A method for correcting deformation values ​​in integrity tests of containment structures based on BP neural networks.

Abstract

This invention discloses a method, apparatus, and equipment for correcting the overall deformation measurement values ​​of a containment structure based on a BP neural network algorithm. It relates to the fields of safety assessment of prestressed concrete structures and machine learning technology. The method includes sampling data during a pre-test and performing first data processing to obtain sample data containing first ambient temperature data, first solar radiation intensity, and first deformation measurement data; training a BP neural network model using the sample data to obtain an optimal BP neural network model; normalizing the measured data during the formal test and inputting it into the optimal BP neural network model for calculation to obtain second deformation prediction data for the formal test; and using the second deformation prediction data to correct the comprehensive deformation measurement values ​​of the containment structure during the formal test to obtain the corrected deformation measurement values ​​of the containment structure under test pressure. This invention has the advantages of high accuracy in containing structure deformation measurement and high practical engineering application value.

Description

Technical Field

[0001] This invention relates to the fields of safety assessment and machine learning technology for prestressed concrete structures, specifically to a method, apparatus, and equipment for correcting the deformation measurement values ​​of a containment structure based on a BP neural network algorithm. Background Technology

[0002] As the last line of defense against the leakage of fission products into the environment, ensuring the integrity of the containment is of paramount importance. To verify whether the overall sealing performance, structural performance, and strength of the containment meet the requirements, an integrity test of the containment structure must be conducted during the unit's first refueling overhaul and every ten years thereafter.

[0003] In this test, the prestressed concrete containment structure primarily bears the test pressure, and its deformation, strain, temperature, and prestressed steel tendon force are measured. This is a typical low-load, small-range precision test with high measurement accuracy requirements. For example, under maximum test internal pressure, the vertical displacement of the dome is only about one ten-thousandth of the containment height, while the deformation measurement accuracy needs to reach one in several hundred thousand. Therefore, it is necessary to consider the influence of comprehensive environmental factors such as ambient temperature and solar radiation on the overall integrity of the containment structure test, and to make reasonable corrections to the containment deformation to obtain measurement results with the required load accuracy.

[0004] Most existing methods for correcting structural deformation measurements are based on linear fitting to obtain an approximate relationship curve between ambient temperature and structural deformation. Therefore, the fitting relationship between the two needs to be determined in advance. Human factors can cause significant interference, and the calculations are cumbersome. Furthermore, existing correction methods do not consider the influence of solar radiation on structural deformation, resulting in large errors. It is difficult to develop a scientific and accurate method for correcting deformation measurements in the overall test of containment structures. Summary of the Invention

[0005] Therefore, in order to overcome the shortcomings of the prior art, the present invention provides a method, device and equipment for correcting the deformation measurement value of the containment structure based on the BP neural network algorithm. By applying the intelligent algorithm to actual engineering, and comprehensively considering the influence of ambient temperature and solar radiation on the structural deformation measurement value, the present invention can automatically determine the relationship between ambient temperature, solar radiation intensity and structural deformation, eliminate the interference of human factors and improve the measurement accuracy of containment structure deformation.

[0006] Therefore, the present invention provides a method for correcting the deformation measurement values ​​of a containment structure based on a BP neural network algorithm, comprising the following steps:

[0007] S1. During the sampling pre-test, data within a continuous time period with weather parameters similar to those during the formal test are collected and processed to obtain sample data. The sample data includes first ambient temperature data, first solar radiation intensity, and first deformation measurement data caused by the first ambient temperature data and the first solar radiation intensity. The weather parameters include ambient temperature, humidity, wind direction, wind force, and solar radiation intensity, etc. The duration of the time period is not less than 72 hours.

[0008] S2. Based on the sample data, train the BP neural network model to obtain the optimal BP neural network model; the input of the BP neural network model is the first ambient temperature data and the first solar radiation intensity, and the output is the first deformation prediction data;

[0009] S3. Collect data during the formal test and perform second data processing to obtain measured data; the measured data includes second ambient temperature data and second solar radiation intensity.

[0010] S4. After normalizing the measured data, input it into the optimal BP neural network model for calculation to obtain the second deformation prediction data caused by the second ambient temperature data and the second solar radiation intensity during the formal test.

[0011] S5. Obtain the comprehensive deformation measurement values ​​of the containment structure during the formal test;

[0012] S6. The comprehensive deformation measurement value is corrected using the second deformation prediction data to obtain the deformation measurement correction value of the containment structure under test pressure.

[0013] Preferably, the data collected during the sampling pre-test process include the first ambient temperature, the first solar radiation intensity, and the first deformation measurement value.

[0014] Preferably, the first data processing includes first outlier processing, deformation hysteresis processing, and first difference processing. The first outlier processing includes replacing any outliers that suddenly increase or decrease during the pre-test process with interpolated data from the data on either side of the outlier. The deformation hysteresis processing includes adjusting the time-history curve of the first deformation measurement value to eliminate deformation hysteresis. The first difference processing includes subtracting the first initial ambient temperature from the first ambient temperature at each time point to obtain first ambient temperature data, and subtracting the initial measurement value from the first deformation measurement value at each time point to obtain first deformation measurement data.

[0015] Preferably, the step of training the BP neural network model based on the sample data to obtain the optimal BP neural network model includes:

[0016] S21. The sample data is normalized and divided into training samples and test samples;

[0017] S22. Construct a BP neural network model with one input layer, one hidden layer, and one output layer connected sequentially, with the output of each layer serving as the input to the next. The input layer has two nodes, used to receive the normalized values ​​of the first ambient temperature data and the first solar radiation intensity, respectively; the output layer has one node, used to output the first deformation prediction data. The activation function from the input layer to the hidden layer is a tangent sigmoid function, and the activation function from the hidden layer to the output layer is a linear function. The training function uses the Levenberg-Marquardt optimization algorithm. The number of nodes in the hidden layer is based on an empirical formula. Determine, where m is the number of nodes in the input layer, n is the number of nodes in the output layer, and a is an integer between [1, 10].

[0018] S23. Input the training samples into the BP neural network model to train the BP neural network, use the test samples to verify the training effect, and determine the best trained BP neural network model by error analysis between the first deformation prediction data and the first deformation measurement data output by the model.

[0019] Preferably, the data collected during the formal test include the second ambient temperature and the second solar radiation intensity.

[0020] Preferably, the second data processing includes second outlier processing and second difference processing. The first outlier processing includes replacing outliers with interpolated values ​​from the data on either side of the outlier if there are sudden increases or decreases in the data during the formal experiment. The second difference processing includes subtracting the second initial ambient temperature from the second ambient temperature at each time point to obtain the second ambient temperature data.

[0021] Preferably, the step of correcting the comprehensive deformation measurement value using the second deformation prediction data to obtain the deformation measurement correction value of the containment structure includes:

[0022] S61. After performing inverse normalization on the second deformation prediction data, remove it from the comprehensive deformation measurement value to obtain the deformation measurement correction value of the containment structure under test pressure.

[0023] The present invention provides a device for correcting deformation measurement values ​​in an integrity test of a containment structure based on a BP neural network algorithm, comprising:

[0024] A pre-test sampling unit is used to conduct a pre-test of the containment structure integrity test during a time period with weather parameters similar to those during the formal test. The unit samples data from the pre-test process and performs first data processing to obtain sample data. The sample data includes first ambient temperature data, first solar radiation intensity, and first deformation measurement data caused by the first ambient temperature data and the first solar radiation intensity. The weather parameters include second ambient temperature, humidity, wind direction, wind force, and second solar radiation intensity. The duration of the time period is no less than 72 hours.

[0025] The BP neural network model training unit is used to train the BP neural network model based on the sample data to obtain the optimal BP neural network model; the input of the BP neural network model is the first ambient temperature data and the first solar radiation intensity, and the output is the first deformation prediction data.

[0026] The formal test sampling unit is used to collect data during the formal test and perform secondary data processing to obtain measured data; the measured data includes secondary ambient temperature data and secondary solar radiation intensity.

[0027] The deformation prediction unit is used to normalize the measured data and input it into the optimal BP neural network model for calculation to obtain the second deformation prediction data caused by the second ambient temperature data and the second solar radiation intensity during the formal test.

[0028] The comprehensive deformation measurement value acquisition unit is used to acquire the comprehensive deformation measurement value of the containment structure during the formal test.

[0029] The deformation measurement correction unit is used to correct the comprehensive deformation measurement value using the second deformation prediction data to obtain the deformation measurement correction value of the containment structure under test pressure.

[0030] Preferably, the data collected during the sampling pre-test process include the first ambient temperature, the first solar radiation intensity, and the first deformation measurement value.

[0031] Preferably, the first data processing includes first outlier processing, deformation hysteresis processing, and first difference processing. The first outlier processing includes replacing any outliers that suddenly increase or decrease during the pre-test process with interpolated data from the data on either side of the outlier. The deformation hysteresis processing includes adjusting the time-history curve of the first deformation measurement value to eliminate deformation hysteresis. The first difference processing includes subtracting the first initial ambient temperature from the first ambient temperature at each time point to obtain first ambient temperature data, and subtracting the initial measurement value from the first deformation measurement value at each time point to obtain first deformation measurement data.

[0032] Preferably, the BP neural network model training unit includes:

[0033] The normalization processing unit is used to normalize the sample data and divide it into training samples and test samples.

[0034] The BP neural network model building unit is used to construct a BP neural network model with an input layer, a hidden layer, and an output layer connected sequentially, with the output of each layer serving as the input to the next layer. The input layer has two nodes, each receiving normalized values ​​of the first ambient temperature and the first solar radiation intensity. The output layer has one node, outputting the first deformation prediction data. The activation function from the input layer to the hidden layer is a tangent sigmoid function, and the activation function from the hidden layer to the output layer is a linear function. The training function uses the Levenberg-Marquardt optimization algorithm. The number of nodes in the hidden layer is determined using an empirical formula. Determine, where m is the number of nodes in the input layer, n is the number of nodes in the output layer, and a is an integer between [1, 10].

[0035] The optimal BP neural network model determination unit is used to input training samples into the BP neural network model, perform BP neural network training, verify the training effect using test samples, and determine the best trained BP neural network model through error analysis between the first deformation prediction data and the first deformation measurement data output by the model.

[0036] Preferably, the data collected during the formal test include the second ambient temperature and the second solar radiation intensity.

[0037] Preferably, the second data processing includes second outlier processing and second difference processing. The first outlier processing includes replacing outliers with interpolated values ​​from the data on either side of the outlier if there are sudden increases or decreases in the data during the formal experiment. The second difference processing includes subtracting the second initial ambient temperature from the second ambient temperature at each time point to obtain the second ambient temperature data.

[0038] Preferably, the deformation measurement correction unit includes:

[0039] The elimination unit is used to remove the second deformation prediction data from the comprehensive deformation measurement value after inverse normalization processing, so as to obtain the deformation measurement correction value of the containment structure.

[0040] The present invention provides a device for correcting deformation measurement values ​​in containment structure integrity tests based on a BP neural network algorithm, comprising:

[0041] One or more processors; and

[0042] A storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-described method for correcting deformation measurement values ​​in the integrity test of containment structures based on the BP neural network algorithm.

[0043] The method, apparatus, and equipment for correcting deformation measurement values ​​in containment structure integrity tests based on the BP neural network algorithm of the present invention have the following advantages:

[0044] 1. By establishing a mapping relationship between ambient temperature, solar radiation intensity and structural deformation measurements through a BP neural network, the deformation measurements during the integrity test of the prestressed concrete containment structure are corrected. This comprehensively considers the influence of ambient temperature and solar radiation on the structural deformation measurements, and does not require prior determination of the functional relationship between them. This improves the accuracy of containment structure deformation measurement, eliminates interference from human factors, and has high innovation and practical engineering application value.

[0045] 2. Applying machine learning algorithms to practical engineering can realize the input of multiple parameters such as ambient temperature and solar radiation intensity, and automatically determine the correspondence between comprehensive environmental factors and structural deformation. This hands over the tedious calculations to the computer, and scientifically and effectively improves the calculation efficiency and accuracy. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the specific embodiments of the present invention, the drawings used in the description of the specific embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0047] Figure 1 This is a flowchart illustrating a specific example of a method for correcting deformation measurement values ​​in an integrity test of a containment structure based on a BP neural network algorithm, as described in this invention.

[0048] Figure 2 The graph shows the measurement results of the ambient temperature and the first deformation measurement value at a certain measuring point of a containment vessel during a pre-test.

[0049] Figure 3 The graph shows the measurement results of the first solar radiation intensity at a certain measuring point in a containment vessel during a pre-test.

[0050] Figure 4 A comparison chart of predicted and expected values ​​for a test sample at a certain measurement point within a containment structure;

[0051] Figure 5The graph shows the measurement results of ambient temperature and second solar radiation intensity at a certain measuring point on a containment structure during a formal test.

[0052] Figure 6 The graph shows the time history of structural displacement changes predicted by a BP neural network at a certain measuring point of a containment structure during a formal test.

[0053] Figure 7 This is a comparison image showing the deformation of a certain measuring point on a containment structure before and after correction. Detailed Implementation

[0054] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] In the description of this invention, it should be noted that the terminology used herein is for the purpose of describing specific embodiments only and is not intended to limit the invention. The use of terms such as "comprising" and / or "including" is intended to indicate the presence of that feature, integer, step, operation, element, and / or component, without excluding the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or other combinations. The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0056] Furthermore, certain diagrams in this specification are flowcharts used to illustrate methods. It should be understood that each block in these flowcharts, and combinations of such blocks, can be implemented by computer program instructions. These computer program instructions can be loaded onto a computer or other programmable device to form a machine, such that instructions executed on the computer or other programmable device form a structure for implementing the functions specified in the flowchart blocks. These computer program instructions can also be stored in a computer-readable storage medium that instructs a computer or other programmable device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium form an article of writing containing an instruction structure for implementing the functions specified in the flowchart blocks. The computer program instructions can also be loaded onto a computer or other programmable device to perform a series of operational steps on the computer or other programmable device to form a computer-implemented process, such that instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the flowchart blocks.

[0057] Accordingly, the blocks in each flowchart support combinations of structures for performing the specified functions and combinations of steps for performing the specified functions. It should also be understood that each block in the flowchart, and combinations of blocks in the flowchart, can be implemented by a dedicated hardware-based computer system, or a combination of dedicated hardware and computer instructions, to perform the specified functions or steps.

[0058] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0059] This embodiment employs a BP neural network algorithm to correct the deformation measurement values ​​of the containment structure integrity test, thus forming a method for correcting the deformation measurement values ​​of the containment structure integrity test based on the BP neural network algorithm. The specific steps are as follows: Figure 1 As shown, it includes:

[0060] S1. During the sampling pre-test, data from a continuous time period with weather parameters similar to those during the formal test (such as the first ambient temperature, the first solar radiation intensity, and the first deformation measurement value) are collected and processed to obtain sample data. The sample data includes the first ambient temperature data, the first solar radiation intensity, and the first deformation measurement data caused by the first ambient temperature data and the first solar radiation intensity.

[0061] The data collected during the preliminary test should be selected from a time period similar to the weather parameters (such as ambient temperature, humidity, wind direction, wind force, solar radiation intensity, etc.) during the formal test, and the time period should not be less than 72 hours. Figure 2 As shown in the figure, the ambient temperature and the measurement results of the first deformation (horizontal radial displacement of the structure) during the pre-test are illustrated. Figure 3 As shown in the figure, the measurement results of the first solar radiation intensity in the preliminary test are presented.

[0062] Preferably, the first data processing includes first outlier processing, deformation lag processing, and first difference processing. Since there are outliers that suddenly increase or decrease in the pre-test and formal test data, the first outlier processing includes replacing the outlier with interpolated data from its left and right sides if there are outliers that suddenly increase or decrease in the data during the pre-test process. Because concrete has poor thermal conductivity, structural displacement changes lag behind changes in ambient temperature and solar radiation intensity. By adjusting the lag time of structural displacement changes, the peak and trough values ​​of the ambient temperature and displacement curves can be approximately overlapped. Therefore, the deformation lag processing includes adjusting the time history curve of the first deformation measurement value to eliminate the deformation lag. The first difference processing includes subtracting the first initial ambient temperature from the first ambient temperature at each time to obtain the first ambient temperature data (ambient temperature difference) and subtracting the initial measurement value from the first deformation measurement value at each time to obtain the first deformation measurement data (structural displacement measurement value difference).

[0063] S2. Based on the sample data, train the BP neural network model to obtain the optimal BP neural network model. The input layer of the BP neural network model has two nodes, used to input the first ambient temperature data and the first solar radiation intensity, respectively, and the output layer has one node, used to output the first deformation prediction data. Specifically, this includes the following steps:

[0064] S21. The sample data is normalized and divided into training samples and test samples. For example, the training samples can be the first 70% of the normalized sample data, and the test samples can be the last 30% of the normalized sample data.

[0065] S22. Construct a BP neural network model with one input layer, one hidden layer, and one output layer connected sequentially, with the output of each layer serving as the input to the next. The input layer has two nodes, used to receive the normalized values ​​of the first ambient temperature data and the first solar radiation intensity, respectively; the output layer has one node, used to output the first deformation prediction data. The activation function from the input layer to the hidden layer is a tangent sigmoid function, and the activation function from the hidden layer to the output layer is a linear function. The training function uses the Levenberg-Marquardt optimization algorithm. The number of nodes in the hidden layer is based on an empirical formula. The optimal number of hidden layer nodes is determined by the following formula: m is the number of nodes in the input layer, n is the number of nodes in the output layer, and a is an integer between 1 and 10. Therefore, the optimal number of hidden layer nodes is calculated to be between 2 and 11. This optimal number of hidden layer nodes is then determined using the established BP neural network.

[0066] S23. Input the training samples into the BP neural network model to train the BP neural network. Verify the training effect using test samples. Determine the best trained BP neural network model by analyzing the error between the first deformation prediction data and the first deformation measurement data output by the model. For example, the maximum number of training iterations for the BP neural network model is 1000, the learning rate is 0.01, and the minimum training error is 0.000001. In a specific example, 202 sets of data from a certain measurement point of a containment vessel were selected, including 143 sets of training samples and 60 sets of test samples. After inputting the test samples into the best BP neural network model, the first parallel prediction data (predicted value) output by the model is obtained. The comparison between the predicted value and the expected value (the first deformation measurement data corresponding to the test sample) is as follows. Figure 4 As shown, the error between the predicted value and the expected value meets the requirements.

[0067] The table below compares five types of errors—sum of squares, mean absolute error, mean square error, root mean square error, and mean percentage error—under three correction methods: linear fitting, single-factor BP neural network for ambient temperature, and multi-factor BP neural network for ambient temperature and solar radiation intensity.

[0068]

[0069] As shown in the table above, the linear fitting error is greater than that of the BP neural network that only considers ambient temperature. The BP neural network that only considers ambient temperature is greater than that of the BP neural network that considers both ambient temperature and solar radiation intensity. This indicates that the BP neural network structure deformation correction method that considers both ambient temperature and solar radiation intensity has a much lower error than the previous two correction methods. Therefore, it can be concluded that the correction method in this embodiment significantly improves the accuracy of deformation (structural displacement) measurement.

[0070] S3. Collect data during the formal experiment (such as the second ambient temperature and the second solar radiation intensity) and perform the second data processing to obtain measured data. The measured data includes the second ambient temperature data and the second solar radiation intensity. Figure 5 As shown in the figure, the measurements of ambient temperature and second solar radiation intensity during the formal test are presented.

[0071] Preferably, the second data processing includes second outlier processing and second difference processing. The first outlier processing includes replacing outliers with interpolated values ​​from the data on either side of the outlier if there are sudden increases or decreases in the data during the formal experiment. The second difference processing includes subtracting the second initial ambient temperature from the second ambient temperature at each time point to obtain the second ambient temperature data.

[0072] S4. After normalizing the measured data, input it into the optimal BP neural network model for calculation to obtain the second deformation prediction data caused by the second ambient temperature data and the second solar radiation intensity during the formal experiment. Figure 6 As shown in the figure, the BP predicted displacement during the formal test is obtained after inverse normalization of the second deformation prediction data.

[0073] S5. Obtain the comprehensive deformation measurement values ​​of the containment structure during the formal test (caused by the second ambient temperature, the second solar radiation intensity, and the test pressure, etc.).

[0074] S6. Correct the comprehensive deformation measurement value using the second deformation prediction data. Specifically, subtract the structural displacement changes caused by ambient temperature and solar radiation intensity during the formal test from the comprehensive deformation measurement value. This involves subtracting the second deformation prediction data from the comprehensive deformation measurement value and performing inverse normalization to obtain the corrected deformation measurement value of the containment structure under test pressure. The effects before and after correction are as follows: Figure 7 As shown.

[0075] The aforementioned method for correcting the deformation measurement values ​​of containment structures based on the BP neural network algorithm comprehensively considers the influence of ambient temperature and solar radiation on the structural deformation measurement values, and does not require prior determination of the functional relationship between them. This improves the accuracy of containment structure deformation measurement, eliminates interference from human factors, and has high innovation and practical engineering application value.

[0076] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for correcting deformation measurement values in a containment structure integrity test based on a BP neural network algorithm, characterized in that, The method comprises the following steps: sampling data in a continuous time period similar to weather parameters during the pre-test process and the formal test period and performing first data processing to obtain sample data; the sample data comprises first ambient temperature data, first solar radiation intensity, and first deformation measurement data caused by the first ambient temperature data and the first solar radiation intensity; the first data processing comprises first outlier processing, deformation lag processing, and first difference processing; the first outlier processing comprises replacing an outlier with its left and right data by interpolation if the data during the pre-test process has an outlier with sudden increase or decrease; the deformation lag processing comprises adjusting a time curve of the first deformation measurement value to eliminate deformation lag; and the first difference processing comprises subtracting a first initial ambient temperature from each time point of the first ambient temperature to obtain the first ambient temperature data, and subtracting an initial measurement value from each time point of the first deformation measurement value to obtain the first deformation measurement data; training a BP neural network model based on the sample data to obtain an optimal BP neural network model; the input of the BP neural network model is the first ambient temperature data and the first solar radiation intensity, and the output is first deformation prediction data; collecting data during the formal test period and performing second data processing to obtain measured data; the measured data comprises second ambient temperature data and second solar radiation intensity; performing normalization processing on the measured data and inputting the normalized measured data into the optimal BP neural network model to obtain second deformation prediction data caused by the second ambient temperature data and the second solar radiation intensity during the formal test period; obtaining a comprehensive deformation measurement value of the containment structure during the formal test period; correcting the comprehensive deformation measurement value by using the second deformation prediction data to obtain a deformation measurement correction value of the containment structure under the test pressure.

2. The method of claim 1, wherein, The length of the time period is not less than 72 hours.

3. The method according to claim 1 or 2, characterized in that, The step of training the BP neural network model based on the sample data to obtain the optimal BP neural network model comprises: performing normalization processing on the sample data and dividing the normalized sample data into training samples and test samples; constructing a BP neural network model having an input layer, a hidden layer, and an output layer connected in sequence, wherein the output of each layer is used as the input of the next layer; the input layer has two nodes for one-to-one receiving normalized values of the first ambient temperature data and the first solar radiation intensity; the output layer has one node for outputting the first deformation prediction data; inputting the training samples into the BP neural network model, performing BP neural network training, verifying the training effect by using the test samples, and determining the trained optimal BP neural network model by analyzing the error between the first deformation prediction data output by the model and the first deformation measurement data.

4. The method of claim 3, wherein, The activation function from the input layer to the hidden layer is a tangent S-shaped function, the activation function from the hidden layer to the output layer is a linear function, and the training function adopts a Levenberg-Marquardt optimization algorithm; the number of nodes of the hidden layer is determined according to an empirical formula determined, wherein, is the number of nodes of the input layer, is the number of nodes of the output layer, is an integer between 1 and 10.

5. The method of claim 4, wherein, The second data processing comprises second outlier processing and second difference processing; the first outlier processing comprises replacing an outlier with its left and right data by interpolation if the data during the formal test period has an outlier with sudden increase or decrease; The second difference processing includes subtracting the second initial environmental temperature from the second environmental temperature at each time point to obtain second environmental temperature data.

6. The method of claim 5, wherein, The step of correcting the integrated deformation measurement value by using the second deformation prediction data to obtain the deformation measurement correction value of the containment structure includes: After the second deformation prediction data is subjected to inverse normalization processing, the second deformation prediction data is removed from the integrated deformation measurement value to obtain the deformation measurement correction value of the containment structure under the test pressure.

7. A containment structure integrity test deformation measurement value correction device based on a BP neural network algorithm, characterized in that, Comprise: A pre-test sampling unit is configured to perform a pre-test of the containment structure integrity test in a time period similar to the weather parameters during the formal test of the containment structure integrity test, sample data during the pre-test and perform first data processing to obtain sample data; the sample data includes first environmental temperature data, first solar radiation intensity and first deformation measurement data caused by the first environmental temperature data and the first solar radiation intensity; the first data processing includes first outlier processing, deformation lag processing and first difference processing; the first outlier processing includes replacing the outlier with interpolation of data on the left and right sides of the outlier if the data during the pre-test has an outlier with sudden increase or decrease; the deformation lag processing includes adjusting the time curve of the first deformation measurement value to eliminate the lag of the deformation; and the first difference processing includes subtracting the first initial environmental temperature from the first environmental temperature at each time point to obtain the first environmental temperature data and subtracting the initial measurement value from the first deformation measurement value at each time point to obtain the first deformation measurement data; A BP neural network model training unit is configured to train the BP neural network model based on the sample data to obtain an optimal BP neural network model; the input of the BP neural network model is the first environmental temperature data and the first solar radiation intensity, and the output is the first deformation prediction data; A formal test sampling unit is configured to sample data during the formal test and perform second data processing to obtain measured data; the measured data includes second environmental temperature data and second solar radiation intensity; A deformation prediction unit is configured to input the measured data after normalization processing into the optimal BP neural network model to obtain second deformation prediction data caused by the second environmental temperature data and the second solar radiation intensity during the formal test; An integrated deformation measurement value acquisition unit is configured to acquire an integrated deformation measurement value of the containment structure during the formal test; A deformation measurement value correction unit is configured to correct the integrated deformation measurement value by using the second deformation prediction data to obtain a deformation measurement correction value of the containment structure under the test pressure.

8. A computer readable storage medium, the storage medium storing instructions, the instructions being executed by a processor to implement the containment structure integrity test deformation measurement value correction method based on the BP neural network algorithm according to any one of claims 1-6.

9. A containment structure integrity test deformation measurement value correction device based on a BP neural network algorithm, characterized in that, Comprise: One or more processors; And A storage device is configured to store one or more programs, which when executed by the one or more processors, cause the one or more processors to implement the method for correcting the deformation measurement value of the containment structure integrity test based on the BP neural network algorithm according to any one of claims 1-6.

Images