Keywords

1 Introduction

The safety and reliability of civil engineering structures have always been the focus in the field of engineering construction Important concerns Point. Traditional civil engineering structure monitoring methods, such as manual inspection and regular detection, have some limitations. First, manual inspection requires a lot of time and manpower investment, and it is difficult to achieve a comprehensive monitoring of the structure. Secondly, regular detection can only provide discrete data points, which can not reflect the health status of the structure in real time, so it may miss the early signals of structural abnormalities [1]. In addition, the traditional monitoring methods also have some difficulties in the detection and diagnosis of complex structural problems, such as micro deformation and hidden damage.

Civil engineering structures carry various loads, such as natural environment (such as wind, earthquake, etc.) and human activities (such as vehicles, pedestrians, etc.), and are affected by climate change, material fatigue, structural aging and other factors for a long time [2]. These external and internal factors may cause damage, deformation or fatigue of the structure, and then adversely affect the safety and reliability of the structure. Among them, human activities such as vehicle traffic and weight placement, as a common source of load, the pressure on the structure can not be ignored. At the same time, with the passage of time, structural aging has also become a problem that can not be ignored. The combined action of these factors may lead to damage, deformation and even destruction of civil engineering structures, which will pose a serious threat to their safety and reliability.

In order to deeply explore the impact of these factors on civil engineering structures, this paper is committed to real-time and accurate detection of the state of the structure through the health monitoring and evaluation method based on machine learning. Machine learning algorithm can intelligently process and analyze a large number of sensor data, extract the characteristics of structural health state, and establish the model related to structural health. By learning the known structural health status and sensor data, the machine learning model can realize the real-time and accurate monitoring and evaluation of structural health status [3]. The practical significance of this research is that it can not only improve the safety of civil engineering structures, but also reduce the economic loss and social impact caused by structural damage, which has important theoretical value and practical significance.

2 Materials and Methods

2.1 Algorithm and Structural Damage Detection

Algorithm

First of all, as for the selection of machine learning algorithms, I plan to use two algorithms: support vector machine (SVM) and random forest. SVM algorithm has good performance in classification problem. It can separate different categories of samples by finding a hyperplane, and maximize the distance between different categories of samples. Random forest is an integrated learning method, which can improve the stability and accuracy of the model by constructing multiple decision trees and integrating their prediction results. The combination of these two algorithms can overcome the limitations of a single algorithm to a certain extent and improve the generalization ability of the model.

Structural Damage Detection

The most unfavorable combination means This part Impending Structural damage, When the structure is damaged, the structural parameters such as mass, stiffness and damping will change to varying degrees, which will lead to the changes of natural frequency, mode shape and impedance of the structure. Many existing damage detection methods are proposed based on these parameters. Zong Zhou Hong and others reviewed the research progress of civil engineering structural damage from the aspects of damage diagnosis, system identification, model modification and sensor layout; Rytter divided damage identification into four progressive levels in his doctoral dissertation: the first level determines whether there is damage in the structure; The second level determines the geometric location of the damage based on the first level; The third level quantifies the severity of damage on the basis of the second level; The fourth level predicts the remaining service life of the structure on the basis of the third level. Farrar and Worden divided damage identification into five processes, including SHM condition monitoring, NDT evaluation, statistical process control and damage prediction [4]. Giraldo divided damage detection and identification into three types in his doctoral dissertation: vibration based method, static based method and direct detection method of structure.

Structural damage detection can be transformed into a mathematical optimization problem in many cases. Genetic algorithm based on Darwin's natural selection and genetic theory is a typical global optimization method. Compared with traditional optimization methods, it has its own characteristics: it has the ability of self-organization, self-adaptive and self-learning, and can automatically discover environmental characteristics and laws according to environmental changes, so it can solve complex unstructured problems; It has essential parallelism and is very suitable for large-scale parallel operation; The search range is wide, which greatly reduces the possibility of falling into the local optimal solution; When solving various kinds of problems, only the objective function needs to be defined, without gradient and other traditional information [5]. Due to the above characteristics of simulating the natural evolution process, genetic algorithm has unique advantages in solving the multi peak complex objective function in damage identification.

Dynamic Control of Structural Damage

As shown in Fig. 1: the dynamic control chart of civil engineering structure health monitoring based on machine learning shows a comprehensive and fine maintenance process. Through GPS data collection by sensors, the system can obtain the state information of the structure in real time. These data are then compared with the state of the initial healthy structure, and the machine learning algorithm is used to analyze a large number of data, so as to accurately identify the abnormal changes or potential damage of the structure [6]. Once the damage affecting the structural safety is found, the system will quickly diagnose the damage, and further use the machine learning model to predict the development trend and impact degree of the damage.

Based on the results of damage diagnosis, the system will intelligently formulate maintenance plans and conduct cost assessment to ensure the economy and efficiency of maintenance work. Subsequently, the actual structural maintenance and repair work are carried out according to the plan to restore the structure to a healthy state. After the maintenance, the structure will enter the operation state again. At this time, the system will continue to conduct data collection and condition monitoring to ensure the long-term safety and stability of the structure [7].

This dynamic control process fully reflects the application value of machine learning in the health monitoring and evaluation of civil engineering structures, which not only improves the accuracy and efficiency of monitoring, but also provides scientific decision support for the maintenance and management of structure.

Fig. 1.
figure 1

Dynamic control process of structural health monitoring system

Full size image

2.2 Structural Health Detection System

Figure 2 shows the full cycle maintenance and health monitoring system of building structure from design and construction to operation and use. In the early stage of the use of building structures, due to the short service life and high structural reliability, large-scale structural inspection and maintenance are not frequent. However, with the increase of service life, minor damages gradually accumulate. If these damages are not monitored and treated in time, they may pose a threat to the overall safety of the structure.

Therefore, it is necessary to introduce the structural health monitoring system. By installing the sensing system, we can monitor the state of the structure in real time, and the preventive maintenance management based on performance design can be implemented [8]. Such a system can not only reduce the cost increase caused by emergencies or improper maintenance, but also ensure the long-term safety and stability of the building structure in the use process.

Compared with the traditional maintenance management based on structural state, the performance-based structural health monitoring system pays more attention to the overall performance and long-term benefits of the structure. Through real-time data monitoring and analysis, we can more accurately predict the life of the structure and possible problems, so as to take corresponding maintenance measures in advance to ensure the safety and reliability of the building structure.

Fig. 2.
figure 2

Full cycle maintenance and health monitoring system

Full size image

To sum up, the structural health monitoring system plays an increasingly important role in the maintenance and management of building structures [9]. Through the implementation of performance-based preventive maintenance management, we can realize the effective monitoring and maintenance of the building structure in the whole cycle, and ensure its safety and stability in the long-term use process.

3 Technology and Application

According to the real-time monitoring requirements of civil engineering structures, Real time monitoring on the machine, We will obtain the real-time data of the floor structure under different loads and environmental conditions. These data will cover the displacement, stress, strain, vibration and other aspects of the structure to comprehensively evaluate the performance and safety of the structure. First, we will monitor the displacement of the floor structure in real time through high-precision displacement sensors. The main beam calculation diagram shown in Fig. 3, These sensors will be installed in key positions, such as beams, plates and other stressed parts, to record the deformation of the structure under load in real time [10]. Through the analysis of displacement data, we can understand the overall stability and deformation trend of the structure, and timely find the potential safety hazards. Secondly, we will use the stress-strain sensor to monitor the stress and strain of the structure in real time. These sensors will be directly installed on materials such as steel and concrete to obtain the stress and strain distribution inside the structure. As shown in Tables 1 and 2, we will collect data on the bending moment and shear force of the main beam and other load-bearing parts that affect the structural stress and strain. Through the analysis of real-time data, we can understand the stress state of the structure under load, make a bending and shear envelope diagram as shown in Fig. 4, and identify the most unfavorable combination of internal forces, judge whether there are problems such as overload or fatigue, and take corresponding measures to repair or strengthen in time. In addition, vibration monitoring is also an important aspect of real-time monitoring of civil engineering structures. We will monitor the vibration of the structure in real time through acceleration sensors and other equipment to obtain the vibration frequency, amplitude and other parameters of the structure. These data can help us understand the response characteristics of the structure under dynamic load, evaluate the seismic performance of the structure, and provide a scientific basis for the optimal design and reinforcement of the structure. In the process of real-time monitoring, we will also consider the impact of environmental factors on structural performance. For example, environmental factors such as temperature and humidity may lead to changes in the performance of materials, and then affect the overall performance of the structure. Therefore, we will carry out real-time monitoring of environmental factors through temperature and humidity sensors and other equipment, and take environmental factors into account in data analysis, so as to more accurately evaluate the performance of the structure [11]. To sum up, by monitoring the displacement, stress, strain, vibration and other data of civil engineering structures in real time, and considering the environmental factors, we can comprehensively evaluate the performance and safety of the structure, and provide a scientific basis for the maintenance and management of the structure. This will help to find and solve potential safety hazards in time, and improve the reliability and durability of civil engineering structures. The following are obtained under the machine monitoring calculation Data and calculations, thus To judge structural health.

Fig. 3.
figure 3

Calculation diagram of main beam

Full size image
Table 1. Bending moment design value of each section of main beam
Full size table
Table 2. Design value of shear force of each section of main beam
Full size table
Fig. 4.
figure 4

Bending moment envelope diagram and shear force envelope diagram

Full size image

4 Conclusions

Industrialized countries have invested a lot of money in the development of civil infrastructure. In order to maintain these investment values, proper maintenance must be paid attention to. SHM has emerged as a tool to support this work. Although many damage detection methods based on structural vibration response and system dynamic parameters have been developed, there are still many difficulties in the practical application of these methods due to the complexity of structural damage and the uncertainty of various influencing factors. SHM based on interdisciplinary will be a more advanced technology. In addition to profound structural knowledge, it also needs to understand the knowledge of other related disciplines [12]. Only the combination of structural vibration theory and signal processing, pattern recognition, artificial intelligence, control theory and material science can improve the accuracy of structural damage detection.

The practical significance of interdisciplinary methods in the field of structural health monitoring is very important. Firstly, by integrating the knowledge and technology of multiple disciplines, we can use the data fusion analysis of different types of sensors and comprehensively consider various factors, so as to improve the accuracy of structural damage identification of the monitoring system. Secondly, the application of intelligent algorithm can help to establish intelligent model, analyze and diagnose the monitoring data automatically, improve the monitoring efficiency and reduce the workload of manual intervention. In addition, there are complex interactions between the structure and the surrounding environment. Through the study of coupling effect, we can better understand the behavior of the structure under different environmental conditions, and improve the depth and accuracy of structural damage detection. In general, the application of interdisciplinary methods will promote the continuous innovation and progress of structural health monitoring technology, bring more opportunities and challenges to the field of building structures, and provide solid support for the sustainable development and safe operation of structures.

From the above analysis and summary of the latest literature on SHM and structural damage detection, we can see that there are still many problems to be solved, which need our continuous research and practice.

1) Multimodal data fusion and analysis: future research can focus on how to better integrate multimodal data obtained by different types of sensors, and develop more efficient data fusion and analysis methods. The importance of this field is that different types of sensors provide complementary information, and the comprehensive use of this information can improve the accuracy and reliability of structural damage detection. Solving this challenge will promote the SHM field to develop towards a more comprehensive and accurate monitoring direction.

2) Intelligent diagnosis and prediction model: future research can focus on developing more intelligent diagnosis and prediction models, combining machine learning, deep learning and other technologies to achieve more accurate and timely prediction of structural health. The importance of this field is that intelligent algorithms can improve the efficiency and automation level of the monitoring system, and provide more timely protection for structural safety. Solving this challenge will promote the SHM field to be intelligent and automated.

3) Research on structural behavior under coupling effect: future research can focus on the complex coupling effect between the structure and the surrounding environment, and further study the influence of these influencing factors on structural behavior. The importance of this field is that the coupling effect is a factor that can not be ignored in structural health monitoring. Only by fully understanding the coupling effect can we more accurately evaluate the health status of structures. Solving this challenge will promote the SHM field to have a deeper understanding of the impact of the environment around the structure.