Logistic regression model-based bridge health state real-time monitoring method and system

Abstract

The invention discloses a bridge health state real-time monitoring method and system based on a logistic regression model, and relates to the technical field of bridge structure health monitoring, and the method comprises the steps: S1, training a logistic regression model based on historical monitoring data and a known bridge health state label; s2, the data acquisition unit acquires monitoring data of the bridge structure in real time; s3, a data processing unit performs preprocessing and feature extraction on the monitoring data obtained in the S2 to obtain feature data S; and S4, the health assessment unit uses the feature data S as the input of the logistic regression model to determine the current health state of the bridge. According to the bridge health state real-time monitoring method and system based on the logistic regression model, the bridge health state can be accurately monitored in real time, potential structure problems can be found in time, and the safety and durability of a bridge are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge structure health monitoring, and more specifically, to a real-time monitoring method and system for bridge health status based on a logistic regression model. Background Art

[0002] Bridges are an important part of modern transportation infrastructure. They not only bear huge loads but also face various environmental impacts, which makes them vulnerable to structural damage. Therefore, real-time monitoring of their health status is of great significance for ensuring traffic safety, reducing accident risks and carrying out maintenance and repair in a timely manner.

[0003] In the existing bridge monitoring process, sensors are usually installed at various important locations to monitor various types of data at each location during the operation of the bridge structure in real time. When analyzing the health status of all locations in the bridge structure, each location in the bridge structure is mainly used as each node in the graph attention network model. The health status of each location in the bridge structure is classified and analyzed to determine the health status of the bridge. However, the above technology still has the following shortcomings: 1. Data collection and processing are not accurate enough Most traditional systems use strain, force and other sensors to monitor bridge structures, without considering the cross-effects of various parameters and lacking multi-dimensional data collection, resulting in incomplete monitoring results. That is, current monitoring systems basically collect single data and draw conclusions based on single analysis, without considering the influence of various parameters, such as the influence of temperature on stress and temperature on strain, and have not achieved truly multi-dimensional data collection.

[0004] At the same time, the noise removal and feature extraction technologies in the data processing process are not mature enough. For example, the mutual influence between different types of signals and data inconsistency may affect the accuracy of the model.

[0005] 2. Model training and updating are not timely Traditional models may lack a dynamic update mechanism and cannot quickly adapt to real-time changes in bridge status. The combination of historical data and real-time data has a lag, which affects the timeliness of model predictions.

[0006] 3. The accuracy of health status assessment is limited Existing assessment systems may use relatively simple models that are difficult to accurately capture complex changes in bridge health status.

[0007] The risk early warning mechanism is not sensitive enough and may miss or misreport potential health problems of bridges. Summary of the invention

[0008] An object of the present invention is to solve at least the above problems and / or disadvantages and to provide at least the advantages which will be described hereinafter.

[0009] In order to achieve these purposes and other advantages of the present invention, a real-time monitoring method for bridge health status based on a logistic regression model is provided, comprising: S1, training a logistic regression model based on historical monitoring data and known bridge health status labels; S2, the data acquisition unit obtains monitoring data of the bridge structure in real time; S3, the data processing unit preprocesses and extracts features from the monitoring data obtained in S2 to obtain feature data S ; S4. The health assessment unit will collect the characteristic data S As the input of the logistic regression model, the current state is analyzed to obtain the corresponding state label y′； If the status mark y′ Comparison feature data set with the logistic regression model S If the marks of 0 are consistent, the bridge is judged to be in a healthy state. Otherwise, there is a health risk and early warning and / or subsequent maintenance are required.

[0010] Preferably, in S1, the known bridge health status label is obtained by: S110, set x 1. x 2… x n They are used to characterize the health status of the bridge. S of n The number of samples will n The number of samples is n A point in a dimensional vector space, then n Generalized equation of a line in dimensional space y It can be characterized by the following formula: In the above formula, i 1. i 2… i i is the model training constant, i 0 is the initial translation constant for model training; S111, select a set of state parameters ( x 0 1. x 0 2… x 0 n ) into the equation of the liney In, if , then specify the state label y =1, the state is a positive example, otherwise y =0, the state is a negative example.

[0011] Preferably, in S1, the training process of the logistic regression model is configured to include: S120, the relationship between the input data and the output results in the logistic regression model is calculated by assuming a function h θ ( x ) to characterize: In the above formula, x 1. x 2. … x n They are used to characterize the health status of the bridge. S of n The number of samples, i 1. i 2. … i i is the model training constant; When predicting the category of a sample using logistic regression, a threshold is set in advance p ,like h θ ( x )≥ p When , the sample is a positive example, otherwise it is a negative example; S121, through the cost function J ( i ) is used to measure the error between the sample prediction value and the true value of the known sample state label during model training: In the above formula, x (i) For the i Sample parameters, y (i) For the i The labels corresponding to the sample parameters; when J ( i ) to the minimum i The combination is used as a linear classification parameter in the logistic regression model to update the threshold p ; S122: Update the threshold value p Substitute into the logistic regression model to establish comparative feature data S 0.

[0012] Preferably, in S3, the preprocessing refers to denoising the original data; The feature extraction refers to extracting strain amplitude, displacement change rate, and temperature change information from the denoised data.

[0013] A real-time monitoring system for bridge health status, which is applied to a real-time monitoring method for bridge health status based on a logistic regression model, comprises: A data acquisition unit for obtaining various monitoring data of the bridge structure; A data processing unit for preprocessing and feature extraction of various collected monitoring data; Model training unit for building logistic regression models; Health assessment unit for analysis of current bridge status based on logistic regression model.

[0014] Preferably, the data acquisition unit includes: strain sensors, displacement sensors and temperature sensors arranged at various locations of the bridge body, bridge frame and bridge pier; The data processing unit, health assessment unit and model training unit are all integrated into a remote central data processing system.

[0015] The present invention includes at least the following beneficial effects: By designing a real-time monitoring system for bridge health status based on a logistic regression model, the present invention can monitor the health status of the bridge in real time and accurately, discover potential structural problems in time, and improve the safety and durability of the bridge.

[0016] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the real-time monitoring method of bridge health status based on the logistic regression model of the present invention. DETAILED DESCRIPTION

[0018] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0019] The present invention relates to a real-time monitoring system for the health status of a bridge based on a logistic regression model. The system combines sensor data and a logistic regression model to evaluate the health status of a bridge structure through real-time data collection and processing.

[0020] Specifically, the real-time monitoring system of bridge health status based on logistic regression model includes: A data acquisition unit is used to obtain various monitoring data of the bridge structure, such as strain, displacement and temperature, etc. The data acquisition unit includes: strain sensors, displacement sensors and temperature sensors arranged at various locations of the bridge body, bridge frame and bridge pier; A data processing unit, used for preprocessing and feature extraction of the collected data; Model training unit, which trains the logistic regression model by using historical data and updates the model in real time; The health assessment unit analyzes the current state based on the logistic regression model and provides a health assessment report and early warning information. The data processing unit, the health assessment unit, and the model training unit are all integrated in a remote central data processing system.

[0021] The present invention aims to provide an efficient and accurate bridge health monitoring method, the working process of which is as follows: Figure 1 As shown, the specific working method is as follows: Step 1: Data Collection The present invention collects corresponding monitoring data at key parts of the bridge by deploying a variety of sensors (such as strain sensors, displacement sensors, and temperature sensors). The present invention realizes data collection in multiple dimensions during bridge monitoring through a variety of sensors. At the same time, the hypothesis function designed in the later stage of this model is used to comprehensively consider the influence of temperature on stress and temperature on strain. That is, the joint influence of multiple parameters is considered through the designed hypothesis function to minimize the cost function (simply understood as error), thereby realizing multi-dimensional data collection in a true sense.

[0022] The data collected by the sensors are transmitted to the central data processing system via a wireless network.

[0023] Step 2: Data processing The data processing unit performs denoising on the raw data to ensure the accuracy of the data, and extracts key features (such as strain amplitude, displacement change rate, temperature change, etc.) as input to the logistic regression model.

[0024] Step 3: Model training The logistic regression model is trained using historical monitoring data and known bridge health status labels. The specific process is as follows: 1. Linear Classification Bridge health status S Depend on n Parameters (bridge measured data x 1, x 2,…, x n ) decides, then this n Parameters n A point in a dimensional vector space, let nA generalized straight line in a dimensional space (hereinafter referred to as a straight line) ,in, i i is the model training constant, i 0 is the initial translation constant for model training, which can be adjusted according to the model training of different bridge types. Figure 1 middle, x i is the measured data of the bridge, such as stress, strain, etc., and i=1,2,3,4… n ,in addition Figure 1 In represents the data weighting process, Represents the process of outputting results during model training. w i is weighted data, Represents the output of the model after training.

[0025] 2. Logistic regression hypothesis function By assuming that the function h θ ( x ), to express the relationship between input data and output results. The hypothesis function of the logistic regression model is: The unknown sample x Input to the logistic regression model, and then the model predicts whether the sample is a positive or negative example. When logistic regression predicts the category of the sample, a threshold is set in advance p , which is used to control the classification boundary and determine whether the bridge is currently in a healthy state: h θ ( x )≥ p It is a positive example when it is detected, otherwise it is a negative example. In this step, automatic updating is used as one of the basic features of artificial intelligence algorithms. This allows the model to continuously optimize parameter selection during operation based on the logistic regression algorithm, thereby solving the technical problem that traditional models in the prior art may lack a dynamic updating mechanism. In addition, this model can be based on the cost function, through the input parameters, combined with the game process of the "cost" function, to minimize the error, thereby realizing the monitoring of complex states and accurately capturing complex changes in the health status of bridges.

[0026] 3. Cost function of logistic regression The cost function is used to measure the size of the error made by the model on the training set, that is, the error between the sample prediction value and the true value.

[0027] in, J ( i )forn The average cost of samples is C i is the cost of a sample, x (i) For the i Sample parameters, y (i) For the i The label corresponding to each sample parameter.

[0028] J ( i )Function design: h θ ( x ) represents the predicted probability that the sample belongs to a certain category, y is the label of the sample, then C The function represents the size of the error made by the model for a sample. When the sample is a positive example, the mark y =1, otherwise mark y =0, when h θ ( x ) is closer to 0, C The larger the size, as shown below: Then we have: By using the above equation, the average error J ( i )The smallest θ combination can be used as the linear classification parameter in the logistic regression model. By iterating the parameters, the optimal linear combination parameters can be achieved. In step three, because the cost function of logistic regression is introduced, the model of the present invention corrects the factors such as "the noise removal and feature extraction technology in the data processing process is not mature enough, such as the mutual influence between different types of signals and data inconsistency" by minimizing the "cost".

[0029] Furthermore, the logistic regression model used in this model assumes that the function h θ ( x ) is a parameter-sensitive function, which will significantly amplify the change in parameters, thereby greatly improving the sensitivity of this model. At the same time, the sensitivity of parameters can be further improved through the "game" process of the cost function.

[0030] Step 4: Health status assessment (1) Specify the judgment standard threshold according to relevant regulations p , and substitute it into the model to determine whether the state (S0) that meets the standard requirements is a positive example or a negative example.

[0031] (2) Input the real-time collected feature data (state S) into the logistic regression model for analysis to obtain the mark of the current state y′ .

[0032] If the S and S0 marks are the same, the current health status is normal, otherwise there is a health risk, and early warning and subsequent maintenance are carried out.

[0033] The above solution is only an illustration of a preferred embodiment, but is not limited thereto. When implementing the present invention, appropriate replacement and / or modification can be performed according to user needs.

[0034] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and the embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily realized. Therefore, without departing from the general concept defined by the claims and equivalent scope, the present invention is not limited to the specific details and the illustrations shown and described here.

Claims

1. A real-time monitoring method for bridge health status based on a logistic regression model, characterized in that: include: S1, training a logistic regression model based on historical monitoring data and known bridge health status labels; S2, the data acquisition unit obtains monitoring data of the bridge structure in real time; S3, the data processing unit preprocesses and extracts features from the monitoring data obtained in S2 to obtain feature data S ; S4. The health assessment unit will collect the characteristic data S As the input of the logistic regression model, the current state is analyzed to obtain the corresponding state label y′； If the status mark y′ Comparison feature data set with the logistic regression model S If the marks of 0 are consistent, the bridge is judged to be in a healthy state. Otherwise, there is a health risk and early warning and / or subsequent maintenance are required.

2. The real-time monitoring method for bridge health status based on logistic regression model according to claim 1 is characterized in that: In S1, the known bridge health status labels are obtained in the following way: S110, set x 1. x 2… x n They are used to characterize the health status of the bridge. S of n The number of samples will n The number of samples is n A point in a dimensional vector space, then n Generalized equation of a line in dimensional space y It can be characterized by the following formula: In the above formula, θ 1. θ 2… θ i is the model training constant, θ 0 is the initial translation constant for model training; S1 11. Select a set of state parameters from the historical monitoring data ( x 0 1. x 0 2… x 0 n ) into the equation of the line y In, if , then specify the state label y =1, the state is a positive example, otherwise y =0, the state is a negative example.

3. The real-time monitoring method for bridge health status based on logistic regression model as claimed in claim 2 is characterized in that: In S1, the training process of the logistic regression model is configured to include: S120, the relationship between the input data and the output results in the logistic regression model is calculated by assuming a function h θ ( x ) to characterize: In the above formula, x 1. x 2. … x n They are used to characterize the health status of the bridge. S of n The number of samples, θ 1. θ 2. … θ i is the model training constant; When predicting the category of a sample using logistic regression, a threshold is set in advance p ,like h θ ( x )≥ p When , the sample is a positive example, otherwise it is a negative example; S121, through the cost function J ( θ ) is used to measure the error between the sample prediction value and the true value of the known sample state label during model training: In the above formula, x (i) For the i Sample parameters, y (i) For the i The labels corresponding to the sample parameters; when J ( θ ) to the minimum θ The combination is used as a linear classification parameter in the logistic regression model to update the threshold p ; S122: Update the threshold value p Substitute into the logistic regression model to establish comparative feature data S 0.

4. The real-time monitoring method for bridge health status based on logistic regression model according to claim 1, characterized in that: In S3, the preprocessing refers to denoising the original data; The feature extraction refers to extracting strain amplitude, displacement change rate, and temperature change information from the denoised data.

5. A real-time monitoring system for bridge health status, which is applied to the real-time monitoring method for bridge health status based on a logistic regression model as claimed in any one of claims 1 to 4, characterized in that: include: A data acquisition unit for obtaining various monitoring data of the bridge structure; A data processing unit for preprocessing and feature extraction of various collected monitoring data; Model training unit for building logistic regression models; Health assessment unit for analysis of current bridge status based on logistic regression model.

6. The real-time monitoring system for bridge health status according to claim 5, characterized in that: The data acquisition unit includes: strain sensors, displacement sensors and temperature sensors arranged at various locations of the bridge body, bridge frame and bridge pier; The data processing unit, health assessment unit and model training unit are all integrated into a remote central data processing system.

Images