Carbon fiber reinforced composite damage detection method based on mamba model

Abstract

The application discloses a carbon fiber reinforced composite material damage detection method based on a Mamba model and relates to the technical field of ultrasonic waveguide detection. The method comprises the following steps: S1: collecting a waveguide signal set on a carbon fiber reinforced composite material with damage; S2: extracting a time characteristic sequence of the waveguide signal set through a waveguide signal processing branch; S3: extracting a spectrum characteristic of the waveguide signal set through a continuous wavelet transform branch; S4: fusing the time characteristic and the spectrum characteristic of the waveguide signal set, comprehensively acquiring multi-dimensional damage characteristics in the waveguide signal, and detecting and positioning defects of the overall carbon fiber reinforced composite material. Through the fusion calculation of the time characteristics of the Mamba, the time domain spectrum characteristics extracted by the continuous wavelet transform and the convolutional neural network, the baseline-free waveguide damage detection and positioning of the carbon fiber reinforced composite material are realized under the condition that the baseline waveguide signal is not used.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ultrasonic waveguide detection, in particular to a carbon fiber reinforced composite material damage detection method based on a Mamba model. BACKGROUND

[0002] Compared with metal materials, carbon fiber reinforced composites have the advantages of light weight, high specific strength, high specific stiffness, corrosion resistance, etc., and are widely used in aerospace and various industrial fields. Due to time-varying load, low-speed impact and harsh operating environment, carbon fiber reinforced composites are prone to invisible damage such as delamination and debonding, which seriously threatens the safety of equipment and personnel. Therefore, it is of great significance to study the non-destructive detection and evaluation technology of carbon fiber reinforced polymer composites for the economic and reliable operation of aerospace and engineering equipment.

[0003] Based on the advantages of long propagation distance, small attenuation, low cost, high sensitivity to various damages, etc., the structure health monitoring technology based on guided waves is widely considered as an ideal choice for detecting damage of plate-shaped composite structures. Guided wave detection is usually realized through a sensor network composed of piezoelectric transducers, in which the transmitter excites guided waves in the structure to be detected, and the receiver collects the corresponding signals for subsequent analysis and processing. The characteristics of guided waves will change due to the presence of damage, which may exhibit linear features such as amplitude attenuation and velocity change. Different damage indices correspond to different characteristic information, and special detection algorithms are designed to achieve accurate damage localization and evaluation.

[0004] Traditional damage identification methods lack effective signal feature extraction techniques, and usually require comparison or subtraction of test signals with baseline signals (signals from healthy structures) to extract damage scattering signals and obtain damage time-of-flight and gravitational wave energy attenuation information. This damage index extraction method usually only utilizes a single change in guided wave characteristics, and cannot accurately construct a high-order nonlinear mapping relationship between guided wave signals and damage location. SUMMARY

[0005] Based on the above-mentioned shortcomings of the prior art, the present application aims to provide a carbon fiber reinforced composite material damage detection method based on a Mamba model to solve the above technical problems.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solution: a carbon fiber reinforced composite material damage detection method based on a Mamba model, comprising:

[0007] S1: collecting a set of guided wave signals on a carbon fiber reinforced composite material with damage;

[0008] S2: extracting a time feature sequence of the set of guided wave signals through a guided wave signal processing branch;

[0009] S3: extracting the spectral features of the guided wave signal set through a continuous wavelet transform branch;

[0010] S4: fusing the time features and the spectral features of the guided wave signal set, comprehensively acquiring the multi-dimensional damage features in the guided wave signal, and detecting and positioning defects of the whole carbon fiber reinforced composite material.

[0011] The application is further provided as, in step S1, four piezoelectric transducers are arranged in a double-transmitting and double-receiving form for collecting a limited guided wave signal set.

[0012] The application is further provided as, in step S2, the time feature sequence of the guided wave signal set is extracted through a guided wave signal processing branch, and a state space model Mamba is used to process long sequence data:

[0013] From time t=0, the time points t=0, 1, 2 output from the state space model are derived;

[0014]

[0015] h t represents the state quantity at time t, x t represents the input control quantity at time t, y t represents the system output at time t, A represents the state transition matrix, B represents the influence of the input control quantity x t on the state quantity h t , C represents the influence of the state quantity h t on the system output y t , B and C specifically represent the weight matrix of the input control quantity x t and the state quantity h t , and map the input control quantity to the system output quantity, represents the average value of the weight matrix, and the average value of a set of weight matrices at different time points is used to smooth the process, so as to reduce the influence of noise;

[0016] The state equation at t=k is extracted:

[0017]

[0018] The convolution x is introduced to perform convolution operation:

[0019]

[0020] wherein, is the transfer function matrix between the input control quantity x t and the system output y t , which is composed of the transition matrix and the weight matrix, and finally outputs the operation result y;

[0021] The guided wave signal set is used as input data f (1) ∈R B×D D represents the sequence length of the guided wave signal set. Average pooling is performed on the data to obtain the dimensionality-reduced feature f. (2) ∈R B×D′ For sequence f (2) Divide into segments f (3) ∈R B×D″ Feature extraction is performed using the input state-space model Mamba: f (4) =M(f (3) ), where M(·) represents the sequenced features f obtained after segmenting the sequence data into the state-space model Mamba for feature extraction. (4) ∈R B×D″ ;

[0022] Stacking and linear processing of serialized features:

[0023] f1 (4) =Reshape(f (4) );f GW =Linear(f1) (4) ),in, For the stacked serialized features, Reshape(·) represents the stacking operation on the features, ensuring the correct alignment of feature dimensions with thread processing. GW Linear(·) represents the linear processing of the serialized features after linear processing, and outputs a feature vector of fixed size through a linear layer.

[0024] The present invention is further configured such that, in step S3, a continuous wavelet transform branch is used to perform time-frequency transformation on the guided wave signal set, and average pooling is performed on the signal to obtain the dimensionality-reduced feature f. (2) ∈R B×D′ Perform continuous wavelet transform on the features and output feature f. CWT (1) ∈R B×DCWT :

[0025]

[0026] f CWT (1) =CWT(a,b)[f (2) ];

[0027] Where CWT(·) represents the continuous wavelet transform of the signal, and t represents the signal L. s In (t), the time variable is ψ(t), which is the wavelet mother function. *(t) represents the complex conjugate of ψ(t), a is a scale parameter, controlling the wavelet width, b is a translation parameter, determining the position of the wavelet in time;

[0028] The output f of the wavelet transform is obtained CWT (1) The average pooling dimension reduction and neural network convolution are performed, and the spectral feature output of the final wavelet transform branch is obtained

[0029] The cross-entropy loss function more suitable for the classification task is used in the continuous wavelet transform branch:

[0030]

[0031] wherein, is a cross-entropy loss function, Y represents the true probability that the signal feature belongs to a certain class, f θ (·) represents the predicted score of the input sample of a certain class, and S(·) represents a Softmax activation function.

[0032] The application is further provided with the step S4, wherein the features of the two branches are fused through a concatenation operation, and a multi-layer linear layer is used for damage classification:

[0033] f Fusion =Concat(f GW ,f CWT );

[0034] f 1 Fusion =Linear1(f Fusion );

[0035] f 2 Fusion =Linear2(f 1 Fusion );

[0036] f Result =Linear3(f 2 Fusion );

[0037] wherein, f Fusion is the feature after the fusion of the two branches through the concatenation operation, Concat(·) represents a connection operation, is the serialized feature after linear processing, is the spectral feature output of the wavelet transform branch, D1 represents the feature dimension size of the guided wave branch, and is set to 1000, D2 represents the spectral feature dimension size of the continuous wavelet transform branch, and is set to 384, Linear i (·) represents the linear layer number, f represents the serialized features after processing through the i-th linear layer. Result ∈R B×N This represents the output damage classification result, where N represents the number of damage categories.

[0038] This invention provides a damage detection method for carbon fiber reinforced composite materials based on the Mamba model. The method involves acquiring guided wave signal sets from damaged carbon fiber reinforced composite materials; extracting the temporal feature sequence of the guided wave signal sets through guided wave signal processing; extracting the spectral features of the guided wave signal sets through continuous wavelet transform; and fusing the temporal and spectral features of the guided wave signal sets to comprehensively obtain multidimensional damage features from the guided wave signals. This allows for defect detection and localization of the entire carbon fiber reinforced composite material, resulting in the following beneficial effects:

[0039] The proposed detection method integrates the state-space model Mamba, continuous wavelet transform, and convolutional neural network to achieve efficient parallel computation. The Mamba method extracts time features from long-distance guided wave sequences with lower computational complexity. Continuous wavelet transform enhances the extraction capability and robustness of signal features. Since the spectrum of continuous wavelet transform is not a time feature map, the convolutional neural network is used to extract time-frequency features from the guided wave signals after continuous wavelet transform. This effectively combines the time features obtained from Mamba with the convolutional neural network to comprehensively acquire multi-scale damage features, extract deeper relational information from the guided wave signals, adaptively analyze guided wave signal information, and accurately extract damage-related feature information, achieving high-precision detection of composite material damage. This method achieves damage detection under conditions of limited guided wave signal sets and no baseline signal, directly using a complete time-series guided wave signal set as input, expanding the engineering application of guided waves in baseline-free environments, and improving the application scope of ultrasonic guided wave non-destructive testing.

[0040] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0042] Figure 1 A flowchart illustrating a damage detection method for carbon fiber reinforced composite materials based on the Mamba model, as an exemplary embodiment of the present invention;

[0043] Figure 2 A schematic diagram of a time feature extraction of a guided wave signal processing branch shown for an exemplary embodiment of the present application;

[0044] Figure 3 A schematic diagram of a model of a continuous wavelet transform branch shown for an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0045] Other advantages and effects of the present application can be easily understood by those skilled in the art from the description of the preferred embodiments of the present application. The present application can also be implemented or applied in other different specific embodiments, and the details in the description can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for illustrating the present application, but not for limiting the protection scope of the present application.

[0046] It should be noted that the diagrams provided in the following embodiments only schematically illustrate the basic concept of the present application, and only the components related to the present application are shown in the diagrams, but not the number, shape and size of the components when actually implemented. The shape, number and ratio of the components when actually implemented can be arbitrarily changed, and the layout pattern of the components can also be more complex.

[0047] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application, however, it is obvious for those skilled in the art that the embodiments of the present application can be implemented without these specific details, and in other embodiments, the known structures and devices are shown in the form of block diagrams instead of details, to avoid making the embodiments of the present application difficult to understand.

[0048] As shown in the figure, the carbon fiber reinforced composite material damage detection method based on the Mamba model includes: Figure 1

[0049] S1: collecting a set of guided wave signals on the carbon fiber reinforced composite material with damage;

[0050] S2: extracting a time feature sequence of the set of guided wave signals through a guided wave signal processing branch;

[0051] S3: extracting a spectral feature of the set of guided wave signals through a continuous wavelet transform branch;

[0052] S4: fusing the time feature and the spectral feature of the set of guided wave signals, comprehensively acquiring multi-dimensional damage features in the guided wave signals, and detecting and positioning defects of the overall carbon fiber reinforced composite material.

[0053] ​Specifically, in step S1, four piezoelectric transducers are first arranged on a composite material plate in a dual-transmitter, dual-receiver configuration. An excitation signal is generated by an arbitrary waveform generator, passed through an ultrasonic signal preamplifier, and applied to the excitation piezoelectric transducers. An oscilloscope collects and receives the ultrasonic signals from the piezoelectric transducers. The ultrasonic signals form a guided wave signal set.

[0054] like Figure 2 As shown, specifically, in step S2, the temporal feature sequence of the guided wave signal set is extracted. Starting from t=0, the output of the state-space model at different time points is derived to obtain the state equation at each time point. By introducing convolution parameters, convolution is incorporated into the calculation and the convolution operation is parallelized. Average pooling is performed on the input guided wave signal set to reduce the data dimensionality and obtain dimensionality-reduced features. The dimensionality-reduced data is segmented and input into the Mamba model for feature extraction to obtain serialized features. Finally, the serialized features are stacked and linearized to obtain the guided wave signal set features.

[0055] The specific steps of its guided wave signal set feature extraction stage are as follows:

[0056] Starting from time t=0, derive the time points t=0, 1, 2 output from the state-space model.

[0057]

[0058] h t Let x represent the state quantity at time t. t y represents the input control quantity at time t. t Let A represent the system output at time t, A represent the state transition matrix, and B represent the input control quantity x. t For state variable h t The influence, C represents the state variable h t For the system output y t The effects of B and C are specifically manifested in the input control quantity x. t With state quantity h t The weight matrix maps the input control quantity to the system output quantity. This represents the average value of the weight matrix. The process is smoothed by averaging a set of weight matrices at different time points to reduce the impact of noise.

[0059] Extract the state equation at t=k:

[0060]

[0061] Extract the state equation at t=k:

[0062]

[0063] Introducing convolution x for convolution operation:

[0064]

[0065] wherein, is an input control variable x t and a transfer function matrix between a system output y t , composed of a transition matrix and a weight matrix, and finally outputs an operation result y;

[0066] The guided wave signal set is taken as input data f (1) ∈R B×D , and D represents the sequence length of the guided wave signal set,

[0067] The data is averaged and pooled to obtain the reduced dimension feature f (2) ∈R B×D′ , the sequence f (2) is segmented into f (3) ∈R B ×D″ The input state space model Mamba is used for feature extraction: f (4) =M(f (3) ), wherein M(·) represents that after the segmented sequence data is input into the state space model Mamba for feature extraction based on the state space model Mamba, the serialized feature f (4) ∈R B×D″ is obtained;

[0068] The serialized feature is stacked and linearly processed:

[0069] f1 (4) =Reshape(f (4) ); f GW =Linear(f1 (4) ), wherein, is the stacked serialized feature, Reshape(·) represents a stacking operation on the feature, so that the feature dimension is correctly connected with the thread processing, f GW is the serialized feature after linear processing, Linear(·) represents linear processing of the serialized feature through a linear layer, and outputs a fixed-size feature vector.

[0070] As shown in FIG. Figure 3 , specifically, in step S3, the spectral feature of the guided wave signal set is extracted through the continuous wavelet transform branch, the guided wave signal set is averaged and pooled to obtain the reduced dimension feature of the guided wave signal set, the guided wave signal set after dimension reduction is subjected to continuous wavelet transform to obtain the spectral feature of the guided wave signal set, the spectral feature is dimensionally reduced to obtain the reduced spectral feature, and then convolution and signal segmentation are performed to extract the signal feature of the continuous wavelet transform spectral feature in the time domain.

[0071] The time domain spectrum feature extraction stage of the guided wave signal set is specifically:

[0072] The signal is averaged and pooled to obtain the dimension reduction feature f (2) ∈R B×D′ The continuous wavelet transform is performed on the feature, and the output feature is obtained

[0073]

[0074]

[0075] f CWT (1) =CWT(a,b)[f (2) ];

[0076] Wherein, CWT(·) represents the continuous wavelet transform of the signal, t represents the time variable in the signal L s (t), ψ(t) is the mother wavelet function, ψ * (t) represents the complex conjugate of ψ(t), a is the scale parameter, which controls the wavelet width, and b is the translation parameter, which determines the position of the wavelet in time;

[0077] The output f CWT (1) of the wavelet transform is obtained, the average pooling dimension reduction and neural network convolution are performed, the spectrum feature with a dimension of (16, 24) is output, the signal is segmented, the dimension is set to 384, and the final spectrum feature output of the wavelet transform branch is obtained

[0078] The cross-entropy loss function more suitable for the classification task is used in the continuous wavelet transform branch:

[0079]

[0080] Wherein, is the cross-entropy loss function, Y represents the true probability that the signal feature belongs to a certain class, f θ (·) represents the predicted score of the input sample of a certain class, and S(·) represents the Softmax activation function.

[0081] In step S4, the time features and spectrum features of the guided wave signal set are fused to comprehensively obtain the multi-dimensional damage features in the guided wave signal, and the overall carbon fiber reinforced composite material is detected and positioned.

[0082] Specifically, the features of the guided wave signal processing branch and the continuous wavelet transform branch are fused through cascading operation, three linear layer operations are performed on the fused guided wave signal set features, the carbon fiber reinforced composite material damage classification is obtained, and the damage detection and positioning are completed.

[0083] The time characteristics and spectrum characteristics of the guided wave signal set are fused, and the defect detection and positioning of the overall carbon fiber reinforced composite material are specifically:

[0084] f Fusion =Concat(f GW ,f CWT );

[0085] f 1 Fusion =Linear1(f Fusion );

[0086] f 2 Fusion =Linear2(f 1 Fusion );

[0087] f Result =Linear3(f 2 Fusion );

[0088] Where f Fusion is the feature after the cascade operation fuses two branches, Concat(·) represents the connection operation, is the serialized feature after linear processing, is the spectrum feature output of the wavelet transform branch, D1 represents the feature dimension size of the guided wave branch, which is set to 1000, D2 represents the frequency spectrum feature dimension size of the continuous wavelet transform branch, which is set to 384, Linear i (·) represents the linear layer number, is the serialized feature after the i-th linear layer processing, f Result ∈R B×N represents the output damage classification result, and N represents the number of damage categories.

[0089] In the detection of the traditional detection technology, the guided wave signal set of the healthy carbon fiber reinforced composite material is taken as the baseline signal, and a neural network model is trained through a large number of healthy guided wave signal sets and damaged guided wave signal sets for damage detection and positioning of the carbon fiber reinforced composite material, which leads to the inability to detect and position the carbon fiber reinforced composite material in the absence of baseline guided wave signals or limited guided wave data sets.

[0090] The time characteristics of the guided wave signal set are extracted using the Mamba model in the prior art, and seven different regions are randomly selected. A damage point is placed at the center of each region, and there are five damage positions in each region. Four groups of guided wave signal set data are generated for each damage scene, and a total of 140 guided wave signal data sets are generated.

[0091] Compared with the prior art, the use of a carbon fiber reinforced composite waveguide method based on the Mamba model for damage detection uses an 80% data set for training, a 10% data set for validation, and a 10% data set for testing, with a validation accuracy of 100% and a test accuracy of 100%, and a detection time of 84.92 seconds.

[0092] In this embodiment, the extracted guided wave signal set is a guided wave signal collected under the condition that the carbon fiber reinforced composite material has defects, and there is no guided wave signal in the healthy state, that is, the baseline-free guided wave damage detection and positioning of the carbon fiber reinforced composite material based on the Mamba model is performed without a baseline signal. Therefore, the time feature sequence of the guided wave signal set is extracted by using a specific formula. By introducing a convolution parameter, the convolution is introduced into the calculation and the convolution operation is parallelized. The input guided wave signal set is averaged and pooled to reduce the data dimension. The Mamba model is inputted for feature extraction, and the serialized features are stacked and linearized to obtain the guided wave signal set features. At the same time, the spectral features of the guided wave signal set are extracted in the time domain by continuous wavelet transform and convolutional neural network, and finally the time features and spectral features of the guided wave signal set are fused to comprehensively obtain the multi-dimensional damage features in the guided wave signal for defect detection and positioning of the overall carbon fiber reinforced composite material. Note that other formulas cannot obtain the depth features and comprehensive information of the damage information in the guided wave signal without a baseline.

[0093] The state space model Mamba is more efficient in processing long sequence guided wave signal sets and has lower computational complexity, which is very beneficial to actual damage detection engineering. The method extracts time features from long distance guided wave sequences by the Mamba method. Continuous wavelet transform enhances the extraction ability and robustness of signal features, and convolutional neural network is used to extract time-frequency features from the guided wave signal after continuous wavelet transform. The time features obtained by Mamba are effectively combined with convolutional neural network to comprehensively obtain multi-scale damage features, extract deeper relationship information in the guided wave signal, adaptively analyze guided wave signal information, accurately extract feature information related to damage, and realize high-precision detection of composite material damage. This method realizes damage detection under the condition of limited guided wave signal set and no baseline signal, directly uses the complete time sequence guided wave signal set as input, expands the engineering application of guided wave in the baseline-free environment, and improves the use range of ultrasonic guided wave nondestructive testing.

[0094] The above-described embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, the above-described embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center through a wired (for example, infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0095] It should be understood that the term "and / or" herein merely describes an association relationship of associated objects, which means that there can be three relationships, for example, A and / or B can represent three cases of A alone, A and B together, and B alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after it, but it can also represent an "and / or" relationship. The specific meaning can be understood according to the context before and after it.

[0096] In this application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b, or c can represent a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0097] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined according to their functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0098] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0099] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0100] In several embodiments provided in the present application, it should be understood that the disclosed system can be implemented in other ways. For example, the above-described device embodiments are only schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0101] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0102] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit.

[0103] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0104] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for damage detection of carbon fiber reinforced composites based on the Mamba model, characterized in that, Comprise: S1: Collecting guided wave signal set on damaged carbon fiber reinforced composite material; S2: Extracting time feature sequence of guided wave signal set through guided wave signal processing branch; In step S2, long sequence data is processed using state space model Mamba to extract time feature sequence of guided wave signal set: From time t=0, derive the time point t=0,1,2 output from the state space model; ； ； ； ； ； ； h t represents the state quantity at time t, x t represents the input control quantity at time t, y t represents the system output at time t, A represents the state transition matrix, B represents the influence of the input control quantity x t on the state quantity h t , C represents the influence of the state quantity h t on the system output y t , B and C specifically represent the weight matrix of the input control quantity x t and the state quantity h t , which maps the input control quantity to the system output quantity, represents the average value of the matrix, and the average value of the weight matrix at different time points is used to smooth the process, so as to reduce the influence of noise; Extract the state equation at t=k: ； Introduce convolution x for convolution operation: ； ； wherein, is an input control quantity x t and a transfer function matrix between a system output y t composed of a transition matrix and a weight matrix, and finally outputs an operation result y; Taking the guided wave signal set as input data , D represents the sequence length of the guided wave signal set, the data is average-pooled to obtain the reduced dimension feature , the sequence is segmented into , and the state space model Mamba is input to extract features: , wherein M(·) represents that the segmented sequence data is input into the state space model Mamba for feature extraction based on the state space model Mamba, and a serialized feature is obtained ; Stack and linearly process the serialized features: ; wherein, is the serialized feature after stacking, Reshape(·) represents a stacking operation on the feature, so that the feature dimension is correctly docked with the thread processing, is the serialized feature after linear processing, Linear(·) represents linear processing of the serialized feature through a linear layer, and outputs a fixed-size feature vector; S3: Extracting spectral features of guided wave signal set through continuous wavelet transform branch; S4: Fusing time features and spectral features of guided wave signal set to comprehensively obtain multi-dimensional damage features in guided wave signal, and detecting and positioning defects of overall carbon fiber reinforced composite material; Fuse features of two branches through cascading operation, and adopt multi-layer linear layer for damage classification: ； ； ； ； wherein, is the feature after fusing two branches of the cascade operation, represents a connection operation, is the serialized feature after linear processing, is the spectral feature output of the wavelet transform branch, D1 represents the feature dimension size of the guided wave branch, which is set to 1000, and D2 represents the frequency spectrum feature dimension size of the continuous wavelet transform branch, which is set to 384, represents a linear layer number, is the feature after the first linear layer, is the serialized feature after linear processing, represents the output damage classification result, and N represents the number of damage categories.

2. The method of claim 1, wherein the Mamba model-based carbon fiber reinforced composite material damage detection method is characterized by, In step S1, four piezoelectric transducers are arranged in a double-transmitting and double-receiving form to collect limited guided wave signal set.

3. The method of claim 1, wherein the Mamba model-based carbon fiber reinforced composite material damage detection method is characterized by, In step S3, using the continuous wavelet transform branch, the time-frequency transform is performed on the set of guided wave signals, the signals are averaged-pooled, and the reduced dimension features are obtained , the continuous wavelet transform is performed on the features, and the features are output : ； ； ； where CWT(·) denotes the continuous wavelet transform of a signal, t denotes the time variable in the signal is the mother wavelet, denotes the complex conjugate of a is the scale parameter, which controls the wavelet width, and b is the translation parameter, which determines the position of the wavelet in time;​ Obtaining the output of wavelet transform , performing average pooling dimension reduction and neural network convolution to obtain the spectral feature output of the final wavelet transform branch ; The cross-entropy loss function more suitable for classification task is used in the continuous wavelet transform branch: ； wherein, is a cross-entropy loss function, Y represents the true probability that a signal feature belongs to a certain class, represents the predicted score of an input sample for a certain class, represents a Softmax activation function.

Images