and δ is the second displacement adjustment factor.

[0067] Specifically, the bridge health assessment function f T (T) of the environmental temperature includes:

[0068]

[0069] Among them, T ref is the reference temperature, T max is the highest temperature that the bridge can withstand, and T min is the lowest temperature that the bridge can withstand.

[0070] Specifically, the bridge health assessment function f H (H) of the environmental humidity includes:

[0071]

[0072] Among them, θ is the humidity adjustment factor, and H ref is the reference humidity, μ is the first time adjustment factor, ω is the second time adjustment factor, and t is the humidity acquisition time.

[0073] Step 103: Input the bridge information into the long short-term memory neural network, calculate the second first bridge health value, and complete the evaluation of the bridge health in combination with the first bridge health value.

[0074] Specifically, the average value of the first bridge health value and the second bridge health value is used as the final bridge health value, and the bridge health is evaluated.

[0075] Specifically,

[0076] Example 2

[0077] As Figure 2 shown, the embodiment of the present invention also provides a Beidou monitoring system for bridge health based on long short-term memory learning, including:

[0078] An information acquisition module, configured to acquire bridge information of a bridge through the Beidou system, where the bridge information includes: environmental wind speed monitored by Beidou, wind direction monitored by Beidou, inclination degree of the bridge monitored by Beidou, displacement of the bridge monitored by Beidou, environmental temperature monitored by Beidou, and environmental humidity monitored by Beidou;

[0079] A first bridge health degree value calculation module, configured to set up a bridge health degree evaluation model and calculate a first bridge health degree value according to the bridge information, where the bridge health degree evaluation model includes: a bridge health degree evaluation function for environmental wind speed, a bridge health degree evaluation function for wind direction, a bridge health degree evaluation function for the inclination degree of the bridge, a bridge health degree evaluation function for the displacement of the bridge, a bridge health degree evaluation function for environmental temperature, and a bridge health degree evaluation function for environmental humidity;

[0080] Specifically, the bridge health degree evaluation model includes:

[0081] Health = f WS (WS)·f WD (WD)·f I (I)·f D (D)·f T (T)·f H (H)

[0082] where Health is the first bridge health degree value, f WS (WS) is the bridge health degree evaluation function for environmental wind speed, WS is the environmental wind speed monitored by Beidou, f WD (WD) is the bridge health degree evaluation function for wind direction, WD is the wind direction monitored by Beidou, f I (I) is the bridge health degree evaluation function for the inclination degree of the bridge, I is the inclination degree of the bridge monitored by Beidou, f D (D) is the bridge health degree evaluation function for the displacement of the bridge, D is the displacement of the bridge monitored by Beidou, f T (T) is the bridge health degree evaluation function for environmental temperature, T is the environmental temperature monitored by Beidou, f H (H) is the bridge health degree evaluation function for environmental humidity, H is the environmental humidity monitored by Beidou.

[0083] Specifically, the bridge health degree evaluation function f WS (WS) includes:

[0084]

[0085] where k is the first wind speed adjustment factor, WS max is the maximum wind speed that the bridge design can withstand, and c is the second wind speed adjustment factor.

[0086] Specifically, the bridge health assessment function f WD (WD) of the wind direction includes:

[0087]

[0088] where WD opt is the optimal wind direction angle.

[0089] Specifically, the bridge health assessment function f I (I) of the inclination degree of the bridge includes:

[0090]

[0091] where α is the first inclination degree adjustment factor, I max is the maximum allowable inclination of the bridge design, and β is the second inclination degree adjustment factor.

[0092] Specifically, the bridge health assessment function f D (D) of the displacement of the bridge includes:

[0093]

[0094] where γ is the first displacement adjustment factor, D max is the maximum allowable displacement of the bridge design, and δ is the second displacement adjustment factor.

[0095] Specifically, the bridge health assessment function f T (T) of the environmental temperature includes:

[0096]

[0097] where T ref is the reference temperature, T max is the highest temperature that the bridge can withstand, and T min is the lowest temperature that the bridge can withstand.

[0098] Specifically, the bridge health assessment function f H (H) of the environmental humidity includes:

[0099]

[0100] where θ is the humidity adjustment factor, H ref is the reference humidity, μ is the first time adjustment factor, ω is the second time adjustment factor, and t is the humidity acquisition time.

[0101] An evaluation module is configured to input the bridge information into a long short-term memory neural network, calculate a second bridge health value, and complete the evaluation of the bridge health in combination with the first bridge health value.

[0102] Specifically, the average value of the first bridge health value and the second bridge health value is used as the final bridge health value, and the evaluation of the bridge health is carried out.

[0103] Embodiment 3

[0104] The embodiment of the present invention further provides a storage medium storing multiple instructions for implementing the bridge health Beidou monitoring method based on long short-term memory learning.

[0105] Optionally, in this embodiment, the above storage medium may be located in any one of the computer terminals in the computer terminal group in the computer network, or in any one of the mobile terminals in the mobile terminal group.

[0106] Optionally, in this embodiment, the storage medium is configured to store program codes for executing the steps of Embodiment 1.

[0107] Embodiment 4

[0108] The embodiment of the present invention further provides an electronic device including a processor and a storage medium connected to the processor. The storage medium stores multiple instructions that can be loaded and executed by the processor so that the processor can execute a bridge health Beidou monitoring method based on long short-term memory learning.

[0109] Specifically, the electronic device in this embodiment may be a computer terminal, and the computer terminal may include: one or more processors and a storage medium.

[0110] Among them, the storage medium can be used to store software programs and modules, such as the bridge health Beidou monitoring method based on long short-term memory learning in the embodiment of the present invention, and the corresponding program instructions / modules. The processor runs the software programs and modules stored in the storage medium to execute various functional applications and data processing, that is, to implement the above-mentioned bridge health Beidou monitoring method based on long short-term memory learning. The storage medium may include a high-speed random access storage medium, and may also include a non-volatile storage medium, such as one or more magnetic storage systems, flash memory, or other non-volatile solid-state storage media. In some instances, the storage medium may further include a storage medium remotely set relative to the processor, and these remote storage media can be connected to the terminal through a network. Examples of the above network include but are not limited to the Internet, enterprise intranet, local area network, mobile communication network and their combinations.

[0111] The processor can execute the steps of Embodiment 1 by transmitting the information and application programs stored in the storage medium through system calls.

[0112] Embodiment 5

[0113] For the monitoring of bridge health, large-span distributed sensors are used to monitor the bridge vibration conditions, wind speed, wind direction, inclination, displacement, temperature, humidity and other conditions all day and all weather. Using the Beidou post-differential positioning technology and multi-sensor fusion monitoring technology, multi-source health data is monitored, and the monitoring data is transmitted back to the data center. A reconstructed bidirectional long short-term neural network is used for modeling to predict the monitoring data. The entire system consists of an automated bridge monitoring network, an intelligent data processing center and an early warning system. The system structure is as Figure 3 shown:

[0114] 1. Beidou Automated Acquisition System

[0115] The bridge monitoring network is arranged in the structure of reference point - working base point - monitoring point, and consists of a Beidou ground-based augmentation station, a Beidou communication base station and bridge sensor monitoring points. The reference point gives the reference system of the bridge position and consists of a Beidou ground-based augmentation station and a Beidou communication base station. The working base point connects the reference point and the sensor observation point. Near the bridge building, the point position is stable during the observation period, and the point position is observed in real time by the reference point. The sensor observation point is buried on the bridge body and can reflect the vibration characteristics of the bridge. The bridge condition can be judged according to its vibration amplitude, displacement change and other characteristics. The system is based on Beidou monitoring equipment and collects monitoring data in real time.

[0116] Based on the Beidou automated monitoring platform, the positioning solution results and real-time monitoring results of bridge sensors are collected and transmitted to the data processing center. The high-precision Beidou automated monitoring system consists of three parts: data acquisition, data solution and data transmission. The acquisition, solution and transmission of bridge monitoring data are completed by a local area network. The system structure is as Figure 4 shown:

[0117] The data acquisition work of the monitoring system is completed at the reference station and the sensor monitoring point. High-precision multi-frequency receivers and sensors are respectively fixedly installed at the reference station and the monitoring station, running continuously all day and all weather. All data is transmitted to the control center through transmission optical fibers (or GPRS / CDMA wireless networks).

[0118] When the sensors at the reference station and the monitoring point are observing, the satellite data received by the receiver can be automatically transmitted to the server, and at the same time, the observation values such as pseudo-range and carrier phase and information such as broadcast ephemeris are automatically stored in the memory of the receiver. Once an abnormal situation occurs in a receiver, the monitoring system will provide an alarm message in real time.

[0119] The monitoring stations and reference stations transmit the data of each monitoring station to the cloud platform in real time through long-distance communication. At the same time, the data (observations, satellite ephemerides, etc.) collected by each of the above receivers are transmitted to the server at regular time intervals set by the cloud platform. Operators can monitor the working conditions of each Beidou receiver at the reference station and monitoring points in real time from the cloud platform and issue relevant instructions to control each receiver. The workstations and microcomputers in the cloud platform can perform data analysis work at regular time intervals set in advance.

[0120] The real-time monitoring system based on multi-sensors consists of three parts: a partial data acquisition unit, a data transmission unit, and a data processing unit (control center). The data acquisition unit consists of various sensors and acquisition instruments. In bridge data monitoring, it can be divided into displacement monitoring sensors, settlement monitoring sensors, and tilt angle monitoring sensors. The sensors used in the automatic online monitoring system convert the measured non-electrical quantities into digital output signals (including direct and indirect conversions). The acquisition instrument collects the output signals of each sensor at certain intervals and transmits them to the control center through the transmission unit. The control center receives the deformation data transmitted by multiple acquisition units, processes and analyzes them, dynamically monitors the real-time data of the bridge body, and issues early warnings in a timely manner when abnormalities are found. The control center can also send instructions to the acquisition instrument to control multiple sensors.

[0121] The Beidou system and the sensor monitoring values are jointly networked and processed. The Beidou observations and the sensor observations are jointly adjusted to ensure the accuracy of the observations and improve the accuracy of the observation results. First, the Beidou and sensor observations are reduced to the reference ellipsoid surface according to a certain mathematical relationship. On the reference ellipsoid surface, the error equations and fixed quantity constraint equations for the joint adjustment of the two types of data are established, the stochastic processing model is determined, and the adjustment is carried out to obtain the result data, and the accuracy error is evaluated.

[0122] The present invention adopts the mode of joint operation of Beidou and intelligent sensors to build an automated deformation software and hardware monitoring system, which collects and uploads the Beidou observations and the monitoring station observations in real time and performs cloud computing. It provides services such as online display of the monitoring project results, query of the calculation results, report statistics, and data download, and can realize the comprehensive and real-time monitoring and early warning of the uneven settlement of the surrounding environment and buildings of the bridge. The monitoring scheme of this system gives full play to the advantages of Beidou in large-area continuity and real-time performance, as well as the stable and reliable high-precision characteristics of sensors within the working radius. While ensuring the monitoring accuracy, it improves the monitoring efficiency and safety.

[0123] 2. Intelligent Data Processing Center

[0124] The data processing center is the core of the Beidou automated monitoring system, mainly realizing the collection of data transmitted back by the bridge monitoring network and the intelligent processing of monitoring data, achieving the reception and storage of original observation values, data decoding, preprocessing, and high-precision positioning calculation, receiving and fusing multi-sensor monitoring data, and obtaining the real-time monitoring results of monitoring points. The real-time high-precision positioning results are used for monitoring the transient (large deformation changes within a short period) conditions of bridge monitoring points.

[0125] Data processing is completed in the control center. The cloud platform (control center) has hardware devices such as a master control computer, storage backup devices, printers, etc. This is the core of the Beidou automated monitoring system, consisting of four modules: data aggregation, data processing, data modeling, and data transmission.

[0126] As Figure 5 shown, the core algorithm of the data processing module adopts the variational autoencoder - bidirectional long short - term memory neural network (VAE - BiLSTM). This module consists of four parts: data preprocessing - network model construction - data learning - test output for the processed monitoring data. The data preprocessing stage includes two steps: standardization and data segmentation. The standardization formula is as follows, where x represents the input sequence, μ represents the mean of the input sequence, σ represents the standard deviation of the sequence, represents the standardized sequence;

[0127]

[0128] The data segmentation part prepares the input for model training, dividing the training set data into three - dimensional data that can be input into the model. Network model construction is the core part of the algorithm. This module uses a reconstructed bidirectional long short - term network to model the training set data. The long short - term memory neural network (LSTM) stores the monitored long - term sequence information through the cell memory state and the hidden state. The cell structure at each time step mainly includes three gates: the input gate, the output gate, and the forget gate, and two states: the cell state and the hidden state. The forget gate decides to selectively forget some information from the memory unit. The output value is between 0 and 1. When the output value is 0, it means all information is forgotten and the information is inhibited from passing through. When the output value is 1, it means all information is retained. The subscript t represents the moment, the subscript f represents the forget gate parameter, σ represents the sigmoid function, h t-1 represents the hidden state of the previous moment, x t represents the current input sequence, W f is the weight matrix of the forget gate, b f is the bias term of the forget gate, f t represents the output value of the forget gate;

[0129] f t =σ(W f ·[h t-1 ,x t+b f )

[0130] The input gate determines the selective update to the current cell state. The output value ranges from 0 to 1, where 0 means completely blocking the value from entering the cell state and 1 means completely allowing it to enter. The subscript t represents the time step, the subscript i represents the input gate parameter, σ represents the sigmoid function, i t represents the output value of the input gate, x t represents the current input sequence, W i is the weight matrix of the input gate, h t-1 represents the hidden state at the previous time step, b i is the bias term of the input gate;

[0131] i t = σ(W i · [h t-1 , x t + b i )

[0132] The cell candidate value is used to store the information at the current time step, ranging from -1 to 1. The subscript t represents the time step, the subscript c represents the cell candidate value parameter, tanh is the hyperbolic tangent function, σ represents the sigmoid function, c t represents the cell candidate value, W c is the weight matrix of the cell candidate value, b c is the bias term of the cell candidate value, h t-1 represents the hidden state at the previous time step, x t represents the current input sequence. Combine the values of the forget gate and the input gate as the updated cell state, denoted by C t ;

[0133] c t = tanh(W c · [h t-1 , x t + b c )

[0134] C t = f t · C t-1 + i t · c t

[0135] The output gate inputs the current input and the hidden state at the previous time step into a fully connected layer, and applies the sigmoid function to obtain a value between 0 and 1. The subscript t represents the time step, the subscript o represents the fully connected layer parameter, tanh is the hyperbolic tangent function, W o is the weight matrix of the fully connected layer, b o is the bias term of the fully connected layer, x tDenote the current input sequence as o t Denote the output value. Obtain a value between -1 and 1 for the current cell state C t through the tanh function and multiply it by the value of the output gate o t to get the final hidden state h t :

[0136] o t = σ(W o [h t-1 , x t +b o )

[0137] h t = o t ·tanh(C t )

[0138] To address noise and other perturbation information in the monitoring data, the bidirectional long short-term network is combined with the variational autoencoder (VAE). The Encoder part encodes the input sequence x and samples a low-dimensional vector z from the prior probability distribution p(z), and adds a constraint to z to make it satisfy the normal distribution, and then generates data through the posterior distribution p(x|z). By comparing the error between the input sequence x and the output sequence , through the training of the network model, the error gradually decreases. During this process, the Bayesian formula is used to calculate the latent distribution p(z|x) as the true distribution, but this distribution is difficult to obtain, so the approximate distribution output by the Encoder model is used to fit p(z|x). The formula is as follows:

[0139]

[0140] p(x) = ∫p(x|z)p(z)dz

[0141] To ensure the continuous transmission of gradients and perform random sampling from the standard normal distribution to ensure the backpropagation of the model, VAE selects the reparameterization method and obtains random sampling using the mean μ and variance σ output by the Encoder network. ε represents the sampling value. The formula is as follows:

[0142] z = μ + ε × σ

[0143] The model optimizer selects Adam. The loss function of the network model consists of two parts. The formula is as follows, the mean squared error loss function (the first part) and the KL divergence (the second part), where f(x) represents the model output value, y represents the true value, μ represents the mean of the network output, and σ represents the variance of the network output. When the value of the loss function satisfies the set threshold, the model stops training and is saved.

[0144]

[0145] 3. Abnormal warning mechanism

[0146] The bridge body is evenly divided into several areas according to the spatial distribution of sensors, so as to facilitate the staff to check the abnormal sections. The number position information and monitoring status of the sensors in each area will be transmitted back to the data center through the monitoring system for summary. Once the network model predicts the pre-abnormal state or the abnormal state has occurred in a certain sequence, the monitoring devices in the corresponding area will be immediately matched, and the operators will be notified in real time to conduct safety inspections.

[0147] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0148] In the above embodiments of the present invention, the descriptions of each embodiment have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0149] In the several embodiments provided by the present invention, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the system embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of units or modules can be in an electrical or other form.

[0150] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0151] In addition, the functional units in each embodiment of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0152] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, read-only storage media (ROM, Read-Only Memory), random access storage media (RAM, Random Access Memory), mobile hard disks, magnetic disks, or optical discs.

[0153] Obviously, the above-mentioned embodiments are merely examples for clear illustration and not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all the implementation manners here. The obvious changes or modifications derived therefrom still fall within the protection scope of the present invention.

Claims

1. A Beidou monitoring method for the health of bridges based on long short-term memory learning, characterized in that, Including: Obtain bridge information of the bridge through the Beidou system, where the bridge information includes: environmental wind speed monitored by Beidou, wind direction monitored by Beidou, inclination degree of the bridge monitored by Beidou, displacement of the bridge monitored by Beidou, environmental temperature monitored by Beidou, and environmental humidity monitored by Beidou; Set up a bridge health evaluation model, and calculate a first bridge health value according to the bridge information, where the bridge health evaluation model includes: a bridge health evaluation function for environmental wind speed, a bridge health evaluation function for wind direction, a bridge health evaluation function for the inclination degree of the bridge, a bridge health evaluation function for the displacement of the bridge, a bridge health evaluation function for environmental temperature, and a bridge health evaluation function for environmental humidity; The bridge health evaluation model includes: Health=f WS (WS)·f WD (WD)·f I (I)·f D (D)·f T (T)·f H (H) Among them, Health is the first bridge health value, f WS (WS) is the bridge health assessment function of the environmental wind speed, WS is the environmental wind speed monitored by Beidou, f WD (WD) is the bridge health assessment function of the wind direction, WD is the wind direction monitored by Beidou, f I (I) is the bridge health assessment function of the bridge's inclination degree, I is the bridge's inclination degree monitored by Beidou, f D (D) is the bridge health assessment function of the bridge's displacement, D is the bridge's displacement monitored by Beidou, f T (T) is the bridge health assessment function of the environmental temperature, T is the environmental temperature monitored by Beidou, f H (H) is the bridge health assessment function of the environmental humidity, H is the environmental humidity monitored by Beidou; Input the bridge information into a long short-term memory neural network, calculate a second bridge health value, and complete the evaluation of the bridge health in combination with the first bridge health value.

2. The Beidou monitoring method for bridge health based on long- and short-term memory learning according to claim 1, characterized in that The bridge health assessment function f of the ambient wind speed WS (WS) includes: where t is the first wind speed adjustment factor, WS max is the maximum wind speed that the bridge design can withstand, and c is the second wind speed adjustment factor.

3. The Beidou monitoring method for bridge health degree with long- and short-term memory learning as described in claim 1, characterized in that, The bridge health assessment function f of the wind direction WD (WD) includes: Among them, WD opt is the optimal wind direction angle.

4. The Beidou monitoring method for bridge health degree with long and short-term memory learning as claimed in claim 1, wherein The bridge health assessment function f for the inclination degree of the bridge I (I) includes: Among them, α is the first inclination degree adjustment factor, and I max is the maximum allowable inclination of the bridge design, and β is the second inclination degree adjustment factor.

5. The Beidou monitoring method for bridge health based on long- and short-term memory learning as claimed in claim 1, wherein The bridge health assessment function f of the displacement of the bridge D (D) includes: where γ is the first displacement adjustment factor, D max is the maximum allowable displacement for bridge design, and δ is the second displacement adjustment factor.

6. The Beidou monitoring method for bridge health degree with long- and short-term memory learning as described in claim 1, characterized in that, The bridge health assessment function f T (T) includes: Among them, T ref is the reference temperature, T max is the highest temperature that the bridge can withstand, and T min is the lowest temperature that the bridge can withstand.

7. The Beidou monitoring method for bridge health based on long and short-term memory learning according to claim 1, characterized in that The bridge health assessment function f H of the environmental humidity includes: Among them, θ is the humidity adjustment factor, H ref is the reference humidity, μ is the first time adjustment factor, ω is the second time adjustment factor, and t is the humidity acquisition time.

8. The Beidou monitoring method for bridge health degree with long- and short-term memory learning as described in claim 1, characterized in that, Including: Take the average value of the first bridge health value and the second bridge health value as the final bridge health value, and conduct an evaluation of the bridge health.

9. A Beidou monitoring system for bridge health based on long short-term memory learning, characterized in that, Including: An information acquisition module, which is used to obtain bridge information of the bridge through the Beidou system, where the bridge information includes: environmental wind speed monitored by Beidou, wind direction monitored by Beidou, inclination degree of the bridge monitored by Beidou, displacement of the bridge monitored by Beidou, environmental temperature monitored by Beidou, and environmental humidity monitored by Beidou; A module for calculating the first bridge health value, which is used to set up a bridge health evaluation model and calculate the first bridge health value according to the bridge information, where the bridge health evaluation model includes: a bridge health evaluation function for environmental wind speed, a bridge health evaluation function for wind direction, a bridge health evaluation function for the inclination degree of the bridge, a bridge health evaluation function for the displacement of the bridge, a bridge health evaluation function for environmental temperature, and a bridge health evaluation function for environmental humidity; The bridge health evaluation model includes: Health=f WS (WS)·f WD (WD)·f I (I)·f D (D)·f T (T)·f H (H) Among them, Health is the first bridge health value, f WS (WS) is the bridge health assessment function of the environmental wind speed. WS is the environmental wind speed monitored by Beidou, f WD (WD) is the bridge health assessment function of the wind direction. WD is the wind direction monitored by Beidou, f I (I) is the bridge health assessment function of the inclination degree of the bridge. I is the inclination degree of the bridge monitored by Beidou, f D (D) is the bridge health assessment function of the displacement of the bridge. D is the displacement of the bridge monitored by Beidou, f T (T) is the bridge health assessment function of the environmental temperature. T is the environmental temperature monitored by Beidou, f H (H) is the bridge health assessment function of the environmental humidity. H is the environmental humidity monitored by Beidou; An evaluation module, which is used to input the bridge information into a long short-term memory neural network, calculate a second bridge health value, and complete the evaluation of the bridge health in combination with the first bridge health value.

Images