Concrete continuous rigid frame bridge temperature-induced deflection prediction and safety early warning method based on health monitoring

Abstract

The invention provides a concrete continuous rigid frame bridge temperature-induced deflection prediction and safety early warning method based on health monitoring, which aims at a concrete continuous rigid frame bridge type and extracts specific indexes reflecting structural performance based on mass measured data acquired by a bridge health monitoring system. Extracting section temperature gradient information of the concrete continuous rigid frame bridge through kernel principal component analysis, and performing wavelet decomposition on deflection data to extract temperature-dependent deflection; the structure performance change is reflected through the prediction model of the temperature-induced deflection, the long-term change rule of the bridge structure performance is effectively analyzed, and a bridge temperature-induced deflection index prediction model with a clear physical meaning is constructed; setting a bridge safety early warning threshold value based on the model prediction error, and performing multi-stage early warning on the bridge state by simply adjusting parameters according to the actual operation condition of the bridge structure and the requirements of bridge owners; the function of monitoring bridge health is realized, important guidance is provided for bridge operation management, and the system has a wide application prospect.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge structure health monitoring, and in particular relates to a temperature-induced deflection prediction and safety early warning method for a concrete continuous rigid frame bridge based on health monitoring. Background Art

[0002] Concrete continuous rigid frame bridges have great bending stiffness along the bridge and torsional stiffness in the transverse direction of the bridge, which can well meet the force requirements of larger spans and are a very vital bridge structure. Performance degradation of bridge structures will inevitably occur during operation, and safety warnings are of great practical significance for ensuring the safe operation of bridges. With the development of structural health monitoring technology, health monitoring systems are installed for bridge structures to monitor bridge structure response indicators in real time and for a long time, providing original data support for bridge safety warnings. Therefore, based on the data collected by the bridge health monitoring system, the characteristics of long-term performance changes of bridge structures are studied, which provides new ideas for realizing bridge structure safety warnings.

[0003] The deflection response caused by temperature has low-frequency characteristics and is an indicator that can be used to evaluate the degradation of bridge performance. Based on the characteristics of temperature-induced deflection, starting from the frequency domain characteristics, the decoupling of temperature-induced deflection from other deflection components in the measured signal can be achieved. This theory provides a basis for establishing a prediction model for temperature-induced deflection. Due to the material properties, the time lag effect of temperature-induced deflection of concrete continuous rigid frame bridges is more significant, that is, the change in temperature-induced deflection usually lags behind the change in temperature. At the same time, affected by factors such as sunlight and ambient temperature, there is a significant temperature gradient in the cross-section of concrete continuous rigid frame bridges. These characteristics of concrete continuous rigid frame bridges pose challenges to the accurate prediction of temperature-induced deflection through simple models such as linear regression. Therefore, developing and training a model suitable for predicting the temperature-induced deflection of concrete continuous rigid frame bridges is an important part of achieving subsequent bridge safety warnings.

[0004] Safety warning for bridge structures is a key means to ensure safe operation. How to set the safety warning threshold is the key to whether the warning system can operate normally. Based on the massive monitoring data provided by the health system, it is feasible to extract bridge structure information from it, accurately predict the future response of the bridge structure, and then achieve early warning of the bridge structure. The temperature-induced deflection prediction model established by the monitoring data corresponding to the past time series reflects the expected results of the health status of the bridge. The prediction error of the accurate prediction model fluctuates only within a small range and has obvious statistical laws. Therefore, based on the statistical laws of the temperature-induced deflection prediction results, setting the bridge safety warning threshold is a feasible method. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a temperature-induced deflection prediction and safety early warning method for a concrete continuous rigid frame bridge based on health monitoring, which is used for monitoring the health of the bridge.

[0006] The technical solution adopted by the present invention to solve the above technical problems is: a method for predicting temperature-induced deflection and safety early warning of a concrete continuous rigid frame bridge based on health monitoring, comprising the following steps:

[0007] S1: Determine the monitoring indicators and system solutions according to the purpose of bridge health monitoring, configure sensors on the bridge body and collect health monitoring indicator data;

[0008] S2: Extract the characteristics of health monitoring data, including performing kernel principal component analysis on temperature data to extract cross-sectional temperature gradient and performing wavelet decomposition on deflection data to extract temperature-induced deflection;

[0009] S3: Using the temperature gradient of the bridge section as input and the temperature-induced deflection as output, a support vector machine prediction model is trained. The support vector machine prediction model has nonlinear prediction properties.

[0010] S4: Temperature-induced deflection is selected as the safety warning indicator of concrete continuous rigid frame bridges. The statistical law of model prediction error is analyzed according to the results of model prediction, and the bridge safety warning threshold is set based on the 3-sigma principle.

[0011] According to the above scheme, in step S1, the specific steps are:

[0012] S11: Determine the monitoring indicators and health monitoring system configuration plan, and install temperature sensors and deflection sensors on the actual bridge;

[0013] S12: Collect health monitoring indicator data and divide the monitoring data into training set and test set according to the accuracy requirements of the prediction model.

[0014] Furthermore, in step S2, the specific steps are:

[0015] S21: Reduce the dimension of the temperature data of the bridge collected by the temperature sensor through kernel principal component analysis, remove data redundancy, extract the temperature gradient information of the bridge section, and discretely express the temperature gradient of the bridge section;

[0016] S22: According to the frequency domain characteristics of the deflection data of the bridge collected by the deflection sensor, the deflection data is decoupled by wavelet decomposition to extract the temperature-induced deflection.

[0017] Furthermore, in step S21, the specific steps of extracting the temperature gradient of the bridge cross section are:

[0018] Assume that N is the dimension of the original temperature data, M is the number of sampling points, and the temperature monitoring value X∈R N×M Map it to a high-dimensional space F that makes it linearly separable;

[0019] A linear principal component analysis is performed on φ(X) in F to obtain a vector p∈F that maximizes the variance of the temperature gradient t=φ(X)p of the bridge section.

[0020] Furthermore, in step S22, the specific steps of extracting the temperature-induced deflection are:

[0021] Assume D v is the vehicle-induced deflection, D t is the temperature-induced deflection, D c is the beam deflection caused by concrete shrinkage creep, δ is the measurement error, and the deflection data D of the bridge is a Divided into:

[0022] D a =D v +D t +D c +δ (1)

[0023] The measurement error is regarded as a random distribution with an expected value of 0 and a constant variance, and the influence of concrete shrinkage creep and measurement error on bridge monitoring data is ignored; let a and b be scale factors and translation factors, W f (a, b) are wavelet coefficients, ψ(e) is the basic wavelet, F(w) is the Fourier transform of ψ(e), and Cψ is the intermediate variable; decoupling D by wavelet decomposition v and D t , extract the temperature-induced deflection D t for:

[0024]

[0025]

[0026] Furthermore, in step S3, the specific steps are:

[0027] S31: taking the temperature gradient and the temperature-induced deflection as input and output respectively into the support vector machine prediction model, using the training set to train the prediction model, and using the test set to test the prediction accuracy of the prediction model;

[0028] S32: Solve the nonlinear regression problem by using support vector machine; find the optimal hyperplane in the feature space to distinguish positive samples from negative samples, and ensure that the distance from the data point closest to the hyperplane to the hyperplane is maximized to achieve nonlinear prediction of temperature-induced deflection;

[0029] Let x p is the input feature vector, y p is the corresponding target value; w is the normal vector of the hyperplane; m is the intercept of the hyperplane; ε pis a slack variable used to process the error and distribution of data points; C is a regularization parameter used to control the complexity of the regressor; i is the number of calculations, q is the upper limit of the number of calculations, i = 1, 2, ..., q; the nonlinear prediction process of temperature-induced deflection is:

[0030]

[0031] S33: Check the accuracy of the prediction model; let m be the number of sample points, g be the current number of calculations, g = 1, 2, ..., m; is the monitoring value of temperature-induced deflection; is the predicted value of temperature-induced deflection; is the average value of the deflection monitoring value; the evaluation index determination coefficient R 2 And the mean absolute error MAE are:

[0032]

[0033]

[0034] If R 2 If it is greater than the preset value and MAE is less than the preset value, the accuracy of the prediction model meets the expected requirements, and subsequent error analysis and safety threshold setting are performed; otherwise, data is collected again and the prediction model is trained.

[0035] According to the above scheme, in step S4, the specific steps are:

[0036] S41: Analysis of prediction error of temperature-induced deflection prediction model:

[0037]

[0038] S42: According to the statistical law of prediction error Δ, the safety warning threshold of concrete continuous rigid frame bridge is determined based on the 3-sigma principle; let Φ be the cumulative distribution function; β be the safety warning threshold δ x The significance level is used for operators to adjust to achieve multi-level early warning of bridge status; S is the predicted value of temperature-induced deflection; α is the standard deviation magnification factor, which is determined by the inverse function of the cumulative distribution function; σ x' is the standard deviation of the prediction error; the upper limit UWT and lower limit LWT of the safety warning threshold are:

[0039] Φ(UWT≥δ x ≥LWT)=β (8),

[0040] α=Φ -1 (β) (9),

[0041] UWT=y S +ασ x' (10)

[0042] LWT=y S -ασ x' (11).

[0043] Furthermore, the step S42 further includes the following steps:

[0044] Adjust the standard deviation magnification factor α to achieve multi-level early warning of the bridge safety status.

[0045] A computer memory stores a computer program executable by a computer processor, wherein the computer program executes a temperature-induced deflection prediction and safety early warning method for a concrete continuous rigid frame bridge based on health monitoring.

[0046] A health monitoring-based concrete continuous rigid frame bridge temperature-induced deflection prediction and safety warning system, the system comprises a processor and a memory, the memory stores computer instructions, the processor is used to execute the computer instructions stored in the memory, when the computer instructions are executed by the processor, the system implements the steps of a health monitoring-based concrete continuous rigid frame bridge temperature-induced deflection prediction and safety warning method.

[0047] The beneficial effects of the present invention are:

[0048] 1. The method for predicting temperature-induced deflection and providing safety warning for concrete continuous rigid frame bridges based on health monitoring of the present invention is targeted at concrete continuous rigid frame bridges. The method extracts specific indicators reflecting structural performance based on massive measured data collected by a bridge health monitoring system, effectively analyzes the long-term variation law of bridge structural performance, and constructs a prediction model for temperature-induced deflection indicators of bridges with clear physical meanings. The method sets a bridge safety warning threshold based on the model prediction error, and performs multi-level warnings on the bridge status by simply adjusting parameters according to the actual operating conditions of the bridge structure and the needs of the bridge owner. The method realizes the function of monitoring bridge health, provides important guidance for bridge operation and management, and has broad application prospects.

[0049] 2. The present invention fully considers factors such as the bridge structure layout, bridge site environment, and line requirements based on the actual conditions of different bridges, determines a suitable health monitoring plan for the bridge, completes sensor installation, collects monitoring index data, and divides the collected monitoring data into training sets and test sets according to the needs of predictive model training, laying the foundation for subsequent temperature-induced deflection prediction and bridge safety early warning.

[0050] 3. The present invention is based on the measured temperature and deflection data of bridge health monitoring, extracts the temperature gradient information of the cross-section of the concrete continuous rigid frame bridge through kernel principal component analysis, and extracts the temperature-induced deflection by wavelet decomposition of the deflection data; trains a support vector machine model with temperature gradient information as input and temperature-induced deflection as output, and verifies the accuracy of the prediction model. The prediction model of temperature-induced deflection reflects the changes in structural performance, overcomes the time lag effect, and realizes the accurate prediction of the temperature-induced deflection of the bridge.

[0051] 4. The present invention determines the bridge safety warning threshold based on the 3-sigma principle by analyzing the error of the temperature-induced deflection prediction model. The theory is clear and complete, so it has strong applicability. The bridge safety warning threshold no longer depends on engineering experience or relevant specifications. The prediction model of the future state is established based on the health monitoring data of the bridge's past state, which can more intuitively reflect the true state of the bridge structure. The set warning threshold has the characteristics of dynamic change over time, which can better meet the actual engineering needs. At the same time, by simply modifying the parameters to achieve multi-level warning of the bridge status, it provides more reaction time for bridge operation and management, and provides data support for bridge safety warning, which is widely applicable to concrete continuous rigid frame bridges. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a flow chart of an embodiment of the present invention.

[0053] Figure 2 It is a structural layout diagram of a three-span continuous rigid frame bridge according to an embodiment of the present invention.

[0054] Figure 3 It is a layout diagram of the health monitoring system for the bridge II section according to an embodiment of the present invention.

[0055] Figure 4 4 is a diagram showing the results of a kernel principal component analysis of temperature monitoring values ​​according to an embodiment of the present invention.

[0056] Figure 5 4 is a diagram of the deflection wavelet decomposition result of an embodiment of the present invention.

[0057] Figure 6 4 is a diagram of the prediction results of the temperature-induced deflection of an embodiment of the present invention.

[0058] Figure 7 It is a probability statistical diagram of the error of the temperature-induced deflection prediction result of an embodiment of the present invention.

[0059] Figure 8 It is a safety warning threshold diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0061] like Figure 2 As shown, this embodiment predicts the temperature-induced deflection of a three-span continuous rigid frame bridge and sets a safety warning threshold for it to verify the superiority of the proposed method.

[0062] See also Figure 1 , the specific steps of the embodiment of the present invention are as follows:

[0063] S1. Determine the indicators and solution configuration of the bridge health monitoring system based on the actual project;

[0064] Based on the purpose of bridge health monitoring, the monitoring indicators and health monitoring system configuration plan are determined, and the temperature and deflection sensors are installed on the actual bridge; the health monitoring indicator data are collected, and the monitoring data are divided into training sets and test sets according to the accuracy requirements of the prediction model.

[0065] S11, determine the monitoring indicators and health monitoring system configuration plan, complete the sensor installation on the actual bridge, that is, the data acquisition module; in this embodiment, 12 temperature sensors and 1 deflection sensor are arranged in the bridge II section. The health monitoring system of this embodiment is arranged as follows Figure 3 shown.

[0066] S12, collecting health monitoring index data, and dividing the monitoring data into a training set and a test set according to the accuracy requirements of the prediction model. In this embodiment, the training set and the test set are divided according to a ratio of 9:1.

[0067] S2. Extract features from health monitoring data. Feature extraction includes performing kernel principal component analysis on temperature data to extract cross-sectional temperature gradient information and performing wavelet decomposition on deflection data to extract temperature-induced deflection. Specifically, it includes: reducing the dimension of temperature monitoring data through kernel principal component analysis algorithm, removing redundancy of temperature monitoring data, and extracting temperature gradient information of concrete continuous rigid frame bridge cross section; performing wavelet decomposition and decoupling on deflection monitoring data to extract temperature-induced deflection.

[0068] S21. Based on the temperature data of the concrete continuous rigid frame bridge section collected by the data acquisition module, the temperature monitoring data is reduced in dimension through kernel principal component analysis to remove the redundancy of the temperature monitoring data, extract the temperature gradient information of the concrete continuous rigid frame bridge section, and realize the discrete expression of the temperature gradient of the bridge section based on the measured data.

[0069] Specifically, the process of extracting bridge section temperature gradient information is as follows:

[0070] The temperature monitoring value X∈R in the low-dimensional space is linearly inseparable through the nonlinear mapping function φ. N×MMap it to a high-dimensional space F that makes it linearly separable, where N is the dimension of the original temperature data and M is the number of sampling points. Perform a linear principal component analysis on φ(X) in F to obtain a vector p∈F, so that t=φ(X)p variance is maximized. t is the temperature gradient information of the bridge section described in the present invention in F. The results of the kernel principal component analysis of the temperature monitoring value in this embodiment are as follows Figure 4 shown.

[0071] S22. According to the deflection data of the concrete continuous rigid frame bridge obtained by the data acquisition module, based on the frequency domain characteristics of the measured deflection signal, the deflection signal is decoupled by wavelet decomposition to extract the temperature-induced deflection.

[0072] Specifically, the present invention converts the bridge deflection monitoring data D a Divided into 4 components:

[0073] D a =D v +D t +D c +δ (1)

[0074] Where D c is the beam deflection caused by concrete shrinkage creep; δ is the measurement error. The present invention considers that concrete shrinkage creep is a long-term influencing factor on beam deflection, and the measurement error can be regarded as a random distribution with an expected value of 0 and a constant variance, so the influence of concrete shrinkage creep and measurement error on bridge monitoring data is ignored. Further, D is decomposed by wavelet v and D t Decouple and extract the temperature-induced deflection D t The wavelet decomposition result of the deflection monitoring value in this embodiment is as follows: Figure 5 shown.

[0075] Specifically, the process of extracting the temperature effect of bridge deflection is as follows:

[0076]

[0077]

[0078] Where a and b are scale factors and translation factors; W f (a, b) are wavelet coefficients; ψ(e) is the basic wavelet. F(w) is the Fourier transform of ψ(e).

[0079] S3. Based on the cross-sectional temperature gradient and temperature-induced deflection obtained from the extracted data features, the cross-sectional temperature gradient is used as input and the temperature-induced deflection is used as output to train the nonlinear prediction model / support vector machine prediction model, where the training set is used to train the model and the test set is used to verify the prediction accuracy of the model. If the accuracy meets the expected requirements, subsequent error analysis and safety threshold setting are performed, otherwise the health monitoring system should be reprocessed and the prediction model should be retrained.

[0080] S31. The temperature gradient information of the bridge section obtained by kernel principal component analysis and the temperature-induced deflection obtained by wavelet decomposition are respectively brought into the support vector machine model as input and output, and the prediction model is trained according to the training set and test set divided in step S12 and the model accuracy is tested.

[0081] S32. Due to the significant time lag effect of concrete continuous rigid frame bridges, temperature and temperature-induced deflection show a highly nonlinear relationship. The present invention solves the nonlinear regression problem by using a support vector machine. The core idea is to find an optimal hyperplane in the feature space, distinguish between positive and negative samples, and ensure that the distance from the data point closest to the hyperplane to the hyperplane is maximized.

[0082] Specifically, the implementation process of nonlinear prediction of temperature-induced deflection is as follows:

[0083]

[0084] Among them, x p is the input feature vector, y p is the corresponding target value; w is the normal vector of the hyperplane; m is the intercept of the hyperplane; ε is the relaxation variable used to process the error and distribution of data points; C is the regularization parameter used to control the complexity of the regressor. The prediction results of the temperature-induced deflection in this embodiment are as follows Figure 6 shown.

[0085] S33, further, test the accuracy of the prediction model, the evaluation index determination coefficient R 2 And the mean absolute error MAE is:

[0086]

[0087]

[0088] Where m is the number of sample points; is the monitoring value of temperature-induced deflection; is the predicted value of temperature-induced deflection; is the average value of the deflection monitoring value. 2 If the value is greater than the preset value and the MAE is less than the preset value, it means that the accuracy of the prediction model meets the expected requirements. The accuracy test results of the prediction model in this embodiment are shown in Table 1.

[0089] Table 1 Prediction model accuracy test results

[0090] Training set Test Set <![CDATA[R 2 ]]> 0.9930 0.9917 MAE 0.0978 0.0951

[0091] S4. Select temperature-induced deflection as the safety warning indicator of concrete continuous rigid frame bridges, analyze the statistical law of model prediction error according to the results of model prediction, and set the bridge safety warning threshold based on the 3-sigma principle.

[0092] S41. Analyze the prediction error of the temperature-induced deflection prediction model. Specifically, the prediction error is

[0093]

[0094] S42. Further, according to the statistical law of the prediction error Δ, the error probability statistics of the temperature-induced deflection prediction result of this embodiment are as follows: Figure 7 Based on the 3-sigma principle, the safety warning threshold of the concrete continuous rigid frame bridge is determined. Specifically, the upper limit UWT and lower limit LWT of the safety warning threshold are:

[0095] Φ(UWT≥δ x ≥LWT)=β (8)

[0096] α=Φ -1 (β) (9)

[0097] UWT=y S +ασ x' (10)

[0098] LWT=y S -ασ x' (11)

[0099] Among them, Φ is the cumulative distribution function; β is the significance level of safety warning, which is set by the bridge owner according to the bridge status. Adjusting β can achieve multi-level warning of bridge status; y S is the predicted value of temperature-induced deflection; α is the standard deviation magnification factor, which is determined by the inverse function of the cumulative distribution function; σ x' is the standard deviation of the prediction error. Taking α as 3, the bridge safety warning threshold in this embodiment is as follows: Figure 8 shown.

[0100] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0101] It is understandable that the multi-level warning of bridge safety status can be achieved by simply adjusting the parameter α, which provides more time for bridge operation safety management, and will not be elaborated here.

[0102] The above embodiments are only used to illustrate the design ideas and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, any equivalent changes or modifications made based on the principles and design ideas disclosed by the present invention are within the protection scope of the present invention.

Claims

1. A method for predicting temperature-induced deflection and safety early warning of a concrete continuous rigid frame bridge based on health monitoring, characterized in that: The following steps are involved: S1: Determine the monitoring indicators and system solutions according to the purpose of bridge health monitoring, configure sensors on the bridge body and collect health monitoring indicator data; S2: Extract the characteristics of health monitoring data, including performing kernel principal component analysis on temperature data to extract cross-sectional temperature gradient and performing wavelet decomposition on deflection data to extract temperature-induced deflection; S3: Using the temperature gradient of the bridge section as input and the temperature-induced deflection as output, a support vector machine prediction model is trained. The support vector machine prediction model has nonlinear prediction properties. S4: Temperature-induced deflection is selected as the safety warning indicator of concrete continuous rigid frame bridges. The statistical law of model prediction error is analyzed according to the results of model prediction, and the bridge safety warning threshold is set based on the 3-sigma principle.

2. According to claim 1, a method for predicting temperature-induced deflection and safety early warning of a concrete continuous rigid frame bridge based on health monitoring, characterized in that: In the step S1, the specific steps are: S11: Determine the monitoring indicators and health monitoring system configuration plan, and install temperature sensors and deflection sensors on the actual bridge; S12: Collect health monitoring indicator data and divide the monitoring data into training set and test set according to the accuracy requirements of the prediction model.

3. The method for predicting temperature-induced deflection and safety early warning of a concrete continuous rigid frame bridge based on health monitoring according to claim 2 is characterized in that: In the step S2, the specific steps are: S21: Reduce the dimension of the temperature data of the bridge collected by the temperature sensor through kernel principal component analysis, remove data redundancy, extract the temperature gradient information of the bridge section, and discretely express the temperature gradient of the bridge section; S22: According to the frequency domain characteristics of the deflection data of the bridge collected by the deflection sensor, the deflection data is decoupled by wavelet decomposition to extract the temperature-induced deflection.

4. The method for predicting temperature-induced deflection and safety early warning of a concrete continuous rigid frame bridge based on health monitoring according to claim 3 is characterized by: In step S21, the specific steps of extracting the temperature gradient of the bridge cross section are: Assume that N is the dimension of the original temperature data, M is the number of sampling points, and the temperature monitoring value X∈R N×M Map it to a high-dimensional space F that makes it linearly separable; A linear principal component analysis is performed on φ(X) in F to obtain a vector p∈F that maximizes the variance of the temperature gradient t=φ(X)p of the bridge section.

5. The method for predicting temperature-induced deflection and safety early warning of a concrete continuous rigid frame bridge based on health monitoring according to claim 3 is characterized by: In step S22, the specific steps of extracting the temperature-induced deflection are: let D v is the vehicle-induced deflection, D t is the temperature-induced deflection, D c is the beam deflection caused by concrete shrinkage creep, δ is the measurement error, and the deflection data D of the bridge is a Divided into: D a =D v +D t +D c +δ (1) The measurement error is regarded as a random distribution with an expected value of 0 and a constant variance, and the influence of concrete shrinkage creep and measurement error on bridge monitoring data is ignored; let a and b be scale factors and translation factors, W f (a, b) are wavelet coefficients, ψ(e) is the basic wavelet, F(w) is the Fourier transform of ψ(e), and Cψ is the intermediate variable; decoupling D by wavelet decomposition v and D t , extract the temperature-induced deflection D t for:

6. The method for predicting temperature-induced deflection and safety early warning of a concrete continuous rigid frame bridge based on health monitoring according to claim 2 is characterized in that: In the step S3, the specific steps are: S31: taking the temperature gradient and the temperature-induced deflection as input and output respectively into the support vector machine prediction model, using the training set to train the prediction model, and using the test set to test the prediction accuracy of the prediction model; S32: Solve the nonlinear regression problem by using support vector machine; find the optimal hyperplane in the feature space to distinguish positive samples from negative samples, and ensure that the distance from the data point closest to the hyperplane to the hyperplane is maximized to achieve nonlinear prediction of temperature-induced deflection; Let x p is the input feature vector, y p is the corresponding target value; w is the normal vector of the hyperplane; m is the intercept of the hyperplane; ε p is a slack variable used to process the error and distribution of data points; C is a regularization parameter used to control the complexity of the regressor; i is the number of calculations, q is the upper limit of the number of calculations, i = 1, 2, ..., q; the nonlinear prediction process of temperature-induced deflection is: S33: Check the accuracy of the prediction model; let m be the number of sample points, g be the current number of calculations, g = 1, 2, ..., m; is the monitoring value of temperature-induced deflection; is the predicted value of temperature-induced deflection; is the average value of the deflection monitoring value; the evaluation index determination coefficient R 2 And the mean absolute error MAE are: If R 2 If it is greater than the preset value and MAE is less than the preset value, the accuracy of the prediction model meets the expected requirements, and subsequent error analysis and safety threshold setting are performed; otherwise, data is collected again and the prediction model is trained.

7. The method for predicting temperature-induced deflection and safety early warning of a concrete continuous rigid frame bridge based on health monitoring according to claim 1 is characterized in that: In the step S4, the specific steps are: S41: Analysis of prediction error of temperature-induced deflection prediction model: S42: According to the statistical law of prediction error Δ, the safety warning threshold of concrete continuous rigid frame bridge is determined based on the 3-sigma principle; let Φ be the cumulative distribution function; β be the safety warning threshold δ x The significance level is used for operators to adjust to achieve multi-level early warning of bridge status; S is the predicted value of temperature-induced deflection; α is the standard deviation magnification factor, which is determined by the inverse function of the cumulative distribution function; σ x' is the standard deviation of the prediction error; the upper limit UWT and lower limit LWT of the safety warning threshold are: Φ(UWT≥δ x ≥LWT)=β (8), a=Φ -1 (b) (9), UWT=y S +as x' (10), LWT=y S -as x' (11).

8. The method for predicting temperature-induced deflection and safety early warning of a concrete continuous rigid frame bridge based on health monitoring according to claim 7 is characterized in that: The step S42 further includes the following steps: adjusting the standard deviation magnification coefficient α to achieve multi-level warning of the bridge safety status.

9. A computer memory, characterized in that: A computer program executable by a computer processor is stored therein, and the computer program executes a method for predicting temperature-induced deflection and providing safety warning for a concrete continuous rigid frame bridge based on health monitoring as described in any one of claims 1 to 8.

10. A temperature-induced deflection prediction and safety warning system for a concrete continuous rigid frame bridge based on health monitoring, the system comprising a processor and a memory, characterized in that: The memory stores computer instructions, and the processor is used to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of a method for predicting temperature-induced deflection and safety warning of a concrete continuous rigid frame bridge based on health monitoring as described in any one of claims 1 to 8.

Images