A metal corrosion acoustic emission monitoring method, device and storage medium

Abstract

The present invention relates to the field of metal monitoring, and in particular to a metal corrosion acoustic emission monitoring method, device and storage medium. The method comprises: obtaining the original acoustic emission signal of the metal corrosion process; performing wavelet denoising on the original acoustic emission signal to obtain a denoised signal; segmenting the denoised signal by setting a threshold parameter to obtain a plurality of segmented signals; extracting chaotic characteristic parameters from the plurality of segmented signals to obtain chaotic characteristic parameters; constructing a characteristic vector according to the chaotic characteristic parameters, and obtaining a classification model according to the characteristic vector; and predicting metal corrosion using a nonlinear dynamic material damage model. The present invention has the beneficial effect of being able to timely and accurately identify and predict early signs of metal corrosion damage.

Description

Technical Field

[0001] The invention relates to the field of metal monitoring, and in particular to a metal corrosion acoustic emission monitoring method, equipment and storage medium. Background Art

[0002] In nuclear power plants and other critical infrastructure, metal materials have long been subjected to the severe test of high temperature, high pressure, strong radiation and corrosive media. Among them, stress corrosion cracking (SCC), as a typical form of corrosion, poses a huge threat to equipment safety. The dynamic behavior of the SCC process is complex and changeable, including both sudden rapid crack expansion and continuous plastic deformation or other gradual damage processes. These different damage modes have different manifestations in acoustic emission (AE) signals. However, traditional AE signal processing methods often have difficulty in accurately distinguishing and quantifying these complex signals, especially when sudden modes and continuous modes coexist, resulting in insufficient timeliness and accuracy in the identification and prediction of early signs of damage. Summary of the invention

[0003] The purpose of the present invention is to provide a metal corrosion acoustic emission monitoring method, device and storage medium to solve the technical problem that the prior art is not timely and accurate enough in identifying and predicting early signs of metal corrosion damage.

[0004] The present invention provides a method for monitoring acoustic emission of metal corrosion, comprising the following steps:

[0005] S1. Obtain the original acoustic emission signal of the metal corrosion process;

[0006] S2, performing wavelet denoising on the original acoustic emission signal to obtain a denoised signal;

[0007] S3, segmenting the denoised signal by setting a threshold parameter to obtain multiple segmented signals;

[0008] S4, extracting chaotic characteristic parameters from the multiple segmented signals to obtain chaotic characteristic parameters;

[0009] S5, constructing a feature vector according to the chaotic feature parameters, and obtaining a classification model based on the feature vector;

[0010] S6. Construct a nonlinear dynamic material damage model based on the eigenvector to predict the trend of metal corrosion.

[0011] A storage medium stores instructions and data for implementing a metal corrosion acoustic emission monitoring method.

[0012] A metal corrosion acoustic emission monitoring device comprises: a processor and the storage medium; the processor loads and executes instructions and data in the storage medium to implement a metal corrosion acoustic emission monitoring method.

[0013] The beneficial effects provided by the present invention are:

[0014] Theoretical innovation and integration: This method innovatively combines chaos theory with metal corrosion monitoring technology, introduces chaos characteristic parameter analysis in the field of metal corrosion acoustic emission signal processing for the first time, and opens up a new way to study the corrosion process from a nonlinear dynamic perspective. Highly sensitive and accurate evaluation: By analyzing chaotic parameters such as Lyapunov exponents and correlation dimensions, this method can extremely sensitively capture subtle changes in the metal corrosion process, accurately evaluate the corrosion state even in the early stages of corrosion, and improve the accuracy of detection and early warning capabilities. Prediction and prevention strategies: Continuously monitoring chaotic characteristic parameters enables this method to have the ability to predict the development trend of metal corrosion, providing a scientific basis for formulating preventive maintenance strategies, helping to reduce sudden failures and reduce maintenance costs. Wide application potential: It shows good versatility and is suitable for a variety of metal materials and different corrosion environments. Regardless of the degree of corrosion, it can effectively detect and expand the scope of application of the technology. Efficient data processing capabilities: For massive acoustic emission signal data, this method provides efficient processing methods to meet the needs of real-time monitoring and rapid response, and enhances the practicality of online analysis. Promoting the understanding of corrosion mechanisms: Through in-depth analysis of chaos parameters, a new perspective is provided for the basic research of metal corrosion, which helps to understand the physical and chemical mechanisms of corrosion at a deeper level and promote the progress of corrosion science. By integrating chaos theory, an important step has been taken in improving the scientificity, effectiveness and foresight of metal corrosion monitoring, making innovative contributions to the optimization of industrial safety, materials science and maintenance strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic flow chart of the method of the present invention;

[0016] Figure 2 It is a working schematic diagram of the hardware device of the present invention. DETAILED DESCRIPTION

[0017] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0018] Please refer to Figure 1 , Figure 1 : is a flow chart of the method of the present invention; the present invention provides a metal corrosion acoustic emission monitoring method, the method specifically comprises the following steps:

[0019] S1. Obtain the original acoustic emission signal of the metal corrosion process;

[0020] S2, performing wavelet denoising on the original acoustic emission signal to obtain a denoised signal;

[0021] It should be noted that for the original acoustic emission (AE) signal , the wavelet threshold denoising method is used to process and obtain the denoised signal , the specific process can be expressed as:

[0022] (1)

[0023] in, and denote wavelet transform and inverse wavelet transform respectively; is the threshold factor; σ is the standard deviation of the noise; is a sign function.

[0024] It should be noted that the present invention involves a variety of wavelet operators when using the wavelet threshold denoising method for processing. Wavelet transform (W) and inverse wavelet transform play a key role in this process. The following are some common and widely used wavelet basis functions or operators:

[0025] 1. Haar wavelet: This is the simplest wavelet basis function with compact support and orthogonality. It is often used in teaching and simple signal processing tasks.

[0026] 2. Daubechies wavelet: This is an orthogonal wavelet family constructed by Ingrid Daubechies. They have good time-frequency localization properties and are available in a variety of orders to suit different signal processing needs. They are very popular in image and signal processing.

[0027] 3. Symlet wavelet: Symlet wavelet is an improvement on Daubechies wavelet, providing better symmetry while maintaining orthogonality and compact support.

[0028] 4. Coiflet wavelet: Coiflet wavelet is also a wavelet basis constructed by Daubechies. It has better symmetry than Daubechies wavelet and provides better performance in some applications.

[0029] As an example, if the 4th order Daubechies wavelet (DB4) is used for denoising, the filter coefficients are defined as follows:

[0030] Scaling function coefficients (low-pass filter coefficients) :

[0031] (2)

[0032] (3)

[0033] (4)

[0034] (5)

[0035] Wavelet function coefficients (high-pass filter coefficients) Available through Export:

[0036] (6)

[0037] In the iterative decomposition of discrete wavelet transform (DWT), the steps are as follows:

[0038] ①Decomposition:

[0039] Approximation signal (low frequency part): (7)

[0040] Detail signal (high frequency part): (8)

[0041] ②Threshold processing: Perform threshold processing,

[0042] (9)

[0043] ③Reconstruction:

[0044] pass and Refactoring ,

[0045] (10)

[0046] in is the number of decomposition levels.

[0047] This process utilizes the multi-scale characteristics of wavelets to achieve effective separation of signals and noise, providing clear data for subsequent signal processing.

[0048] S3, segmenting the denoised signal by setting a threshold parameter to obtain multiple segmented signals;

[0049] In the present invention, the signal after denoising is assumed to be , by setting the threshold , identify the start and end points of the AE event and segment the signal into multiple AE events.

[0050] S4, extracting chaotic characteristic parameters from the multiple segmented signals to obtain chaotic characteristic parameters;

[0051] It should be noted that chaos characteristic parameters are quantitative indicators used to describe the characteristics of chaotic systems. In chaos theory, chaos characteristic parameters help analyze the complexity and unpredictability of dynamic systems. These parameters usually include the maximum Lyapunov exponent, correlation dimension, and recursion graph analysis.

[0052] The phase space of chaos is a multidimensional space where each dimension represents a state variable of the system. In chaos theory, phase space is used to represent all possible states of a dynamic system.

[0053] However, for complex systems, it is often impossible to directly observe all state variables. Therefore, phase space reconstruction is needed to recover the dynamic characteristics of the system from partial observation data.

[0054] According to Takens' theorem, even if a multidimensional dynamic system can only be observed through a one-dimensional time series, the phase space can be reconstructed by selecting appropriate delay time and embedding dimension to reveal the true dynamics of the system. The reason for this is that the actual observed data is often a projection of the multidimensional dynamic system in one dimension, and phase space reconstruction can help better understand the true dynamic behavior of the system.

[0055] The Lyapunov index is an important indicator to measure the degree of chaos in a system. It indicates the average divergence or convergence rate of similar orbits in the phase space.

[0056] The maximum Lyapunov exponent being greater than zero is a basic characteristic of a chaotic system, which means that the system is extremely sensitive to small changes in initial conditions, the so-called "butterfly effect".

[0057] The maximum Lyapunov exponent The calculation process is as follows:

[0058] S311, using Takens theorem for segmentation signal, combined with preset delay time and embedding dimension , reconstruct the phase space; the specific formula is as follows:

[0059] (11).

[0060] S312. For each point in the phase space, find its nearest neighbor and record the distance between them.

[0061] Specifically, for every point in the phase space (Formula 1), find its nearest neighbor , record the distance between them , this distance can be Euclidean distance, Manhattan distance, etc.;

[0062] S313, after waiting for k time steps, measure the distance between the two points again ;

[0063] ①It should be noted that allowing the system to continue to evolve for a period of time (for example, time steps), and then measure the evolved distance between the two points again .

[0064] S314. For each pair of points, calculate the divergence rate: ,in is the sampling time interval;

[0065] S315. Calculate the average divergence rate of all points: ; It is The divergence rate of a point is calculated as follows: , It is The initial distance between a point and its nearest neighbor, It is through The distance between these two points after time steps;

[0066] S316, when ,Will As the maximum Lyapunov exponent It should be noted that when , will approach the maximum Lyapunov exponent of the system In actual calculations, we usually choose a sufficiently large value to approximate this limit.

[0067] It should be noted that the correlation dimension D 2 Calculated using the Grassberger-Procaccia method.

[0068] The correlation dimension is a parameter that describes the complexity of the chaotic attractor structure. It reflects the distribution and aggregation of points in the phase space. The correlation dimension calculated by the Grassberger-Procaccia method can help us understand the dynamic characteristics and complexity of the system.

[0069] (12)

[0070] in, C ( r ) is the correlation integral.

[0071] It should be noted that the recursion graph parameters are obtained by recursion graph analysis; the recursion graph parameters include: recursion rate RR, determination rate DET, average recursion length L and information entropy ENTR.

[0072] Recurrence graph is a visualization tool for analyzing repetitive patterns and structures in time series. Through recurrence graph analysis, a series of parameters can be obtained, such as recurrence rate (RR), determination rate (DET), average recurrence length (L), and information entropy (ENTR).

[0073] These parameters provide important information about the dynamic behavior, stability and complexity of the system.

[0074] Recurrence rate (RR): It represents the proportion of recurrence points in the recurrence graph, reflecting the frequency of repeated patterns in the time series.

[0075] Determinism rate (DET): measures the proportion of recursion points that form a diagonal structure in the recursion graph, indicating the stability and predictability of the system state.

[0076] Average recursion length (L): The average length of the diagonal structure, which reflects the continuity and stability of the system state.

[0077] Information entropy (ENTR): measures the complexity of the distribution of the length of the diagonal in the recursive graph, reflecting the complexity and uncertainty of the system dynamics.

[0078] Recursion Graph (RQA) parameters: including recursion rate (RR), determination rate (DET), average recursion length (L) and information entropy (ENTR), which are obtained through recursion graph analysis.

[0079] Recurrence Plot (RP) drawing

[0080] ①In the embedded phase space, for each pair of time points and ( ), calculate the distance between the corresponding phase space vectors. This distance can be Euclidean distance, Manhattan distance, etc.

[0081] ②Set the threshold and generate the recursive graph: Select a threshold , if the distance between two phase space vectors is less than this threshold, then the corresponding position on the recurrence graph ( , ) are marked black (or dark), indicating that the two states are "recursive", that is, they are close to each other in phase space. If the distance is greater than or equal to the threshold, the corresponding position on the recursion graph is marked white (or light).

[0082] ③ Calculation of Recursive Quantification Analysis (RQA) parameters

[0083] Recurrence Rate (RR): RR is the ratio of black points in the recurrence graph, indicating the frequency of state recurrence in the time series. The calculation formula is: RR = (number of recurrence points) / (number of total points).

[0084] ④ Determinism (DET): DET measures the proportion of recursive points that form a diagonal structure in the recursive graph. The diagonal structure indicates that the system maintains a similar state over a period of time, showing deterministic behavior. The calculation of DET involves the statistics of the diagonal length.

[0085] ⑤ Average Diagonal Line Length (L): L is the average length of all diagonal lines in the recursive graph (usually only diagonal lines with lengths greater than a certain threshold are considered). It reflects how long the system state remains stable.

[0086] ⑥ Entropy (ENTR): ENTR represents the complexity of the distribution of diagonal lengths in a recursive graph. It measures the uniformity or complexity of the distribution of diagonal lengths. If all diagonals are of similar length, the ENTR value is low; if the diagonal lengths vary greatly, the ENTR value is high.

[0087] S5, constructing a feature vector according to the chaotic feature parameters, and obtaining a classification model based on the feature vector;

[0088] It should be noted that step S5 is specifically as follows:

[0089] S51. Construct feature vector ;

[0090] In the present invention, for each AE event, a feature vector is constructed: ;

[0091] S52. Use machine learning algorithms to build classification models based on feature vectors: ;in, is the predicted injury pattern; It is a classification model; is a model parameter. The machine learning algorithm may be a random forest or the like.

[0092] S6. Construct a nonlinear dynamic material damage model based on the eigenvector to predict the trend of metal corrosion.

[0093] In describing the damage evolution of materials or structures, the following nonlinear dynamic models can be used:

[0094]

[0095] in:

[0096] represents the crack length, which is the main variable we are interested in and is used to describe the evolution of damage.

[0097] is time, which indicates the crack expansion process over time.

[0098] It is a time-dependent function representing the externally applied load or stress intensity and has a direct influence on the crack growth.

[0099] and is related to the crack length Related nonlinear functions.

[0100] These functions can be selected in various forms according to actual conditions, such as polynomial, exponential function, logarithmic function, etc., and may even contain more complex chaos-related terms. The selection of these terms aims to more realistically reflect the nonlinear growth of damage and possible irregular behavior.

[0101] The general form of a power function is: .

[0102] in and is a parameter that needs to be estimated through experimental data. is the chaotic eigenvector.

[0103] Please refer to Figure 2 , Figure 2 40 is a schematic diagram of the working of the hardware device of an embodiment of the present invention. The hardware device specifically includes: a metal corrosion acoustic emission monitoring device 401, a processor 402 and a storage medium 403.

[0104] A metal corrosion acoustic emission monitoring device 401: The metal corrosion acoustic emission monitoring device 401 implements the metal corrosion acoustic emission monitoring method.

[0105] Processor 402: The processor 402 loads and executes the instructions and data in the storage medium 403 to implement the metal corrosion acoustic emission monitoring method.

[0106] Storage medium 403: The storage medium 403 stores instructions and data; the storage medium 403 is used to implement the metal corrosion acoustic emission monitoring method.

[0107] In summary, the beneficial effects of the present invention are:

[0108] 1. Theoretical innovation and integration: This method innovatively combines chaos theory with metal corrosion monitoring technology, and introduces chaos characteristic parameter analysis in the field of metal corrosion acoustic emission signal processing for the first time, opening up a new way to study the corrosion process from a nonlinear dynamic perspective.

[0109] 2. Highly sensitive and accurate assessment: By analyzing chaotic parameters such as the Lyapunov exponent and correlation dimension, this method can extremely sensitively capture subtle changes in the metal corrosion process, accurately assess the corrosion status even in the early stages of corrosion, and improve detection accuracy and early warning capabilities.

[0110] 3. Prediction and prevention strategies: Continuous monitoring of chaotic characteristic parameters enables this method to have the ability to predict the development trend of metal corrosion, providing a scientific basis for formulating preventive maintenance strategies, helping to reduce sudden failures and reduce maintenance costs.

[0111] 4. Wide application potential: It shows good versatility and is suitable for a variety of metal materials and different corrosion environments. Regardless of the degree of corrosion, it can effectively perform detection, expanding the scope of application of the technology.

[0112] 5. Efficient data processing capability: This method provides efficient processing means for massive acoustic emission signal data, meeting the needs of real-time monitoring and rapid response, and enhancing the practicality of online analysis.

[0113] 6. Promote the understanding of corrosion mechanism: Through the in-depth analysis of chaos parameters, a new perspective is provided for the basic research of metal corrosion, which helps to understand the physical and chemical mechanism of corrosion at a deeper level and promote the progress of corrosion science.

[0114] By integrating chaos theory, the present invention has taken an important step in improving the scientificity, effectiveness and foresight of metal corrosion monitoring, and has made innovative contributions to the optimization of industrial safety, materials science and maintenance strategies.

[0115] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for monitoring acoustic emission of metal corrosion, characterized in that: The following steps are involved: S1. Obtain the original acoustic emission signal of the metal corrosion process; S2, performing wavelet denoising on the original acoustic emission signal to obtain a denoised signal; S3, segmenting the denoised signal by setting a threshold parameter to obtain multiple segmented signals; S4, extracting chaotic characteristic parameters from the multiple segmented signals to obtain chaotic characteristic parameters; S5, constructing a feature vector according to the chaotic feature parameters, and obtaining a classification model based on the feature vector; S6. Construct a nonlinear dynamic material damage model based on the eigenvector to predict the trend of metal corrosion; In step S4, the chaotic characteristic parameters include: maximum Lyapunov exponent , correlation dimension D 2 and recursive graph parameters; the recursive graph parameters include: recursive rate RR, determination rate DET, average recursive length L and information entropy ENTR; The maximum Lyapunov exponent The calculation process is as follows: S311, using Takens theorem for segmentation signal, combined with preset delay time and embedding dimension , reconstruct the phase space; S312. For each point in the phase space, find its nearest neighbor and record the distance between them. ; S313, after waiting for k time steps, measure the distance between the two points again ; S314. For each pair of points, calculate the divergence rate: ,in is the sampling time interval; S315. Calculate the average divergence rate of all points: ; It is The divergence rate of a point is calculated as follows: , It is The initial distance between a point and its nearest neighbor, It is through The distance between these two points after time steps; S316, when ,Will As the maximum Lyapunov exponent ; The nonlinear dynamic material damage model is as follows: , in: represents the crack length, which is the main variable we are interested in and is used to describe the evolution of damage; is time, which indicates the crack expansion process over time; It is a time-dependent function representing the externally applied load or stress intensity, which has a direct effect on the crack growth; and is related to the crack length Related nonlinear functions; in , and is a parameter estimated from experimental data, is the eigenvector.

2. A metal corrosion acoustic emission monitoring method according to claim 1, characterized in that: The operators used in the wavelet denoising include: Haar wavelet operator, Daubechies wavelet operator, Symlet wavelet operator and Coiflet wavelet operator.

3. A metal corrosion acoustic emission monitoring method according to claim 1, characterized in that: Correlation Dimension D 2 Calculated using the Grassberger-Procaccia method.

4. A metal corrosion acoustic emission monitoring method according to claim 1, characterized in that: Recursion graph parameters are obtained using recursion graph analysis.

5. A metal corrosion acoustic emission monitoring method according to claim 1, characterized in that: Step S5 is specifically as follows: S51. Construct feature vector ; S52. Use machine learning algorithms to build classification models based on feature vectors: ;in, is the predicted injury pattern; It is a classification model; are model parameters.

6. A storage medium, characterized in that: The storage medium stores instructions and data for implementing a metal corrosion acoustic emission monitoring method as described in any one of claims 1 to 5.

7. A metal corrosion acoustic emission monitoring device, characterized in that: include: Processor and storage medium; the processor loads and executes instructions and data in the storage medium to implement a metal corrosion acoustic emission monitoring method as described in any one of claims 1 to 5.

Images