Structural hidden defect identification system based on nonlinear characteristics of vibration signals

Abstract

This invention relates to the field of defect identification technology and discloses a system for identifying hidden structural defects based on the nonlinear characteristics of vibration signals. The system includes a system module assembly, comprising a vibration signal collection module connected to an analysis module. The analysis module employs a high-precision time-frequency analysis method and, after analyzing the collected vibration signals, obtains a time-series spectrum module. This time-series spectrum module is connected to a detection module, which includes a ridge detection algorithm module. The ridge detection algorithm module obtains frequency and energy curves or skeleton curves reflecting the nonlinear characteristics of the structure based on the instantaneous characteristics of the ridge identification signal. This invention uses the time-series spectrum obtained by the high-precision time-frequency analysis method as data samples for deep learning. It trains the data samples and extracts features by constructing models such as convolutional neural networks, and achieves intelligent localization and quantitative analysis of hidden defects based on the extracted features.

Description

Technical Field

[0001] This invention relates to the field of defect identification technology, and in particular to a structural hidden defect identification system based on the nonlinear characteristics of vibration signals. Background Technology

[0002] Steel-concrete composite structures are widely used as load-bearing components due to their excellent structural performance. However, due to the exposure to the environment, the structure inevitably suffers from the vibration and impact of external loads, periodic temperature, and other effects. At the same time, due to unscientific casting processes and other construction defects and concrete shrinkage, hidden defects such as bond slip and debonding may occur at the interface of the steel-concrete composite structure. The dynamic characteristics of the structure usually exhibit nonlinear characteristics.

[0003] To address the difficulty in detecting hidden defects at the interface of steel-concrete composite structures, it is essential to propose a structural hidden defect identification system based on the nonlinear characteristics of vibration signals. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] To address the shortcomings of existing technologies, this invention provides a structural hidden defect identification system based on the nonlinear characteristics of vibration signals, which solves the problem of difficulty in detecting hidden defects at the interface of steel-concrete composite structures.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A structural hidden defect identification system based on the nonlinear characteristics of vibration signals includes a system module assembly. The system module assembly includes a vibration signal collection module connected to an analysis module. The analysis module employs a high-precision time-frequency analysis method and obtains a time-series spectrum module after analyzing the collected vibration signals. The time-series spectrum module is connected to a detection module, which includes a ridge detection algorithm module. This ridge detection algorithm module obtains frequency and energy curves or skeleton curves reflecting the nonlinear characteristics of the structure based on the instantaneous characteristics of the ridge detection signal. The system module assembly also includes a model building module, which includes a learning module based on a convolutional neural network. The learning module includes a data collection module and a feature extraction module. The data collection module is connected to the time-series spectrum module, and the data collection module is connected to the feature extraction module. The feature extraction module is connected to a training module and a modeling module. After model building, the modeling module trains the model through the training module. The training module is connected to a validation module, which uses features extracted by the feature extraction module to train and validate the model, evaluating its accuracy. The validation module is also connected to a testing module, which tests and evaluates the model's generalization ability.

[0009] Based on the aforementioned scheme, the high-precision time-frequency analysis method extends the analysis of the one-dimensional time-domain signal collected by the vibration signal collection module to a two-dimensional time-frequency domain plane.

[0010] Furthermore, the high-precision time-frequency analysis method is one of the following: synchronous compressed wavelet transform, synchronous compressed short-time Fourier transform, time-frequency rearrangement, or local maximization synchronous compressed transform.

[0011] As a further embodiment of the present invention, the ridge detection algorithm module obtains the instantaneous frequency of the vibration signal through the time-series spectrum module, and reconstructs the signal through the obtained instantaneous frequency ridge, thereby obtaining the amplitude of the signal.

[0012] Based on the aforementioned scheme, the instantaneous characteristics include instantaneous frequency, instantaneous damping, and instantaneous amplitude.

[0013] Furthermore, the system module assembly includes a signal transmission module, which is connected to the time-series spectrogram module and the learning module.

[0014] As a further embodiment of the present invention, the system module assembly includes an information sending module, an information receiving module, an information processing module, and a storage module, and the storage module is connected to a power supply module.

[0015] The beneficial effects of this invention are as follows:

[0016] 1. In this invention, a high-precision time-frequency analysis method is used to analyze the vibration response of the structure, and a high-resolution time-frequency spectrum of the non-stationary vibration response signal of the interface peeling is obtained. This solves problems such as ridge line extraction and nonlinear instantaneous feature identification of the signal, and links the interface peeling problem with nonlinear feature identification to determine whether there is damage to the interface.

[0017] 2. The deep learning technology of this invention can realize intelligent target detection and damage classification. It uses the time-frequency spectrum obtained by the high-precision time-frequency analysis method as the data sample for deep learning. It trains the data sample and extracts features by constructing models such as convolutional neural networks. Based on the extracted features, it realizes intelligent localization and quantitative analysis of hidden defects.

[0018] 3. In this invention, the high-precision time-frequency analysis method extends the analysis of the one-dimensional time-domain signal collected by the vibration signal collection module to a two-dimensional time-frequency domain plane. This can better reflect the subtle changes in the energy of the signal with time and frequency, reflect the instantaneous characteristics of the structure, obtain a high-definition time-frequency spectrum, and further extract high-precision time-frequency ridges, providing a practical method for the identification of nonlinear systems.

[0019] 4. This invention solves the problem that the interface of SCCS is prone to peeling. When the structure vibrates, the complex effects of interface slippage, debonding, and loosening can cause the structure to exhibit nonlinear vibration characteristics with instantaneous frequency changing with amplitude. By identifying nonlinear characteristics, the invention reflects whether there is a peeling problem at the interface. The nonlinear characteristics of the defect are obtained by using time-frequency analysis. Deep learning technology is applied to reveal the different dynamic behaviors of the defect under the coupling of multiple loads and environment, so as to achieve the purpose of intelligent identification of hidden defects. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the system framework of the structural hidden defect identification system based on the nonlinear characteristics of vibration signals proposed in this invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. It should be noted that, unless otherwise expressly specified and limited, the terms "installation", "connection", and "setting" should be interpreted broadly. For those skilled in the art, the specific meaning of the above terms in this patent can be understood according to the specific circumstances.

[0022] Example 1

[0023] Reference Figure 1A structural hidden defect identification system based on the nonlinear characteristics of vibration signals includes a system module assembly. The system module assembly includes a vibration signal collection module connected to an analysis module. The analysis module employs a high-precision time-frequency analysis method and, after analyzing the collected vibration signals, obtains a time-series spectrum module. The time-series spectrum module is connected to a detection module, which includes a ridge detection algorithm module. This ridge detection algorithm module obtains frequency and energy curves or skeleton curves reflecting the nonlinear characteristics of the structure based on the instantaneous characteristics of the ridge identification signal. It uses a high-precision time-frequency analysis method to analyze the vibration response of the structure, obtaining a high-resolution time-spectrum map of the nonstationary vibration response signal after interface peeling. This solves problems such as ridge extraction and identification of instantaneous nonlinear signal features, linking the interface peeling problem with nonlinear feature identification to determine whether damage exists at the interface. The system module assembly also includes a model building module. The model building module includes a learning module based on convolutional neural networks. This learning module comprises a data collection module and a feature extraction module. The data collection module is connected to the time-series graph module, and the data collection module is also connected to the feature extraction module. The feature extraction module is connected to both the training and modeling modules. After model building, the modeling module trains the model through the training module. The training module is connected to a validation module, which uses features extracted by the feature extraction module for training and validation to evaluate model accuracy. The validation module is also connected to a testing module, which tests and evaluates the model's generalization ability. Deep learning technology can achieve intelligent target detection and damage classification. High-precision time-frequency analysis methods are used to obtain time-spectrum graphs as data samples for deep learning. Convolutional neural networks and other models are constructed to train the data samples and extract features. Based on the extracted features, intelligent localization and quantitative analysis of hidden defects are achieved.

[0024] In this invention, the high-precision time-frequency analysis method extends the analysis of one-dimensional time-domain signals collected by the vibration signal collection module to a two-dimensional time-frequency domain plane. This allows for a better reflection of the subtle changes in signal energy with time and frequency, and of the instantaneous characteristics of the structure. The high-precision time-frequency analysis method is a synchronous compressed wavelet transform method. These algorithms have high resolution, can obtain high-definition time-frequency spectra, and can further extract high-precision time-frequency ridges, providing a practical method for the identification of nonlinear systems.

[0025] Example 2

[0026] Reference Figure 1A structural hidden defect identification system based on the nonlinear characteristics of vibration signals includes a system module assembly. The system module assembly includes a vibration signal collection module connected to an analysis module. The analysis module employs a high-precision time-frequency analysis method and, after analyzing the collected vibration signals, obtains a time-series spectrum module. The time-series spectrum module is connected to a detection module, which includes a ridge detection algorithm module. This ridge detection algorithm module obtains frequency and energy curves or skeleton curves reflecting the nonlinear characteristics of the structure based on the instantaneous characteristics of the ridge identification signal. It uses a high-precision time-frequency analysis method to analyze the vibration response of the structure, obtaining a high-resolution time-spectrum map of the nonstationary vibration response signal after interface peeling. This solves problems such as ridge extraction and identification of instantaneous nonlinear signal features, linking the interface peeling problem with nonlinear feature identification to determine whether damage exists at the interface. The system module assembly also includes a model building module. The model building module includes a learning module based on convolutional neural networks. This learning module comprises a data collection module and a feature extraction module. The data collection module is connected to the time-series graph module, and the data collection module is also connected to the feature extraction module. The feature extraction module is connected to both the training and modeling modules. After model building, the modeling module trains the model through the training module. The training module is connected to a validation module, which uses features extracted by the feature extraction module for training and validation to evaluate model accuracy. The validation module is also connected to a testing module, which tests and evaluates the model's generalization ability. Deep learning technology can achieve intelligent target detection and damage classification. High-precision time-frequency analysis methods are used to obtain time-spectrum graphs as data samples for deep learning. Convolutional neural networks and other models are constructed to train the data samples and extract features. Based on the extracted features, intelligent localization and quantitative analysis of hidden defects are achieved.

[0027] In this invention, the high-precision time-frequency analysis method extends the analysis of one-dimensional time-domain signals collected by the vibration signal collection module to a two-dimensional time-frequency domain plane. This allows for a better reflection of the subtle changes in signal energy with time and frequency, and of the instantaneous characteristics of the structure. The high-precision time-frequency analysis method employs a local maximization synchronous compression transform algorithm. These algorithms have high resolution, enabling the acquisition of high-definition time-frequency spectra and the extraction of high-precision time-frequency ridges, thus providing a practical method for the identification of nonlinear systems.

[0028] In particular, the ridge detection algorithm module obtains the instantaneous frequency of the vibration signal through the time-series spectrum module, and reconstructs the signal by using the obtained instantaneous frequency ridge, thereby obtaining the signal amplitude. This solves the problem that the interface of SCCS is prone to peeling. When the structure vibrates, the complex effects of interface slippage, debonding, loosening and other factors cause the vibration of the structure to exhibit nonlinear vibration characteristics in which the instantaneous frequency changes with the amplitude. By identifying nonlinear characteristics, the algorithm reflects whether there is a peeling problem at the interface. The instantaneous characteristics include instantaneous frequency, instantaneous damping and instantaneous amplitude.

[0029] The system module assembly includes a signal transmission module, which is connected to the time-series spectrum module and the learning module. It uses time-frequency analysis to obtain the nonlinear characteristics of defects and applies deep learning technology to reveal the different dynamic behaviors of defects under the coupling of multiple loads and environment, so as to achieve the purpose of intelligent identification of hidden defects. The system module assembly includes an information sending module, an information receiving module, an information processing module and a storage module, and the storage module is connected to the power supply module.

[0030] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A structural hidden defect identification system based on the nonlinear characteristics of vibration signals, comprising a system module assembly, characterized in that, The system module assembly includes a vibration signal collection module connected to an analysis module. The analysis module employs a high-precision time-frequency analysis method and, after analyzing the collected vibration signals, obtains a time-series spectrum module. This time-series spectrum module is connected to a detection module, which includes a ridge detection algorithm module. This algorithm module obtains frequency and energy curves or skeleton curves reflecting the nonlinear characteristics of the structure based on the instantaneous characteristics of the ridge detection signal. The system module assembly also includes a model building module, which includes a learning module based on a convolutional neural network. This learning module includes a data collection module and a feature extraction module. The data collection module is connected to the time-series spectrum module. The data collection module is connected to the feature extraction module, which in turn is connected to the training module and the modeling module. After the model is built, it is trained by the training module, which is also connected to the validation module. The model is trained and validated using the features extracted by the feature extraction module to evaluate its accuracy. The validation module is also connected to the testing module to test and evaluate the model's generalization ability. The high-precision time-frequency analysis method extends the analysis of the one-dimensional time-domain signal collected by the vibration signal collection module to a two-dimensional time-frequency domain plane. The ridge detection algorithm module obtains the instantaneous frequency of the vibration signal through the time-series spectrum module and reconstructs the signal using the obtained instantaneous frequency ridge to obtain the signal amplitude.

2. The structural hidden defect identification system based on the nonlinear characteristics of vibration signals according to claim 1, characterized in that, The high-precision time-frequency analysis method is one of the following: synchronous compressed wavelet transform, synchronous compressed short-time Fourier transform, time-frequency rearrangement, and local maximization synchronous compressed transform.

3. The structural hidden defect identification system based on the nonlinear characteristics of vibration signals according to claim 2, characterized in that, The instantaneous characteristics include instantaneous frequency, instantaneous damping, and instantaneous amplitude.

4. The structural hidden defect identification system based on the nonlinear characteristics of vibration signals according to claim 1, characterized in that, The system module assembly includes a signal transmission module, which is connected to the time-series spectrogram module and the learning module.

5. The structural hidden defect identification system based on the nonlinear characteristics of vibration signals according to claim 1, characterized in that, The system module assembly includes an information sending module, an information receiving module, an information processing module, and a storage module, and the storage module is connected to a power supply module.

Images