A bridge temperature monitoring data reconstruction method and system

Abstract

The present invention relates to the field of data processing technology, and in particular to a method and system for reconstructing bridge temperature monitoring data. The method comprises the following steps: obtaining the original structural temperature data of the bridge through monitoring equipment, and eliminating the time lag effect of the original structural temperature data; using a data decomposition model to decompose the structural temperature data to obtain a structural temperature trend item; constructing a bridge temperature monitoring data reconstruction model, constructing a data set according to the number of channels and the structural temperature trend item, and training and verifying the bridge temperature monitoring data reconstruction model based on the data set; evaluating the trained bridge temperature monitoring data reconstruction model, and determining the optimal number of channels for the data set in combination with the evaluation results; and completing the reconstruction of the bridge temperature monitoring data based on the optimal number of channels and real-time monitoring data. The present invention determines the decomposition quantity using the structural temperature and the ambient temperature, and then constructs a data set according to multi-channel correlation to train the data reconstruction model, thereby solving the problem of high-precision reconstruction of bridge temperature monitoring data.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for reconstructing bridge temperature monitoring data. Background Art

[0002] Data integrity is crucial in structural health monitoring. High-quality data can be used for condition assessment, damage identification and maintenance decisions. However, during long-term monitoring, the complex service environment causes a lot of noise and fluctuations in the data, and data loss caused by sensor failure or transmission failure occurs frequently.

[0003] Temperature is an important environmental factor in bridge health monitoring. The temperature effect of bridge structure is significant during operation and may even cover various load responses. The above problems may cause false alarms in the bridge monitoring system or incorrectly evaluate the service status of the structure, causing serious economic losses and social impacts. Therefore, it is necessary to propose an effective method to reconstruct temperature data. Summary of the invention

[0004] In view of the shortcomings of existing methods and the needs of practical applications, in order to improve the temperature data reconstruction capability and solve the problem of high-precision reconstruction of bridge temperature monitoring data. On the one hand, the present invention provides a bridge temperature monitoring data reconstruction method, comprising the following steps:

[0005] The original structural temperature data of the bridge is obtained through monitoring equipment, and the time lag effect of the original structural temperature data is eliminated; the structural temperature data is decomposed to obtain the structural temperature trend item by using the data decomposition model; a bridge temperature monitoring data reconstruction model is constructed, and a data set is constructed according to the number of channels and the structural temperature trend item, and the bridge temperature monitoring data reconstruction model is trained and verified based on the data set; the trained bridge temperature monitoring data reconstruction model is evaluated, and the optimal number of channels of the data set is determined in combination with the evaluation results; the bridge temperature monitoring data reconstruction is completed based on the optimal number of channels and real-time monitoring data. The present invention determines the decomposition quantity by using the structural temperature and the ambient temperature to realize the physical-guided temperature trend recognition, and then considers the correlation of the sensor data at different spatial positions of the same section, that is, the multi-channel correlation to construct a data set to train the data reconstruction model, and then reconstruct the temperature data of the fault channel, and determines the accuracy of the reconstruction result of the bridge temperature monitoring data through a variety of evaluation methods, which effectively solves the problem of high-precision reconstruction of bridge temperature monitoring data.

[0006] Optionally, the bridge temperature monitoring data reconstruction method eliminates the time lag effect of the original structural temperature data, comprising the following steps:

[0007] An ambient temperature matrix is ​​constructed based on ambient temperature; a structural temperature matrix is ​​constructed based on structural temperature; an offset based on ambient temperature is obtained according to the correlation between the ambient temperature matrix and the structural temperature matrix; and the time lag effect of the original structural temperature data is eliminated by using the offset. The present invention helps to improve the accuracy of monitoring data by eliminating the time lag effect of temperature data, provides a more reliable basis for structural health monitoring, and further improves the accuracy of data reconstruction of the present invention.

[0008] Optionally, the correlation is calculated to satisfy the following formula:

[0009] Among them, r represents the degree of correlation, x represents the structural temperature data, v represents the ambient temperature data, σ x represents the variance of the structural temperature data, σ v The correlation calculation involved in the present invention is simple and has high applicability, which further improves the calculation efficiency of the present invention.

[0010] Optionally, the data decomposition model is a variational modal decomposition model. The data decomposition model provided by the present invention can effectively separate data of different time scales, easily specify the number of modes for decomposition, and is conducive to obtaining an optimal decomposition model.

[0011] Optionally, the step of utilizing a data decomposition model to decompose the structural temperature data to obtain a structural temperature trend item comprises the following steps:

[0012] The variation layer number of the data decomposition model is set, the structural temperature data is decomposed, and multiple first components are obtained; according to the correlation between the first component and the ambient temperature, the optimal decomposition layer number of the data decomposition model is evaluated, and then the structural temperature trend item is obtained. The present invention sets multiple variation layer numbers for decomposition respectively, and then uses the correlation for evaluation to obtain the optimal decomposition layer number, which is simple in calculation and has strong applicability.

[0013] Optionally, the bridge temperature monitoring data reconstruction model is a temporal convolutional network; the present invention is based on a convolutional neural network, and enhances learning performance by expanding the receptive field and stacking residual blocks, which can effectively capture the dependencies between time series, avoid gradient diffusion and parallel calculation output, and further improve the data reconstruction efficiency of the present invention.

[0014] Optionally, constructing a data set according to the number of channels and the structural temperature trend item comprises the following steps:

[0015] The channels are sorted according to their correlation and multiple channel numbers are set; based on the sorting result and the channel numbers, multiple data sets are constructed using the structural temperature trend items. The present invention improves data learning capability and further improves data reconstruction accuracy by selecting the optimal channel number.

[0016] Optionally, the evaluation trained bridge temperature monitoring data reconstruction model satisfies the following formula:

[0017]

[0018] Among them, MAE represents the mean absolute error, m represents the number of missing data, O i represents the i-th original data, P i represents the i-th reconstructed data, RMSE represents the root mean square error, R 2 represents the coefficient of determination, The present invention evaluates the training model through multiple evaluation models, which is conducive to obtaining the optimal number of channels and further conducive to improving the efficiency of the present invention.

[0019] Optionally, the bridge temperature monitoring data reconstruction method further includes introducing an evaluation model to evaluate the reconstructed data;

[0020] The evaluation model includes autocorrelation function, power spectrum density and probability density function. The present invention evaluates the reconstructed data through multiple evaluation functions, which ensures the rationality of the data and further improves the accuracy of the present invention.

[0021] In the second aspect, in order to efficiently execute the bridge temperature monitoring data reconstruction method provided by the present invention, the present invention also provides a bridge temperature monitoring data reconstruction system, including a processor, an input device, an output device and a memory, wherein the processor, the input device, the output device and the memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the bridge temperature monitoring data reconstruction method as described in the first aspect of the present invention. The bridge temperature monitoring data reconstruction system of the present invention has a compact structure and stable performance, and can stably execute the bridge temperature monitoring data reconstruction method provided by the present invention, further improving the overall applicability and practical application capabilities of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flow chart of a bridge temperature monitoring data reconstruction method provided by an embodiment of the present invention;

[0023] Figure 2 A layout diagram of a bridge temperature monitoring device provided by an embodiment of the present invention;

[0024] Figure 3 A framework diagram of a bridge temperature monitoring data reconstruction system provided by an embodiment of the present invention;

[0025] Figure 4A schematic diagram of the structure of a bridge temperature monitoring data reconstruction device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are only for illustration and are not intended to limit the present invention. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is obvious to those of ordinary skill in the art that these specific details do not need to be adopted to implement the present invention. In other examples, in order to avoid confusing the present invention, known circuits, software or methods are not specifically described.

[0027] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "in one embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or subcombination. In addition, it should be understood by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale.

[0028] See also Figure 1 In order to solve the problem of high-precision reconstruction of bridge temperature monitoring data, the present invention provides a bridge temperature monitoring data reconstruction method, such as Figure 1 As shown, the method comprises the following steps:

[0029] S1. Acquire original structural temperature data of the bridge through monitoring equipment, and eliminate time lag effect of the original structural temperature data.

[0030] Specifically, according to the bridge structure and temperature monitoring needs, multiple temperature monitoring sensors are deployed in the bridge monitoring section. Figure 2In a specific embodiment, the Egongyan rail-specific bridge is a portal double-tower suspension bridge with a total length of 1650.5 meters, a main bridge length of 1120 meters, a bridge deck width of 22 meters, a main beam of a steel beam-concrete beam hybrid structure, and a main pier of a reinforced concrete structure. Among them, the five-span continuous steel box girder self-anchored suspension bridge with a main span of 600 meters has the largest span in the world among similar bridges. The whole bridge has a total of 16 monitoring sections, each section is equipped with 8 structural temperature measurement points, and an ambient temperature measurement point is arranged in the mid-span section. Taking the 9 sensors in the mid-span section of the main bridge as an example, 8 of them are inside the box girder and 1 is outside the box girder. The sensors in the box girder are numbered from top to bottom as #1 to #8. The temperature data obtained is the original structural temperature data of the bridge, and the temperature data obtained by the sensor outside the box girder is called the ambient temperature.

[0031] It should be understood that due to the different thermal conductivity coefficients of different materials or different thermal boundary conditions, the temperature change of the structure lags behind the ambient temperature. This is because the bridge deck pavement covers the heat, which requires more time to transfer to the steel box girder, and because it is in the shadow surface, the time lag effect is more obvious at the bottom of the steel box girder; when cooling, the air inside the steel box girder does not circulate, so the internal temperature drops slowly and still lags behind the ambient temperature. Therefore, the structural temperature lags behind the ambient temperature as a whole. Furthermore, after obtaining the original structural temperature data of the bridge, the time lag effect of the original structural temperature data is eliminated to obtain the structural temperature data, which includes the following steps:

[0032] An ambient temperature matrix is ​​constructed based on the ambient temperature; a structural temperature matrix is ​​constructed based on the structural temperature; an offset based on the ambient temperature is obtained according to the correlation between the ambient temperature matrix and the structural temperature matrix; and the time lag effect of the original structural temperature data is eliminated by using the offset.

[0033] Specifically, according to the monitoring frequency, the number of monitoring points and the monitoring time, the corresponding monitoring data is obtained, and the correlation matrix is ​​established based on the monitoring data. In this embodiment, there are 9 sensors, the sampling time is 61 days from May to June, the sampling interval is 5 minutes, the number of sampling times per day is 288 times, and the data length is 61*288=17568, then the ambient temperature matrix is ​​a (17568,1) matrix, and the structural temperature matrix is ​​a (17568,8) matrix.

[0034] Create an empty matrix A to place the correlation coefficient between the structural temperature and the ambient temperature after each offset. The correlation coefficient between the structural temperature and the ambient temperature is used as an evaluation index. When the offset is 0, the correlation coefficient is the original ambient temperature and the structural temperature, and the correlation coefficient is stored in the first row of the matrix A. When the offset is 1, the 2-17568 data of the structural temperature is taken to perform correlation analysis with the 1-17567 data of the ambient temperature, and the correlation coefficient is stored in the second row of the matrix A. Here, it can be understood that the structural temperature is advanced, and the data at the second time point becomes the data at the first time point. The ambient temperature is subtracted by 1 to meet the same data length. Through this cycle, the correlation coefficient of each offset, each structural temperature sensor and the ambient temperature sensor is stored in the A matrix. Find the largest correlation coefficient and the corresponding index value in the matrix A. The corresponding index value minus 1 can get the offset of each sensor. After determining the offset, offset the data collected by each sensor to eliminate the time lag effect with the ambient temperature. It is understandable that the method of advancing the structural temperature time series, for example, advancing the structural temperature one hour later to the current time, to match the change in ambient temperature is simple and practical to eliminate the time lag effect by translating the structural temperature data in time series, and can quickly and accurately solve the time lag effect problem.

[0035] Taking the data from May 1, 2023 to May 31, 2023 as an example, the structural temperature data collected by each sensor in the beam is offset in time based on the ambient temperature. The reverse time offset of the structural temperature is positive, and the forward time offset is negative. The offset of each sensor is determined based on the maximum correlation coefficient between the temperature of each measuring point and the ambient temperature. The offset and correlation improvement are shown in Table 1.

[0036]

[0037] Table 1 Statistics of sensor offset and correlation improvement

[0038] From Table 1, it can be found that the offsets of #1, #4, and #7 located at the top plate are smaller than those of the other sensors located at the bottom plate. The reason is that the bottom of the box girder is in the shadow surface, and the heat change mostly depends on the heat transfer of the top plate; therefore, the time lag effect at the bottom of the box girder is more significant. After eliminating the time lag effect by this method, the correlation coefficients between the structural temperature and the ambient temperature collected by the 8 sensors in this section are all improved, and the correlation coefficients between the structural temperature and the ambient temperature collected by the sensors at the bottom plate are all increased by more than 10%, among which the #3 sensor has the most obvious improvement, which is increased by 14.09%.

[0039] In an optional embodiment, the correlation between the temperature data is calculated to satisfy the following formula:

[0040] Among them, r represents the degree of correlation, x represents the structural temperature data, v represents the ambient temperature data, σ x represents the variance of the structural temperature data, σ v Represents the ambient temperature data variance.

[0041] S2. Decompose the structural temperature data using a data decomposition model to obtain a structural temperature trend item.

[0042] Specifically, the data decomposition model is a variational mode decomposition model.

[0043] The variational mode decomposition model is a new type of completely non-recursive signal decomposition and estimation method based on the frequency domain. It combines the classic Wiener filtering, Hilbert transform and frequency mixing. Through iterative optimization and reverse error correction, the inherent intrinsic mode function component is defined as an amplitude modulation-frequency modulation signal. For the input signal x(t), the w-th order component u is decomposed. w (t) can be expressed as: w (t) = A w (t)cos(φ w (t)), where A w (t)≥0 is the instantaneous amplitude, φ w (t) is the instantaneous phase, and its derivative function ω w (t) = φ′ w (t) ≥ 0 is the instantaneous center frequency.

[0044] Using Hilbert transform, u w (t) is converted into a complex domain signal. After the mixing operation, the component bandwidth is estimated by calculating the L2 norm of the baseband signal. The constrained variational problem of the original signal decomposition is described as: ||

[0045]

[0046] Among them, {u w}={u1,…,u W} is the set of modal components, {ω w}={ω1,…,ω W} is the set of center frequencies of the corresponding modal components; represents the partial derivative with respect to time t, is the sum of all modes and δ(t) is an impulse function.

[0047] In order to obtain the optimal solution to the variational problem, the quadratic penalty factor α and the Lagrangian multiplication operator λ are introduced to transform the above problem into an unconstrained variational problem augmented Lagrangian operator As shown below:

[0048]

[0049] The alternating direction multiplier optimization algorithm is used to solve the unconstrained variational problem, which can transform λ(t), u in the frequency domain. w (t) and ω w The expression is updated to:

[0050]

[0051]

[0052] in, and are the frequency domain expressions of f(t), u(t) and λ(t) after Fourier transform, and ε is the noise tolerance parameter, which is usually 0.

[0053] Finally, the iterative convergence criterion can be expressed as:

[0054]

[0055] Where η is the convergence threshold when the input signal is applied.

[0056] Further, using the data decomposition model, the structural temperature data is decomposed to obtain the structural temperature trend item, including the following steps:

[0057] Setting the number of variation layers of the data decomposition model, decomposing the structural temperature data, and obtaining a plurality of first components;

[0058] According to the correlation between the first component and the ambient temperature, the optimal decomposition layer number of the data decomposition model is evaluated to obtain the structural temperature trend item.

[0059] Specifically, the number of decomposition layers is first initialized to 1, the structural temperature data to be decomposed is input, and the correlation coefficient between the structural temperature data and the ambient temperature data is calculated.

[0060] Repeat the following steps:

[0061] Increase the number of decomposition layers to obtain the corresponding components, and calculate the correlation coefficient between the first component and the ambient temperature. The first component here is the structural temperature trend term. The structural temperature data is in a strip shape, which makes it difficult to capture the real temperature change. The first component is the real temperature change identified from the structural temperature data, that is, the trend term, which eliminates the influence of various types of noise during data collection and transmission. The number of decomposition layers is determined when the correlation coefficient between the first component and the ambient temperature is the largest.

[0062] S3. Construct a bridge temperature monitoring data reconstruction model, construct a data set according to the number of channels and the structural temperature trend item, and train and verify the bridge temperature monitoring data reconstruction model based on the data set.

[0063] In an embodiment, the bridge temperature monitoring data reconstruction model is a temporal convolutional network.

[0064] Traditional neural networks cannot process in parallel on a large scale. The temporal convolutional network model is based on convolutional neural networks and enhances its performance by expanding the receptive field and stacking residual blocks. It has many advantages, including effectively capturing the dependencies between time series, avoiding gradient diffusion, and parallel computing outputs. Its core components include causal convolution, dilated convolution, and residual blocks.

[0065] To ensure that the length of network input and output is the same and that future data will not leak into the past, the temporal convolutional network uses a one-dimensional fully convolutional neural network as the basic architecture of the network, and introduces a causal convolutional neural network on this basis. In order to adapt to various types of sequences and obtain a larger perceptual field of view, an expansion factor is added to the causal convolution to obtain a larger range of sequence information with fewer layers. For a one-dimensional time series x(t) = {x1, x2, ..., x t}, filter {f1,f2,...,f n-1}, in x s The dilated convolution formula at is:

[0066]

[0067] Among them, * is the convolution operation, d is the expansion factor, which grows exponentially (d=2i), n is the filter size, x s-d·i is a forward connection sequence, so the expansion is equivalent to introducing a fixed step size between every two adjacent filters. When d = 1, the dilated convolution degenerates into a regular convolution. Using a larger expansion can make the output of the top layer represent a wider range of inputs, thereby effectively expanding the receptive field. Due to the input-output characteristics of the dilated convolution, each layer has to fill in the value at the leftmost end of the sequence, the number of which is (n-1)d.

[0068] Since the receptive field of the temporal convolutional network depends on the network depth as well as the filter size and dilation factor, the stability of deeper and larger temporal convolutional networks becomes important; as the number of network layers continues to increase, various benefits are brought about, but problems also arise, such as excessive gradient norm, gradient diffusion, and decreased training set accuracy, etc. Residual connections can effectively solve these problems; in the residual block, the temporal convolutional network has two layers of extended causal convolutions and nonlinearities, so a rectified linear unit is used, and for normalization, weight normalization is applied to the convolutional filter; in addition, a spatial dropout is added after each extended convolution for regularization: in each training step, the entire channel is cleared to zero, and in order to explain the difference in input and output widths, an additional 1x1 convolution is used to ensure that element addition ≥ receives tensors of the same shape.

[0069] In the embodiment, the temporal convolutional network model parameters are as follows: the number of filter layers is 5, the kernel size is 2, the dilation factor d (dilations) is 1, 2, 4, 8, 16, 32, the mean square error (MSE) is selected as the loss function of the training process, and the Adam optimizer is used to perform gradient descent optimization on the random objective function of the model, and the learning rate is 0.00001; the network parameters are initialized by glorot_uniform, and according to the size of the sample, the Batch Size is selected as 64, the epochs is 200, and the Kears2.11.0 deep learning framework is used.

[0070] Further, constructing a data set according to the number of channels and the structural temperature trend item includes the following steps:

[0071] Sort by correlation between channels and set multiple channel numbers;

[0072] Based on the sorting result and the number of channels, multiple data sets are constructed using the structural temperature trend items.

[0073] Specifically, assuming that #1 is the damaged channel, data from May 1, 2023 to May 31, 2023 are selected, the sampling frequency is once every 5 minutes, and the length of data collected by each sensor is 8928. The data set is divided into two parts: training set and validation set. The training set contains the first 80% of this data set, and the validation set contains the last 20% of the data set. Considering the spatiotemporal correlation between sensors, the correlation coefficient between sensor #1 and the remaining sensor data is used as an indicator, and the sensors are reordered in descending order of the correlation coefficient. The corresponding sensors are used as corresponding channels to construct multiple channel order data sets.

[0074] S4. Evaluate the trained bridge temperature monitoring data reconstruction model, and determine the optimal number of channels for the data set based on the evaluation results.

[0075] Specifically, the evaluation trained bridge temperature monitoring data reconstruction model satisfies the following formula:

[0076]

[0077] Among them, MAE represents the mean absolute error, m represents the number of missing data, O i represents the i-th original data, P i represents the i-th reconstructed data, RMAE represents the root mean square error, R 2 represents the coefficient of determination, Represents the average value of the original data.

[0078] In the embodiment, the data reconstruction performance evaluation indicators with different numbers of channels are shown in Table 2.

[0079] Number of channels MAE RMSE <![CDATA[R 2 ]]> 1 0.9213 1.1216 0.8026 2 0.9017 1.0869 0.8198 3 0.6560 0.8009 0.9225 4 0.6273 0.7520 0.9319 5 0.6058 0.7277 0.9382 6 0.6495 0.7994 0.9239 7 0.6734 0.8150 0.9234

[0080] Table 2 Results of data reconstruction performance evaluation indicators with different numbers of channels

[0081] It can be seen from Table 1 that the fitting effect gradually improves from the result of 1 channel to the result of 4 channels, and reaches the optimal value when the number of channels is 5.

[0082] Furthermore, an evaluation model is introduced to evaluate the reconstructed data; the evaluation model includes an autocorrelation function, a power spectrum density and a probability density function.

[0083] S5. Reconstructing the bridge temperature monitoring data based on the optimal number of channels and the real-time monitoring data.

[0084] Specifically, the real-time monitoring data corresponding to the determined optimal number of channels is used as the input data of the bridge temperature monitoring data reconstruction model, and the output data is the reconstructed bridge temperature monitoring data.

[0085] See also Figure 3 In an embodiment, in order to efficiently execute a bridge temperature monitoring data reconstruction method provided by the present invention, the present invention also provides a bridge temperature monitoring data reconstruction system, including: an input device, an output device, a processor, and a memory, wherein the input device, the output device, the processor, and the memory are interconnected, and the memory contains program instructions, and the program instructions are used for the steps of the bridge temperature monitoring data reconstruction method. The bridge temperature monitoring data reconstruction system of the present invention has a compact structure and stable performance, and can stably execute a bridge temperature monitoring data reconstruction method of the present invention, further improving the overall applicability and practical application ability of the present invention.

[0086] In an embodiment, the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. An input device may be used to obtain data information. An output device may be used to output the results obtained by storing program instructions contained in a computer program in a memory provided by the present invention. The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory.

[0087] In yet another alternative embodiment, see Figure 4 In order to efficiently execute the bridge temperature monitoring data reconstruction method provided by the present invention, this embodiment also provides a bridge temperature monitoring data reconstruction device, such as Figure 4 As shown, including:

[0088] The memory 10 is used to store computer programs; the processor 20 is used to execute the computer programs to implement the above bridge temperature monitoring data reconstruction method. The memory 10, the processor 20, the communication interface 31 and the communication bus 32. The memory 10, the processor 20 and the communication interface 31 all communicate with each other through the communication bus 32.

[0089] In an embodiment, the memory 10 is used to store one or more program instructions. The memory 10 may store program instructions for implementing the following functions:

[0090] The original structural temperature data of the bridge is obtained through monitoring equipment, and the time lag effect of the original structural temperature data is eliminated; the structural temperature data is decomposed by using a data decomposition model to obtain a structural temperature trend item; a bridge temperature monitoring data reconstruction model is constructed, a data set is constructed according to the number of channels and the structural temperature trend item, and the bridge temperature monitoring data reconstruction model is trained and verified based on the data set; the trained bridge temperature monitoring data reconstruction model is evaluated, and the optimal number of channels for the data set is determined based on the evaluation results; and the bridge temperature monitoring data reconstruction is completed based on the optimal number of channels and real-time monitoring data.

[0091] In one possible implementation, the memory 10 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, and at least one application required for a function, etc.; the data storage area may store data created during use. In addition, the memory 10 may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include NVRAM. The memory stores an operating system and operating instructions, executable modules or data structures, or a subset thereof, or an extended set thereof, wherein the operating instructions may include various operating instructions for implementing various operations. The operating system may include various system programs for implementing various basic tasks and processing hardware-based tasks.

[0092] The processor 20 may be a central processing unit (CPU), an application specific integrated circuit, a digital signal processor, a field programmable gate array or other programmable logic device, and the processor 20 may be a microprocessor or any conventional processor, etc. The processor 20 may call a program stored in the memory 10. The communication interface 31 may be an interface of a communication module, and is used to connect with other devices or systems.

[0093] Of course, it should be noted that Figure 4 The structure shown does not constitute a limitation on the bridge temperature monitoring data reconstruction device in this embodiment. In actual applications, the bridge temperature monitoring data reconstruction device may include Figure 4 More or fewer components than shown, or combinations of certain components.

[0094] The embodiment further provides a storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the above-mentioned bridge temperature monitoring data reconstruction method are implemented.

[0095] The storage medium may include: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0096] In summary, the present invention uses the structural temperature and the ambient temperature to determine the decomposition quantity, realizes the physical-guided temperature trend recognition, and then considers the correlation of the sensor data at different spatial positions of the same section, that is, the multi-channel correlation, to construct a data set to train the data reconstruction model, and then reconstruct the temperature data of the fault channel, and determines the accuracy of the reconstruction result of the bridge temperature monitoring data through a variety of evaluation methods, effectively solving the problem of high-precision reconstruction of bridge temperature monitoring data. Therefore, the present invention effectively overcomes the various shortcomings of the prior art and has a high industrial utilization value.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope recorded in the present invention.

Claims

1. A bridge temperature monitoring data reconstruction method, characterized in that: The bridge temperature monitoring data reconstruction method comprises the following steps: Acquire original structural temperature data of the bridge through monitoring equipment, and eliminate time lag effect of the original structural temperature data; Using the data decomposition model, the structural temperature data is decomposed to obtain the structural temperature trend item; Constructing a bridge temperature monitoring data reconstruction model, constructing a data set according to the number of channels and the structural temperature trend item, and training and verifying the bridge temperature monitoring data reconstruction model based on the data set; Evaluate the trained bridge temperature monitoring data reconstruction model, and determine the optimal number of channels for the data set based on the evaluation results; Reconstructing the bridge temperature monitoring data according to the optimal number of channels and the real-time monitoring data; The step of eliminating the time lag effect of the original structural temperature data comprises the following steps: Construct an ambient temperature matrix based on the ambient temperature; Constructing a structural temperature matrix based on the structural temperature; According to the correlation between the ambient temperature matrix and the structural temperature matrix, an offset based on the ambient temperature is obtained; Using the offset, the time lag effect of the original structural temperature data is eliminated; The correlation is calculated to satisfy the following formula: , in, Indicates the degree of relevance, Represents structural temperature data, Indicates ambient temperature data, represents the variance of the structural temperature data, Represents the variance of ambient temperature data; The data decomposition model is a variational mode decomposition model; The method of using the data decomposition model to decompose the structural temperature data to obtain the structural temperature trend item includes the following steps: Setting the number of variation layers of the data decomposition model, decomposing the structural temperature data, and obtaining a plurality of first components; According to the correlation between the first component and the ambient temperature, the optimal decomposition layer number of the data decomposition model is evaluated to obtain the structural temperature trend item.

2. A bridge temperature monitoring data reconstruction method according to claim 1, characterized in that: The bridge temperature monitoring data reconstruction model is a time convolutional network.

3. A bridge temperature monitoring data reconstruction method according to claim 2, characterized in that: The step of constructing a data set according to the number of channels and the structural temperature trend item comprises the following steps: Sort by correlation between channels and set multiple channel numbers; Based on the sorting result and the number of channels, multiple data sets are constructed using the structural temperature trend items.

4. The bridge temperature monitoring data reconstruction method according to claim 1 is characterized in that: The evaluation trained bridge temperature monitoring data reconstruction model satisfies the following formula: , , ,in, represents the mean absolute error, Indicates the number of missing data. Indicates The original data, Indicates Reconstructed data, represents the root mean square error, represents the coefficient of determination, Represents the average value of the original data.

5. The bridge temperature monitoring data reconstruction method according to claim 1 is characterized in that: It also includes the introduction of an evaluation model to evaluate the reconstructed data; The evaluation model includes an autocorrelation function, a power spectrum density and a probability density function.

6. A bridge temperature monitoring data reconstruction system, characterized in that: The bridge temperature monitoring data reconstruction system includes: an input device, an output device, a processor, and a memory. The input device, the output device, the processor, and the memory are interconnected. The memory includes program instructions, and the program instructions are used to execute the bridge temperature monitoring data reconstruction method according to any one of claims 1 to 5.

Images