Structural damage identification method and system for eliminating temperature interference based on blind source separation

Abstract

The invention discloses a blind source separation-based structural damage identification method and system for eliminating temperature interference, and the method comprises the steps: obtaining displacement time-history response data at a key position of a target structure under environment excitation, obtaining the inherent frequency of each order of the target structure according to the displacement time-history response data, and constructing a structural modal frequency signal; whitening the structural modal frequency signal to obtain a delay covariance matrix after whitening the structural modal frequency signal; eigenvalue decomposition is carried out on the delay covariance matrix after whitening processing, an optimal approximation algorithm is adopted to obtain an optimal eigenvector matrix, and a blind source separation matrix is constructed based on the optimal eigenvector matrix and a whitening matrix during whitening processing; and extracting an estimation signal of a source signal in the structural modal frequency signal according to the blind source separation matrix, and judging the damage development condition of the target structure according to the estimation signal. The method can effectively eliminate the interference of the temperature factor on the damage identification method based on the dynamic characteristics, and has good robustness and anti-noise capability.

Description

Technical Field

[0001] The present invention relates to the technical field of structural health monitoring, and in particular to a structural damage identification method based on blind source separation to eliminate temperature interference. Background Art

[0002] As the total amount of bridge construction in my country continues to increase, bridge health monitoring and maintenance has become a new focus in the industry. With the development and application of communication, sensor technology and big data technology, massive amounts of structural health monitoring data are extracted and stored, which contain important information that can reflect the service status and damage of the structure, and urgently need to be processed by efficient and accurate damage identification algorithms. In actual situations, bridge structures are under the coupling of various natural factors and load factors, and are affected by a large number of unknown external noise factors. Therefore, there are many uncertainties in the real structural state, which makes the damage identification method based on dynamic characteristics unsatisfactory in actual monitoring.

[0003] Among various external interference factors, temperature can significantly change the modal parameters of the structure and is one of the main environmental factors. The phenomenon of modal parameter changes caused by temperature effects masking damage is common. Therefore, establishing a complete identification system that eliminates temperature interference and improves damage identification accuracy is a top priority for promoting the development of bridge damage identification. However, existing structural damage identification methods, including some methods that use blind source separation technology to extract modal frequencies from structural dynamic responses to determine damage, have failed to propose a better way to eliminate temperature interference, and therefore cannot obtain accurate and reliable results in actual engineering problems.

[0004] In addition, it is noted that some scholars have conducted research on using the AMUSE algorithm in blind source separation to eliminate temperature interference. However, the AMUSE algorithm only uses two covariance matrices with different time delays, which is easily interfered by various other factors and has disadvantages in stability and noise resistance. As reflected in the results of blind source separation, this method is difficult to identify small changes in modal frequency, and the results of damage are not directly and obviously expressed, so there is a lot of room for improvement.

[0005] In summary, the existing structural damage identification technology has poor ability to eliminate temperature interference and poor anti-interference ability. Summary of the invention

[0006] The present invention proposes a structural damage identification method and system based on blind source separation to eliminate temperature interference. The method introduces the blind source separation technology in the field of signal processing into a damage identification method using structural dynamic characteristics as damage indicators, and adopts a second-order blind identification (SOBI) algorithm with better stability and functionality to unmix the original structural natural frequency signal and separately separate the structural damage signal. By observing whether the damage signal has jumps, it is judged whether the structure is damaged and the time when the damage occurs, thereby achieving the effect of removing temperature noise interference and accurately judging whether the structure is damaged, and having good robustness and noise resistance.

[0007] The present invention provides a method for identifying structural damage by eliminating temperature interference based on blind source separation, comprising the following steps:

[0008] Obtain the displacement time-history response data at the key positions of the target structure under environmental excitation, obtain the natural frequencies of each order of the target structure based on the displacement time-history response data, and construct the structural modal frequency signal;

[0009] The structural modal frequency signal is whitened to obtain the delay covariance matrix of the structural modal frequency signal after the whitening process; the delay covariance matrix after the whitening process is decomposed by eigenvalues, and the optimal eigenvector matrix is ​​obtained by using the optimal approximation algorithm, and a blind source separation matrix is ​​constructed based on the optimal eigenvector matrix and the whitening matrix during the whitening process;

[0010] The estimated signal of the source signal in the structural modal frequency signal is extracted according to the blind source separation matrix, and the damage development of the target structure is judged according to the estimated signal.

[0011] Preferably, the method for acquiring the structural modal frequency signal comprises the following steps:

[0012] Arrange acceleration sensors at key locations of the target structure to collect displacement time-history response data under environmental excitation;

[0013] The random subspace method or peak picking method is used to obtain the natural frequencies of each order of the target structure and construct the structural modal frequency signal.

[0014] Preferably, the whitening process of the structural modal frequency signal comprises the following steps:

[0015] Determine the structural modal frequency signal as:

[0016] x(t)=As(t)

[0017] Among them, A is the mixing matrix, s(t) is the source signal, and x(t) is the observation signal, that is, the structural modal frequency signal;

[0018] Whiten x(t):

[0019]

[0020] Among them, V is the whitening matrix, removing The correlation of each signal component makes its covariance matrix the unit matrix.

[0021] Preferably, the construction of the blind source separation matrix comprises the following steps:

[0022] The delay covariance matrix of the structural modal frequency signal is determined as:

[0023] R s (τ)=s(t)s T (t-τ)

[0024] The delay covariance matrix of the structural modal frequency signal after whitening is determined as:

[0025]

[0026] Among them, τ takes a set of different values ​​τ1,τ2,τ3,…,τ p , used to represent different time delays;

[0027] Perform eigenvalue decomposition on the delay covariance matrix after whitening and obtain:

[0028]

[0029]

[0030] Among them, U is an orthogonal matrix, that is, the optimal eigenvector matrix; among them, R s (τ i ) is a diagonal matrix;

[0031] According to the delay covariance matrix after whitening, the optimal approximation algorithm is used to find the optimal eigenvector matrix U and obtain the blind source separation matrix W:

[0032] W=U T V.

[0033] Preferably, the estimated signal of the source signal in the structural modal frequency signal is extracted according to the blind source separation matrix, as shown in the following formula:

[0034]

[0035] in, is an estimate of the source signal s(t).

[0036] Preferably, judging the damage development of the target structure according to the estimated signal comprises the following steps:

[0037] Whether the target structure is damaged and the damage development situation are determined based on whether the estimated signal has a jump. The moment when the jump occurs is the moment when the target structure is damaged.

[0038] The present invention also provides a structural damage identification system based on blind source separation to eliminate temperature interference, the system comprising:

[0039] processor;

[0040] a memory having stored thereon a computer program executable on the processor;

[0041] Wherein, when the computer program is executed by the processor, the steps of the structural damage identification method based on blind source separation and eliminating temperature interference are implemented.

[0042] The present invention also provides a computer-readable storage medium, characterized in that a data processing program is stored on the computer-readable storage medium, and when the data processing program is executed by a processor, the steps of the method for identifying structural damage based on blind source separation and eliminating temperature interference are implemented.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The present invention discloses a structural damage identification method and system based on blind source separation to eliminate temperature interference, which breaks the limitation of the existing damage identification technology that fails to eliminate the interference of environmental temperature factors, introduces the blind source separation technology in the field of signal processing into structural damage detection with structural dynamic characteristics as damage indicators, and overcomes the shortcomings of the existing AMUSE algorithm for eliminating temperature interference. The highly robust second-order blind identification (SOBI) algorithm is used to unmix the original structural natural frequency signal, and accurately judge whether the structure is damaged and the time when the damage occurs from the separately extracted structural damage signal, thereby achieving the goal of removing the interference of temperature noise, and exploring a feasible method for eliminating environmental noise in real structural health monitoring data. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a specific flow chart of the method of the present invention;

[0046] Figure 2 It is the temperature change curve of a certain place;

[0047] Figure 3 is the modeling model of the single damaged cantilever beam in this embodiment;

[0048] Figure 4 is the sixth-order modal frequency curve of the single-damaged cantilever beam in this embodiment;

[0049] Figure 5is the blind source separation result of the modal frequency curve of the single damaged cantilever beam in this embodiment;

[0050] Figure 6 is the sixth-order modal frequency curve of the single-damaged cantilever beam considering noise in this embodiment;

[0051] Figure 7 is the blind source separation result of the modal frequency curve of the single damaged cantilever beam considering noise in this embodiment;

[0052] Figure 8 is the modeling model of two damaged cantilever beams considering damage development in this embodiment;

[0053] Fig. 9 is the sixth-order modal frequency curve of the two damaged cantilever beams considering the damage development in this embodiment;

[0054] Fig.10 1 is the blind source separation result of the modal frequency curves of the two damaged cantilever beams considering the damage development in this embodiment. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0056] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0057] like Figure 1 As shown, an embodiment of the present invention provides a method for identifying structural damage by eliminating temperature interference based on blind source separation, comprising:

[0058] S1: Obtain the displacement time-history response data at the key positions of the target structure under environmental excitation, obtain the natural frequencies of each order of the target structure according to the displacement time-history response data, and construct the structural modal frequency signal.

[0059] S2: Whiten the structural modal frequency signal to obtain the delay covariance matrix of the structural modal frequency signal after whitening; perform eigenvalue decomposition on the delay covariance matrix after whitening, use the optimal approximation algorithm to obtain the optimal eigenvector matrix, and construct a blind source separation matrix based on the optimal eigenvector matrix and the whitening matrix during whitening.

[0060] S3: Extract the estimated signal of the source signal in the structural modal frequency signal according to the blind source separation matrix, and judge the damage development of the target structure according to the estimated signal.

[0061] The specific implementation is described in detail below in conjunction with specific embodiments.

[0062] The influence of ambient temperature on the structural modal frequency is mainly reflected in that the temperature changes the elastic modulus of the structural material (such as concrete), and the change in the elastic modulus of the material causes the change in the structural modal frequency; and the modal frequency change value caused by temperature factors is greater than the modal frequency change value caused by damage. The change in the structural modal frequency caused by damage will be masked by the temperature effect. Therefore, it is impossible to directly judge whether the structure is damaged from the modal frequency curve.

[0063] The embodiment simulates the damage identification scenarios of a cantilever beam structure under the influence of a temperature change curve in a certain place, in three cases: single damage, single damage considering noise influence, and double damage developed from a single damage. The modal frequencies of the beam structure in the above three cases are obtained, and blind source separation is applied to eliminate temperature interference and identify structural damage. The detailed steps are as follows:

[0064] Step 1: Under environmental excitation, collect the acceleration sensor data of the structure to obtain the displacement time-history response data, use the random subspace method or the peak picking method to calculate the natural frequencies of each order of the structure, correspond each modal frequency to the acquisition time, and establish the structural modal frequency signal;

[0065] Step 2: Construct a second-order blind identification algorithm in blind source separation technology, and establish a damage identification algorithm based on blind source separation of structural modal frequency signals. The algorithm steps are as follows:

[0066] (1) According to the basic assumptions of mixed signals in blind source separation, we can know that:

[0067] x(t)=As(t)

[0068] Where A is the mixing matrix, s(t) is the source signal, and x(t) is the observed signal;

[0069] (2) Whiten x(t):

[0070]

[0071] Where V is the whitening matrix, removing The correlation of each signal component and make its covariance matrix the unit matrix;

[0072] (3) The delay covariance matrix of the source signal can be written as:

[0073] R s (τ)=s(t)s T (t-τ)

[0074] The delay covariance matrix of the source signal after whitening preprocessing can be written as:

[0075]

[0076] Take a set of different values ​​for τ 1, τ 2, τ 3, ..., τ p For different time delays, we derive:

[0077]

[0078]

[0079] Where U is an orthogonal matrix, and it is known that R s (τ i ) is a diagonal matrix;

[0080] (4) Based on Using the optimal approximation algorithm, find the optimal matrix U and calculate the blind source separation matrix W:

[0081] W=U T V

[0082] Finally, the source signal is estimated:

[0083]

[0084] in is an estimate of the source signal s(t);

[0085] Step 3: Use the above algorithm to perform blind source separation on the structural modal frequency signal to separate the individual structural damage signal. Since damage will cause the structural modal frequency to change suddenly at the time of damage, it will be reflected in the blind source separation result as a jump phenomenon of the damage signal, so it is determined whether the structure is damaged and when the damage occurred.

[0086] In this embodiment, structural damage identification tests were conducted on a single damaged cantilever beam, a single damaged cantilever beam considering noise, and a double damaged cantilever beam considering damage development based on the above method. Figure 3 As shown, Figure 4 and Figure 5 The sixth-order modal frequency curve and blind source separation result of a single damaged cantilever beam are shown in Figure 1. Figure 6 and Figure 7 The model diagram of two damaged cantilever beams is shown in Figure 8 As shown, Fig. 9 and Fig.10 They are the sixth-order modal frequency curves of the two damaged cantilever beams considering damage development and the blind source separation results.

[0087] Depend on Figure 3-10It can be obtained that: for single damage and single damage considering noise, the modal frequency curve cannot determine whether there is damage and when the damage occurs; after eliminating temperature interference through this method, it can be clearly seen from the jump of the damage signal that the structure has damage on the 300th day. For the two damages developed from a single damage, the third-order natural frequency curve can roughly determine that the structure has damage on the 300th day, but it cannot be determined whether the damage is initial damage or developed damage, and only the occurrence of one damage can be determined; after eliminating temperature interference through this method, it can be clearly seen from the jump of the damage signal that the structure has the first damage on the 200th day and develops into two damages on the 300th day, so as to determine the development of the damage. The results show that this method has good robustness, proving that this method can detect single damage and damage development under different noise conditions.

[0088] Compared with traditional methods that cannot identify structural damage due to temperature interference, the method of the present invention successfully eliminates the masking of damage conditions by temperature factors in the structural modal frequency curve. This method can directly use the structural dynamic characteristic data collected by sensors for damage detection, does not rely on numerical or physical models, and the algorithm is lightweight and fast, and has good noise robustness, and can achieve accurate and reliable damage identification effects in actual engineering problems. In addition, compared with existing research on blind source separation of modal frequencies using the AMUSE algorithm, this method uses a second-order blind identification (SOBI) algorithm with better stability and stronger functionality, which overcomes the shortcomings of the AMUSE algorithm, such as indirect and obvious result expression, inability to identify small changes in modal frequencies, and weak noise resistance, and is more suitable for actual engineering needs.

[0089] The above disclosure is only a few specific embodiments of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the patent principle of the present invention. These improvements and modifications should also be regarded as the protection scope of the patent of the present invention.

Claims

1. A structural damage identification method based on blind source separation to eliminate temperature interference, characterized in that: The following steps are involved: Obtain the displacement time-history response data at the key positions of the target structure under environmental excitation, obtain the natural frequencies of each order of the target structure based on the displacement time-history response data, and construct the structural modal frequency signal; The structural modal frequency signal is whitened to obtain the delay covariance matrix of the structural modal frequency signal after the whitening process; the delay covariance matrix after the whitening process is decomposed by eigenvalues, and the optimal eigenvector matrix is ​​obtained by using the optimal approximation algorithm, and a blind source separation matrix is ​​constructed based on the optimal eigenvector matrix and the whitening matrix during the whitening process; The estimated signal of the source signal in the structural modal frequency signal is extracted according to the blind source separation matrix, and the damage development of the target structure is judged according to the estimated signal.

2. The method for identifying structural damage based on blind source separation and eliminating temperature interference according to claim 1 is characterized in that: The method for acquiring the structural modal frequency signal comprises the following steps: Arrange acceleration sensors at key locations of the target structure to collect displacement time-history response data under environmental excitation; The random subspace method or peak picking method is used to obtain the natural frequencies of each order of the target structure and construct the structural modal frequency signal.

3. The method for identifying structural damage based on blind source separation and eliminating temperature interference according to claim 1 is characterized in that: The whitening process of the structural modal frequency signal comprises the following steps: Determine the structural modal frequency signal as: x(t)=As(t) Among them, A is the mixing matrix, s(t) is the source signal, and x(t) is the observation signal, that is, the structural modal frequency signal; Whiten x(t): Among them, V is the whitening matrix, removing The correlation of each signal component makes its covariance matrix the unit matrix.

4. The method for identifying structural damage based on blind source separation and eliminating temperature interference according to claim 3 is characterized in that: The construction of the blind source separation matrix comprises the following steps: The delay covariance matrix of the structural modal frequency signal is determined as: R s (τ)=s(t)s T (t-t) The delay covariance matrix of the structural modal frequency signal after whitening is determined as: Among them, τ takes a set of different values ​​τ1,τ2,τ3,…,τ p , used to represent different time delays; Perform eigenvalue decomposition on the delay covariance matrix after whitening and obtain: Among them, U is an orthogonal matrix, that is, the optimal eigenvector matrix; among them, R s (τ i ) is a diagonal matrix; According to the delay covariance matrix after whitening, the optimal approximation algorithm is used to find the optimal eigenvector matrix U and obtain the blind source separation matrix W: W=U T V。 5. The method for identifying structural damage based on blind source separation and eliminating temperature interference according to claim 4 is characterized in that: The estimated signal of the source signal in the structural modal frequency signal is extracted according to the blind source separation matrix, as shown in the following formula: in, is an estimate of the source signal s(t).

6. The method for identifying structural damage based on blind source separation and eliminating temperature interference according to claim 1 is characterized in that: The method of judging the damage development of the target structure according to the estimated signal comprises the following steps: Whether the target structure is damaged and the damage development situation are determined based on whether the estimated signal has a jump. The moment when the jump occurs is the moment when the target structure is damaged.

7. A structural damage identification system based on blind source separation to eliminate temperature interference, characterized in that: The system comprises: processor; a memory having stored thereon a computer program executable on the processor; Wherein, when the computer program is executed by the processor, the steps of the method for identifying structural damage based on blind source separation and eliminating temperature interference as described in any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a data processing program, and when the data processing program is executed by the processor, the steps of the method for identifying structural damage based on blind source separation and eliminating temperature interference as described in any one of claims 1 to 6 are implemented.

Images