Method and system for detecting internal and external surface defects of TA22 titanium alloy tubes

Abstract

A method and system for detecting internal and external surface defects of TA22 titanium alloy tubes provided by the present application relate to the technical field of defect detection; it includes: extracting defect features from the enhanced signal data by using a convolutional neural network according to the obtained defect features; introducing the characteristic factors of TA22 titanium alloy materials to recheck and determine the defect features to obtain the defect analysis results after recheck and determination; predicting the expansion of defects based on the defect analysis results after recheck and determination. The present application introduces the characteristic factors of TA22 titanium alloy materials, combines the step-by-step recheck and determination process of the internal and external surfaces, and optimizes the accuracy of defect feature analysis; comprehensively combines the crack expansion factor and the stress intensity factor with the expansion saturation factor to realize the prediction of defect expansion; greatly improves the detection efficiency through customized design, and realizes the rapid closed-loop from defect detection to report generation.

Description

Technical Field

[0001] The present application relates to the technical field of defect detection, and specifically relates to a method and system for detecting internal and external surface defects of TA22 titanium alloy tubes. Background Art

[0002] TA22 titanium alloy has been widely used in high-performance industrial fields such as aerospace and chemical equipment due to its excellent strength-to-weight ratio, high corrosion resistance, and good processing performance;

[0003] However, due to the high activity and complex structural characteristics of titanium alloy materials, they are extremely susceptible to processing defects during production and use. Conventional processing defects include cracks, pores, and inclusions; processing defects may cause a significant decline in the mechanical properties of the material, thereby threatening the safety of equipment or structures.

[0004] Currently, the detection method using ultrasonic detection has certain defects, specifically as follows:

[0005] The signal processing ability is limited. Conventional methods are vulnerable to noise interference for defect signals on complex surfaces and inside materials, and it is difficult to achieve high-precision defect identification;

[0006] The defect feature extraction is incomplete, lacking in-depth analysis of multi-modal features of defect signals, and it is difficult to effectively distinguish defect types and determine specific parameters of defects;

[0007] The review mechanism is missing. The detection results mainly rely on the output of a single method, lacking a review mechanism for judgment results, and lacking a review standard for specific materials. A single method cannot adapt to the review and defect inspection of specific materials, and there is a risk of false detection or missed detection. Summary of the Invention

[0008] The present application provides a method and system for detecting internal and external surface defects of TA22 titanium alloy tubes, aiming to solve the technical problem of limited signal processing ability in related technologies, avoid the problem of incomplete defect feature extraction, and solve the subsequent problems caused by the lack of a review mechanism.

[0009] To achieve the above object, in the present application, a method for detecting internal and external surface defects of TA22 titanium alloy tubes is provided, including the following:

[0010] Arrange ultrasonic sensors, and obtain signal data on the internal and external surfaces of TA22 titanium alloy tubes according to the ultrasonic sensors.

[0011] According to the signal data, perform denoising processing on it using an adaptive filtering algorithm.

[0012] According to the denoised signal data, perform enhancement processing on it using a time-frequency analysis algorithm.

[0013] Based on the enhanced signal data, a convolutional neural network is used to extract defect features therefrom, and defect features are obtained according to the extraction.

[0014] The characteristic factors of TA22 titanium alloy material are introduced to recheck and determine the defect features, and the defect analysis result after recheck and determination is obtained.

[0015] Based on the defect analysis result after recheck and determination, the expansion of the defect is predicted.

[0016] A detection report is generated according to the prediction result.

[0017] Preferably, ultrasonic sensors are arranged, and signal data on the inner and outer surfaces of TA22 titanium alloy pipes are obtained according to the ultrasonic sensors. The specific steps are as follows:

[0018] Ultrasonic sensors are evenly arranged on the inner and outer surfaces of TA22 titanium alloy pipes to form a full coverage of detection signal data.

[0019] The transmitting end of the ultrasonic sensor adjusts the transmitting angle so that the ultrasonic signal forms a uniformly distributed acoustic field on the pipe surface.

[0020] The receiving end sensors are arranged in a ring distribution mode and are evenly arranged around the outer surface of the pipe to ensure the spatial coverage rate of signal acquisition.

[0021] The receiving end sensors adjust the receiving angle and sensitivity in real time to adapt to the local changes on the pipe surface. The adjustment formula is:

[0022]

[0023] In the formula, is the adjusted receiving angle of the i-th receiving end sensor, is the initial receiving angle, ε is the sensitivity parameter, which reflects the response ability of the receiving angle to surface changes, and ΔR is the local geometric change amount of the pipe surface; the obtained signal data includes the morphological information on the pipe surface and the echo signal of internal defects, and the signal data provides a data basis for subsequent steps.

[0024] Preferably, according to the signal data, an adaptive filtering algorithm is used to perform denoising processing on it. The denoising formula is:

[0025]

[0026] In the formula, the adaptive filtering algorithm uses the least mean square error algorithm to dynamically update the filtering coefficient and optimize the signal-to-noise ratio of the signal data; represents the denoised signal data, w i (n) is the adaptive weight of the filter, x(n - i) is the input signal data, and M is the order of the filter.

[0027] Preferably, according to the denoised signal data, a time-frequency analysis algorithm is used to enhance it, and its enhancement formula is:

[0028]

[0029] In the formula, the short-time Fourier transform is used to perform time-frequency decomposition on the denoised signal data, enhance the signal strength of the low-frequency part, and suppress high-frequency noise; Y(t, f) represents the time-frequency representation after the short-time Fourier transform, represents the denoised signal data; w(t - n) represents the window function; f represents the frequency, and t represents the center position of the discrete time window.

[0030] Preferably, according to the enhanced signal data, a convolutional neural network is used to extract defect features therefrom. According to the extracted defect features, the specific steps are as follows:

[0031] Configure the convolutional neural network structure, divide the data set, train and optimize the configured convolutional neural network to obtain the optimized convolutional neural network. Input the enhanced signal data into the optimized convolutional neural network. The convolutional neural network extracts features from the enhanced signal data, and based on the extracted defect features, performs visual mapping on the defect features.

[0032] Preferably, an image recognition algorithm is used to review and determine the defect analysis result to obtain the defect analysis result after review and determination. The specific steps are as follows:

[0033] Introduce the characteristic factor of TA22 titanium alloy material to review and determine the defect features to obtain the defect analysis result after review and determination. The specific steps are as follows:

[0034] According to the defect features, perform a review and determination, introduce a roughness correction factor, and review the defect features on the inner surface of TA22 titanium alloy pipes. Specifically:

[0035]

[0036] In the formula, C in represents the comprehensive confidence level of the inner surface defect review, N is the number of inner surface defect features, is the matching value of the i-th inner surface defect feature, is the weight of the i-th inner surface defect feature, which is assigned according to the importance of the feature, is the TA22 titanium alloy inner surface roughness correction factor, and its calculation formula is: In the formula, is the TA22 titanium alloy inner surface roughness correction factor, represents the roughness at the i-th place on the inner surface; Rmax The maximum allowable roughness threshold for the inner surface.

[0037] For the review of inner surface defect features, calculate the comprehensive confidence level C according to the formula in , when C in ≥T in

[0038] If so, determine that the inner surface inspection result is consistent; otherwise, mark it as a divergence point. The defect features marked as divergence points will be fed back to the convolutional neural network and used to adjust the model parameters; the calculation formula for the inner surface determination threshold is:

[0039]

[0040] In the formula, T in is the threshold for inner surface determination, T base is the basic determination threshold, with a value of 0.85; ΔR is the deviation value of the current maximum roughness of the inner surface, obtained by calculating the difference between the actual roughness value of the inner surface and the preset target value of the inner surface roughness of TA22 titanium alloy tubes; R max is the maximum allowable roughness threshold for the inner surface.

[0041] Introduce the fatigue factor to review the outer surface defect features of TA22 titanium alloy tubes, specifically:

[0042]

[0043] In the formula, C out represents the comprehensive confidence level of outer surface defect review; M is the number of outer surface defect features; is the matching value of the i-th outer surface defect feature. It should be noted that takes a value of 0 or 1; is the weight of the i-th outer surface defect feature, assigned according to the importance of the defect feature; is the outer surface fatigue factor, and its calculation formula is: In the formula, is the outer surface fatigue factor, σ yiedld is the yield strength of TA22 titanium alloy, σ i is the stress distribution value at the i-th point; C out represents the comprehensive confidence level of outer surface defect review; M represents the number of outer surface defect features.

[0044] For the review of outer surface defect features, calculate the comprehensive confidence level C out ; when C out ≥T outWhen they are the same, the outer surface inspection result is determined to be consistent; otherwise, it is marked as a divergence point, and the defect features marked as divergence points will be fed back to the convolutional neural network and used to adjust the model parameters; the calculation formula for the outer surface determination threshold is:

[0045]

[0046] In the formula, T out is the threshold for outer surface determination, T base is the basic determination threshold, which has the same value as the inner surface basic determination threshold; σ design is the preset fatigue limit stress, σ current is the currently measured maximum stress value, σ yield is the yield strength of TA22 titanium alloy. Preferably,

[0047] Based on the defect analysis result after review determination, predict the expansion of the defect. The specific steps are as follows:

[0048] According to the defect analysis result, predict the expansion path of the defect. Specifically:

[0049] P(t) = P 0 +(v 0 +β·K·t)·t

[0050] In the formula, P(t) represents the spatial position of the defect at time t, P 0 is the initial position of the defect, v 0 is the initial expansion speed of the defect, β is the crack expansion factor of TA22 titanium alloy, which is calculated by substituting the measured values; K is the defect stress intensity factor, which is calculated from the fatigue performance of the material.

[0051] Predict the expansion size of the defect. Specifically:

[0052]

[0053] In the formula, S(t) is the size of the defect at time t, S 0 is the initial defect size, k is the material expansion rate constant, and λ is the expansion saturation factor, which is used to reflect the deceleration characteristics of the defect expansion.

[0054] Preferably, generate an inspection report according to the prediction result. The specific steps are as follows:

[0055] According to the defect prediction results, defect information is extracted. The defect information includes the specific location of the defect, the defect type, the defect size, the defect propagation path, and the defect evolution trend. The defect information is integrated and a detection report is generated. The detection report includes: a three-dimensional model of the defect distribution, marking the specific location, size, and type of the defect; a dynamic curve graph of the defect propagation trend. After the detection report is generated, it is fed back to the operator, and the operator performs maintenance on the pipe.

[0056] This application further provides an internal and external surface defect detection system for TA22 titanium alloy pipes, including the following units:

[0057] A data acquisition unit, which is used to deploy ultrasonic sensors and obtain signal data on the internal and external surfaces of TA22 titanium alloy pipes according to the ultrasonic sensors.

[0058] A denoising processing unit, which is used to perform denoising processing on the signal data by using an adaptive filtering algorithm according to the signal data.

[0059] An enhancement processing unit, which is used to perform enhancement processing on the denoised signal data by using a time-frequency analysis algorithm according to the denoised signal data.

[0060] A defect extraction unit, which is used to extract defect features from the enhanced signal data by using a convolutional neural network and obtain a defect analysis result according to the extraction.

[0061] A review and determination unit, which is used to review and determine the defect analysis result by using an image recognition algorithm and obtain the defect analysis result after review and determination.

[0062] A prediction unit, which predicts the evolution trend and expansion of the defect based on the defect analysis result after review and determination.

[0063] A report generation unit, which is used to generate a detection report according to the prediction result.

[0064] The beneficial effects of this application are as follows:

[0065] 1. By introducing the characteristic factors of TA22 titanium alloy materials and combining the step-by-step review and determination process of the internal and external surfaces, the accuracy of defect feature analysis is optimized; for the prediction of defect expansion, a dynamic expansion model that conforms to the characteristics of TA22 titanium alloy is established by comprehensively considering the crack expansion factor and the stress intensity factor and combining the expansion saturation factor. It can not only accurately describe the acceleration stage of defect expansion but also effectively predict the process of its stabilization; in addition, the present invention greatly improves the detection efficiency through targeted design, reduces the time cost of the review and prediction processes, and realizes a fast closed-loop from defect detection to report generation; it provides a solution for the detection of internal and external surface defects of TA22 titanium alloy pipes under complex working conditions.

[0066] 2. The present invention uses an adaptive filtering algorithm combined with a time-frequency analysis algorithm to perform multi-stage processing on ultrasonic signals, optimizes the signal-to-noise ratio by dynamically updating the filtering coefficients, and enhances the low-frequency characteristics of the signals using the short-time Fourier transform, effectively suppressing high-frequency noise interference and providing a data basis for subsequent defect feature extraction.

[0067] 3. Based on a convolutional neural network, feature extraction is performed on the enhanced signal data, and precise analysis of the defect position, type, and size is achieved through multiple convolutional kernels; compared with conventional single feature extraction methods, the present invention can simultaneously capture the time-domain characteristics, frequency-domain characteristics, and geometric characteristics of defect signals, significantly improving the comprehensiveness and accuracy of defect feature extraction. To make the above objects, features, and advantages of the present application more obvious and understandable, the following specific preferred embodiments are given, and in conjunction with the accompanying drawings, the detailed description is as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other relevant drawings can also be obtained based on these drawings.

[0069] Figure 1 It is a flowchart of a method for detecting internal and external surface defects of TA22 titanium alloy tubes provided by an embodiment of the present application.

[0070] Figure 2 It is an architecture diagram of a system for detecting internal and external surface defects of TA22 titanium alloy tubes provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0071] The following will describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application.

[0072] Please refer to Figure 1 , Figure 1 It is a flowchart of a method for detecting internal and external surface defects of TA22 titanium alloy tubes provided by an embodiment of the present application.

[0073] In this embodiment, a method for detecting internal and external surface defects of TA22 titanium alloy tubes includes steps S10, S20, S30, S40, S50, S60, and S70.

[0074] Step S10: Arrange ultrasonic sensors, and obtain signal data on the internal and external surfaces of TA22 titanium alloy tubes according to the ultrasonic sensors. The specific steps are as follows:

[0075] Ultrasonic sensors are evenly arranged on the inner and outer surfaces of TA22 titanium alloy tubes to achieve full coverage of the detection signal data.

[0076] It should be noted that to ensure comprehensive and accurate detection, the layout density and position of the ultrasonic sensors need to be selected according to the geometric dimensions of the tubes and the detection requirements; the layout scheme adopts an equally spaced arrangement to ensure the complete coverage of the signal data on the inner and outer surfaces of the tubes.

[0077] The transmitting end of the ultrasonic sensor adjusts the transmitting angle so that the ultrasonic signal data forms a uniformly distributed acoustic field on the surface of the tube.

[0078] It should be noted that the uniformity of the acoustic field is pre-designed through simulation software, and the transmitting angle is controlled by a stepper motor, with an adjustment range of 0° to 45° to cover the detection area of complex curved surfaces; the sensor uses a pulse method when transmitting signal data, and the duration of the signal data is controlled at the microsecond level to ensure the time resolution of the detection.

[0079] The receiving-end sensors are arranged in a circular distribution around the outer surface of the tube, evenly arranged to ensure the spatial coverage rate of signal data acquisition.

[0080] It should be noted that the specific spacing of the circular distribution of the receiving-end sensors is optimized according to the tube diameter and the ultrasonic propagation attenuation characteristics; for tubes with a diameter larger than the conventional model, the spacing of the receiving-end sensors is set to 5 mm to 20 mm to ensure the integrity of the echo signal data.

[0081] The receiving-end sensors adjust the receiving angle and sensitivity in real time to adapt to the local changes on the surface of the tube. The adjustment formula is:

[0082]

[0083] In the formula, is the adjusted receiving angle of the i-th receiving-end sensor, is the initial receiving angle, ε is the sensitivity parameter, reflecting the response ability of the receiving angle to surface changes, and ΔR is the local geometric change amount on the surface of the tube; the obtained signal data includes the morphological information on the surface of the tube and the echo signal of internal defects, and the signal data provides a data basis for the subsequent steps.

[0084] It should be noted that the adjustment of the receiving angle is achieved through the control of a stepper motor, and the adjustment of the sensitivity is completed through the gain control of the amplifier.

[0085] Step S20: According to the signal data, an adaptive filtering algorithm is used to denoise it. The denoising formula is:

[0086]

[0087] In the formula, the adaptive filtering algorithm uses the least mean square error algorithm to dynamically update the filtering coefficients and optimize the signal-to-noise ratio of the signal data; represents the signal data after denoising, and w i (n) is the adaptive weight of the filter, x(n - i) is the input signal data, and V is the order of the filter.

[0088] It should be noted that the adaptive filtering algorithm dynamically adjusts the filter weight w i (n) according to the change of the input signal data x(n) to separate the signal data from the noise; the output of the filter is the signal data after filtering, which has a higher signal-to-noise ratio; the order V of the filter determines the coverage range of the historical data in the signal data. The larger the order, the wider the response range of the filter, but the computational complexity also increases; the specific order is selected according to the characteristics of the actual detected signal data, and the value range is limited between 10 and 50; the adaptive filter dynamically updates the weight w i (n) by the least mean square error algorithm to minimize the error between the output signal data and the desired signal data, thereby optimizing the signal-to-noise ratio; the signal data after denoising not only retains the key features of the original signal data but also suppresses the interference of environmental noise, providing input data with stable quality for subsequent analysis steps.

[0089] Step S30: According to the signal data after denoising, use the time-frequency analysis algorithm to perform enhancement processing on it, and its enhancement formula is:

[0090]

[0091] In the formula, the short-time Fourier transform is used to perform time-frequency decomposition on the signal data after denoising, enhance the signal intensity of the low-frequency part, and suppress the high-frequency noise; U(t, f) represents the time-frequency representation after the short-time Fourier transform, represents the signal data after denoising; w(t - n) represents the window function; f represents the frequency, and t represents the center position of the discrete time window.

[0092] It should be noted that the short-time Fourier transform is a time-frequency analysis method. By segmenting the signal data for analysis, it can simultaneously obtain the time-domain and frequency-domain characteristics of the signal data. The introduction of the window function w(t-τ) ensures the stability of the signal data within the local time window. The enhanced function of the time-frequency analysis algorithm is mainly reflected in the enhancement of the low-frequency part of the signal data and the suppression of high-frequency noise. By dynamically adjusting the amplitude of Y(t, f), the algorithm can highlight the signal data with low-frequency characteristics and filter out high-frequency noise at the same time. The result of the short-time Fourier transform is represented in the form of a time-frequency diagram, where the horizontal axis is time, the vertical axis is frequency, and the color or brightness of the image represents the intensity of the signal data. This result provides a basis for the input data of the subsequent convolutional neural network. The application of the time-frequency analysis algorithm needs to combine the characteristics of the ultrasonic signal data, perform the short-time Fourier transform on each segment of the signal data separately, store the processed enhanced signal data in digital form, and set the sampling rate according to the characteristics of the signal data. The lower limit of the sampling rate is set to 1 GHz.

[0093] Step S40: According to the enhanced signal data, use a convolutional neural network to extract its defect features. The specific steps for extracting the obtained defect features are as follows:

[0094] Configure the convolutional neural network structure, segment the data set, train and optimize the configured convolutional neural network to obtain the optimized convolutional neural network. Input the enhanced signal data into the optimized convolutional neural network. The convolutional neural network extracts features from the enhanced signal data, and based on the extracted defect features, perform a visual mapping of the defect features.

[0095] It should be noted that the structure design of the convolutional neural network includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The configuration is optimized according to the characteristics of the enhanced signal data. The convolutional layer is used to extract the local features of the signal data, the pooling layer reduces redundant data through dimensionality reduction, the fully connected layer maps the features into high-dimensional vectors, and the output layer is used to classify the defect features.

[0096] Furthermore, it should be noted that to improve the training effect, the enhanced signal data is segmented into a training set, a validation set, and a test set, with a segmentation ratio of 8:1:1. The training set is used to train the convolutional neural network, the validation set is used to adjust the network parameters, and the test set is used to evaluate the network performance.

[0097] Furthermore, it should be noted that the training process uses the supervised learning method, uses the labeled defect feature data as labels, and updates the network weights through the backpropagation algorithm. The optimization methods include:

[0098] Learning rate adjustment: The initial learning rate is set to 0.001 and gradually decays as the number of training rounds increases; Regularization processing: L2 regularization is introduced to avoid overfitting; Data augmentation: The training samples are extended by translating and scaling the signal data to improve the generalization ability of the network.

[0099] Furthermore, it should be noted that the optimized convolutional neural network receives the time-frequency features of the enhanced signal data and uses them as input data; The convolutional layer extracts the local features of the input signal data through convolutional kernels, and the size, number, and stride of each layer of convolutional kernels are designed according to the characteristics of the signal data; After feature extraction, the generated feature map contains the spatial distribution, type, and signal intensity of the defects.

[0100] Furthermore, it should be noted that the visualization mapping reduces the high-dimensional features to three-dimensional space to generate a 3D point cloud map; Different types of defects are mapped to different colors or brightness levels to facilitate the operator to intuitively understand the type and distribution of the defects.

[0101] Step S50: Introduce the characteristic factor of TA22 titanium alloy material to recheck and determine the defect characteristics, and obtain the defect analysis result after recheck and determination. The specific steps are as follows:

[0102] Recheck and determine according to the defect characteristics, introduce the roughness correction factor, and recheck the inner surface defect characteristics of TA22 titanium alloy pipes, specifically:

[0103]

[0104] In the formula, C in represents the comprehensive confidence level of the inner surface defect recheck, N is the number of inner surface defect characteristics, is the matching value of the i-th inner surface defect characteristic, is the weight of the i-th inner surface defect characteristic, is the roughness correction factor for the inner surface of TA22 titanium alloy, and its calculation formula is: In the formula, is the roughness correction factor for the inner surface of TA22 titanium alloy, represents the roughness at the i-th place on the inner surface; R max is the maximum allowable roughness threshold for the inner surface.

[0105] It should be noted that The value range of

[0106] It should be noted that is calculated through the following formula:

[0107]

[0108] where α i is the influence factor of the i-th defect feature, determined according to the influence degree of the defect on the mechanical properties of the pipe; is the roughness correction factor, used to reflect the importance of the inner surface roughness to specific defect features; N is the total number of inner surface defect features, is the weight of the i-th inner surface defect feature.

[0109] Furthermore, it should be noted that R max is determined by measuring the influence of different roughnesses on the fatigue performance of TA22 titanium alloy, determining the critical roughness value according to the influence, introducing a roughness safety factor based on the critical roughness value, and substituting the critical roughness value and the roughness safety factor into the calculation to obtain the specific value of R max .

[0110] R max = R crit ·(1 - S f )

[0111] where R crit is the critical roughness value determined by experiments, and Sf is the safety factor, with a value range of 0.1 to 0.2.

[0112] For the review of inner surface defect features, calculate the comprehensive confidence level C in according to the formula. When C in ≥ T in

[0113] , it is determined that the inner surface detection result is consistent; otherwise, it is marked as a divergence point, and the defect features marked as divergence points will be fed back to the convolutional neural network and used to adjust the weights of the convolutional neural network; the formula for calculating the inner surface determination threshold is:

[0114]

[0115] where T in is the threshold for inner surface determination, T base is the basic determination threshold, with a value of 0.85; ΔR is the deviation value of the current maximum inner surface roughness; R max is the maximum roughness threshold allowed for the inner surface.

[0116] It should be noted that ΔR is calculated by comparing the actual inner surface roughness value with the preset target value, and the specific formula is:

[0117] ΔR = R actual - R target

[0118] where R actual is the actually measured roughness value of the inner surface, and R targetis the preset target value, and the preset target value determines the range based on the pipe design requirements, usage environment, and processing capabilities of TA22 titanium alloy, and is obtained by combining the measured value with the standard surface roughness range of the pipe. ΔR is the deviation value of the maximum roughness of the current inner surface.

[0119] Introduce the fatigue factor to recheck the external surface defect characteristics of TA22 titanium alloy pipes, specifically:

[0120]

[0121] In the formula, C out represents the comprehensive confidence level of the external surface defect recheck; M is the number of external surface defect characteristics; is the matching value of the i-th external surface defect characteristic, is the weight of the i-th external surface defect characteristic, which is assigned according to the importance of the defect characteristic; is the external surface fatigue factor, and its calculation formula is: In the formula, is the external surface fatigue factor, σ yield is the yield strength of TA22 titanium alloy, σ i is the stress distribution value at the i-th point, C out represents the comprehensive confidence level of the external surface defect recheck, and M represents the number of external surface defect characteristics.

[0122] It should be noted that takes a value of 0 or 1, where 0 indicates non-matching and 1 indicates matching.

[0123] Furthermore, it should be noted that the weight distribution of the external surface defect characteristics is calculated based on the influence degree of the defect characteristics, and the calculation formula is:

[0124]

[0125] In the formula, β i is the sensitivity factor of the i-th defect characteristic, which is determined according to the influence of the defect position on the external surface anti-fatigue performance, is the external surface fatigue factor, is the weight of the external surface defect characteristic.

[0126] It should be noted that takes a value range of 0 or 1, where 0 indicates matching and 1 indicates non-matching.

[0127] Furthermore, it should be noted that the yield strength is obtained through a tensile test, which is carried out according to the standard test procedure ASTM E8 / E8M. The specific steps are as follows:

[0128] A1. Prepare specimens of standard size.

[0129] A2. Apply a load to the specimen using a tensile testing machine and record the relationship between stress and strain.

[0130] A3. Take a 0.2% offset value in the stress-strain curve as the yield strength.

[0131] Furthermore, it should be noted that the stress distribution value is calculated by the finite element method based on stress field analysis, specifically:

[0132]

[0133] In the formula, F is the applied load, A i is the effective cross-sectional area at the location of the defect, and σ i is the stress distribution value.

[0134] For the review of the outer surface defect characteristics, calculate the comprehensive confidence level C according to the formula out ; when C out ≥T out , it is determined that the outer surface detection result is consistent; otherwise, it is marked as a divergence point, and the defect characteristics marked as divergence points will be fed back to the convolutional neural network and used to adjust the weights of the convolutional neural network; the formula for calculating the outer surface determination threshold is:

[0135]

[0136] In the formula, T out is the threshold for outer surface determination, T base is the basic determination threshold, with a value of 0.85; σ design is the preset fatigue limit stress, σ current is the currently measured maximum stress value, and σ yield is the yield strength of TA22 titanium alloy.

[0137] It should be noted that the fatigue limit stress is based on the actual fatigue test data of TA22 titanium alloy and is set in combination with the application environment:

[0138] B1. Determine the fatigue strength of the material at different numbers of cycles through fatigue tests.

[0139] B2. Combine the service environment and introduce the fatigue safety factor U f for calculation:

[0140] σ design = σ fatigue ·(1 - U f )

[0141] In the formula, σ fatigue is the fatigue strength determined by experiments, U f is the fatigue safety factor, and σ design is the fatigue limit stress.

[0142] Step S60: Based on the defect analysis results after review and determination, predict the expansion of the defect. The specific steps are as follows:

[0143] According to the defect analysis results, predict the expansion path of the defect, specifically:

[0144] P(t) = P 0 +(v 0 +β·K·t)·t

[0145] In the formula, P(t) represents the spatial position of the defect at time t, P 0 is the initial position of the defect, v 0 is the initial expansion speed of the defect, β is the crack growth factor of TA22 titanium alloy; K is the stress intensity factor of the defect.

[0146] It should be noted that the crack growth factor is obtained based on fatigue crack growth tests and by fitting the Paris law formula:

[0147] β = C·(ΔK) m

[0148] In the formula, C and m are experimental fitting constants, ΔK is the stress intensity factor range, and β is the crack growth factor of TA22 titanium alloy.

[0149] Furthermore, it should be noted that the stress intensity factor of the defect is calculated by the following formula:

[0150]

[0151] In the formula, σ is the stress acting on the tip of the defect, a is the half-length of the defect, and K represents the stress intensity factor of the defect.

[0152] Predict the expansion size of the defect, specifically:

[0153]

[0154] In the formula, S(t) is the size of the defect at time t, S 0 is the initial defect size, k is the material expansion rate constant, and λ is the expansion saturation factor.

[0155] It should be noted that the material expansion rate constant is obtained by obtaining crack growth rate data through fatigue crack growth experiments; and based on the Paris law formula determine k; introduce the environmental correction factor η in combination with the working environment of the material, and finally calculate k = k base ·η.

[0156] It should be further noted that the extended saturation factor records the stage when crack growth tends to saturate, and determines λ according to the formula and combines the maximum allowable crack length a max and material properties to calculate By introducing the environmental correction factor μ, finally calculate λ = λ base ·μ to obtain.

[0157] Step S70: Generate a detection report according to the prediction result. The specific steps are as follows:

[0158] According to the defect prediction result, extract the defect information, which includes the specific location, defect type, defect size, defect propagation path and defect evolution trend of the defect. Integrate the defect information and generate a detection report, which includes: a three-dimensional model of the defect distribution, marking the specific location, defect size and defect type of the defect; a dynamic curve graph of the defect propagation trend; after the detection report is generated, feedback the detection report to the operator, and the operator maintains the pipe.

[0159] It should be further noted that the generation of the detection report is based on a standardized data template, including the following parts:

[0160] A three-dimensional model of the defect distribution, generated by a three-dimensional modeling tool, marking the location and range of the defect in different colors; a dynamic curve graph of the defect propagation trend, visualizing the prediction result as an expansion trend curve changing with time, intuitively showing the future change of the defect; statistical information on the defect type and size, listing the specific parameters and detection results of various defects in tabular form.

[0161] It should be further noted that the detection report is transmitted to the operator in real time through a digital platform. The operator formulates a corresponding maintenance plan according to the defect distribution and trend analysis results in the report.

[0162] Thus, a method for detecting internal and external surface defects of TA22 titanium alloy pipes is completed.

[0163] Please also refer to Figure 2, this application further provides an internal and external surface defect detection system for TA22 titanium alloy tubes, including the following units: a data acquisition unit for arranging ultrasonic sensors and acquiring signal data of the internal and external surfaces of TA22 titanium alloy tubes according to the ultrasonic sensors; a denoising processing unit for denoising the signal data by using an adaptive filtering algorithm according to the signal data; an enhancement processing unit for enhancing the denoised signal data by using a time-frequency analysis algorithm; a defect extraction unit for extracting defect features from the enhanced signal data by using a convolutional neural network and obtaining the extracted defect features; a review and determination unit for introducing the characteristic factors of TA22 titanium alloy materials to review and determine the defect features and obtaining the defect analysis result after review and determination; a prediction unit for predicting the expansion of the defect based on the defect analysis result after review and determination; and a report generation unit for generating a detection report according to the prediction result.

[0164] Thus, an internal and external surface defect detection system for TA22 titanium alloy tubes is completed.

[0165] It should be understood that the "one embodiment" or "an embodiment" or "a feasible implementation manner" or "some implementation manners" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the embodiments of the present invention. Therefore, the "in one embodiment" or "in an embodiment" or "in a feasible implementation manner" or "in some implementation manners" that appear throughout the specification do not necessarily refer to the same embodiment. In addition, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the steps and modules involved are not necessarily essential to the embodiments of the present invention.

[0166] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented. These instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0167] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the functions in the process Figure 1 one process or multiple processes and / or blocksFigure 1 The functions specified in one or more boxes.

[0168] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 process or multiple processes and / or boxes Figure 1 or more boxes.

[0169] The above is only a specific implementation manner of the present invention, but the protection scope of the embodiments of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the embodiments of the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the embodiments of the present invention. Therefore, the protection scope of the embodiments of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for detecting inner and outer surface defects of TA22 titanium alloy pipes, characterized in that: These include: Arrange ultrasonic sensors to obtain signal data of the inner and outer surfaces of TA22 titanium alloy pipes according to the ultrasonic sensors; According to the signal data, an adaptive filtering algorithm is used to perform denoising; According to the denoised signal data, the time-frequency analysis algorithm is used to enhance it; According to the enhanced signal data, a convolutional neural network is used to extract defect features, and the defect features are obtained based on the extraction; The characteristic factors of TA22 titanium alloy material are introduced to review and determine the defect characteristics, and the defect analysis results after review and determination are obtained, including: According to the defect characteristics, the review and judgment are carried out, and the roughness correction factor is introduced to review the inner surface defect characteristics of TA22 titanium alloy pipes, specifically: In the formula, C in represents the comprehensive confidence of the internal surface defect review, N is the number of internal surface defect features, is the matching value of the i-th inner surface defect feature, is the weight of the i-th inner surface defect feature, is the correction factor of the inner surface roughness of TA22 titanium alloy; For the internal surface defect feature review, the comprehensive confidence C is calculated according to the formula in , when C in ≥T in When , the inner surface detection result is determined to be consistent; otherwise, it is marked as a divergence point. The defect features marked as divergence points will be fed back to the convolutional neural network and used to adjust the convolutional neural network model weights; the calculation formula for the inner surface judgment threshold is: Where, T in is the threshold for inner surface determination, T base is the basic judgment threshold, which is 0.85; ΔR is the deviation value of the maximum roughness of the current inner surface, R max is the maximum roughness threshold allowed for the inner surface; The fatigue factor was introduced to review the external surface defect characteristics of TA22 titanium alloy tubes, specifically: In the formula, C out Represents the comprehensive confidence of the external surface defect review; M is the number of external surface defect features; is the matching value of the i-th external surface defect feature, is the weight of the i-th external surface defect feature, is the outer surface fatigue factor; For the review of external surface defect characteristics, the comprehensive confidence C is calculated according to the formula out ; When C out ≥T out When , the outer surface detection result is determined to be consistent; otherwise, it is marked as a divergence point. The defect features marked as divergence points will be fed back to the convolutional neural network and used to adjust the model parameters; the calculation formula for the outer surface judgment threshold is: Where, T out is the threshold for external surface determination, T base is the basic judgment threshold, which is consistent with the value of the inner surface basic judgment threshold; σ design is the preset fatigue limit stress, σ current is the maximum stress value currently measured, σ yield is the yield strength of TA22 titanium alloy; Based on the defect analysis results after review and determination, the defect expansion is predicted, including: According to the defect analysis results, the defect expansion path is predicted, specifically: P(t)=P0+(v0+β·K·t)·t Where, P(t) represents the spatial position of the defect at time t, P0 is the initial position of the defect, v0 is the initial expansion speed of the defect, β is the crack growth factor of TA22 titanium alloy, and K is the defect stress intensity factor; The prediction of the defect expansion size is as follows: Where S(t) is the defect size at time t, S0 is the initial defect size, k is the material expansion rate constant, and λ is the expansion saturation factor; Generate a test report based on the prediction results.

2. A method for detecting inner and outer surface defects of a TA22 titanium alloy pipe as claimed in claim 1, characterized in that: Arrange ultrasonic sensors and obtain signal data of the inner and outer surfaces of TA22 titanium alloy pipes according to the ultrasonic sensors. The specific steps are as follows: Ultrasonic sensors are evenly arranged on the inner and outer surfaces of TA22 titanium alloy pipes to form full coverage of detection signal data; The transmitting end of the ultrasonic sensor adjusts the transmitting angle so that the ultrasonic signal forms a uniformly distributed sound wave field on the surface of the pipe; The sensors at the receiving end are arranged in a circular distribution manner, evenly arranged around the outer surface of the pipe to ensure the spatial coverage of signal collection; The receiving end sensor adjusts the receiving angle and sensitivity in real time to adapt to the local changes on the pipe surface. The adjustment formula is: In the formula, is the adjusted receiving angle of the i-th receiving end sensor, is the initial receiving angle, ε is the sensitivity parameter, which reflects the responsiveness of the receiving angle to surface changes, and ΔR is the local geometric change of the pipe surface. The acquired signal data contains the morphological information of the pipe surface and the echo signal of the internal defects. The signal data provides a data benchmark for subsequent steps.

3. The method for detecting inner and outer surface defects of a TA22 titanium alloy pipe according to claim 1, characterized in that: According to the signal data, an adaptive filtering algorithm is used to denoise it, and the denoising formula is: In the formula, the adaptive filtering algorithm uses the minimum mean square error algorithm to dynamically update the filter coefficients and optimize the signal-to-noise ratio of the signal data; represents the signal data after denoising, w i (n) is the adaptive weight of the filter, x(ni) is the input signal data, and M is the order of the filter.

4. A method for detecting inner and outer surface defects of a TA22 titanium alloy pipe as claimed in claim 3, characterized in that: According to the denoised signal data, the time-frequency analysis algorithm is used to enhance it, and the enhancement formula is: In the formula, short-time Fourier transform is used to perform time-frequency decomposition on the denoised signal data, enhance the signal strength of the low-frequency part, and suppress high-frequency noise; Y(t, f) represents the time-frequency representation after short-time Fourier transform, represents the signal data after denoising; w(tn) represents the window function; f represents the frequency, and t represents the center position of the discrete time window.

5. A method for detecting inner and outer surface defects of a TA22 titanium alloy tube as claimed in claim 4, characterized in that: According to the enhanced signal data, a convolutional neural network is used to extract defect features, and the defect features are obtained based on the extraction. The specific steps are as follows: Configure the convolutional neural network structure, segment the data set, train and optimize the configured convolutional neural network, obtain the optimized convolutional neural network, input the enhanced signal data into the optimized convolutional neural network, and use the convolutional neural network to extract features from the enhanced signal data. Based on the extraction, the defect features are obtained and visualized.

6. A method for detecting inner and outer surface defects of a TA22 titanium alloy pipe as claimed in claim 1, characterized in that: Generate a test report based on the prediction results. The specific steps are as follows: According to the defect prediction results, the defect information is extracted, including the specific location of the defect, the defect type, the defect size, the defect expansion path and the defect evolution trend. The defect information is integrated and a test report is generated. The test report includes: a three-dimensional model of the defect distribution, marking the specific location, defect size and defect type of the defect; a dynamic curve chart of the defect expansion trend; after the test report is generated, the test report is fed back to the operator, who then maintains the pipe.

7. A system for detecting inner and outer surface defects of a TA22 titanium alloy pipe, the system being applied to the method for detecting inner and outer surface defects of a TA22 titanium alloy pipe as claimed in any one of claims 1 to 6, characterized in that: The following units are included: A data acquisition unit, used to arrange ultrasonic sensors and acquire signal data of the inner and outer surfaces of the TA22 titanium alloy pipe according to the ultrasonic sensors; A denoising processing unit, used for denoising the signal data by using an adaptive filtering algorithm; An enhancement processing unit, used to enhance the denoised signal data by using a time-frequency analysis algorithm; A defect extraction unit is used to extract defect features from the enhanced signal data using a convolutional neural network, and obtain defect analysis results based on the extraction; The review and determination unit is used to review and determine the defect analysis result by using an image recognition algorithm to obtain the defect analysis result after review and determination, specifically including: According to the defect characteristics, the review and judgment are carried out, and the roughness correction factor is introduced to review the inner surface defect characteristics of TA22 titanium alloy pipes, specifically: In the formula, C in represents the comprehensive confidence of the internal surface defect review, N is the number of internal surface defect features, is the matching value of the i-th inner surface defect feature, is the weight of the i-th inner surface defect feature, is the correction factor of the inner surface roughness of TA22 titanium alloy; For the internal surface defect feature review, the comprehensive confidence C is calculated according to the formula in , when C in ≥T in When , the inner surface detection result is determined to be consistent; otherwise, it is marked as a divergence point. The defect features marked as divergence points will be fed back to the convolutional neural network and used to adjust the convolutional neural network model weights; the calculation formula for the inner surface judgment threshold is: Where, T in is the threshold for inner surface determination, T base is the basic judgment threshold, which is 0.85; ΔR is the deviation value of the maximum roughness of the current inner surface, R max is the maximum roughness threshold allowed for the inner surface; The fatigue factor was introduced to review the external surface defect characteristics of TA22 titanium alloy tubes, specifically: In the formula, C out Represents the comprehensive confidence of the external surface defect review; M is the number of external surface defect features; is the matching value of the i-th external surface defect feature, is the weight of the i-th external surface defect feature, is the outer surface fatigue factor; For the review of external surface defect characteristics, the comprehensive confidence C is calculated according to the formula out ; When C out ≥T out When , the outer surface detection result is determined to be consistent; otherwise, it is marked as a divergence point. The defect features marked as divergence points will be fed back to the convolutional neural network and used to adjust the model parameters; the calculation formula for the outer surface judgment threshold is: Where, T out is the threshold for external surface determination, T base is the basic judgment threshold, which is consistent with the value of the inner surface basic judgment threshold; σ design is the preset fatigue limit stress, σ current is the maximum stress value currently measured, σ yield is the yield strength of TA22 titanium alloy; The prediction unit predicts the evolution trend and expansion of defects based on the defect analysis results after review and judgment, including: According to the defect analysis results, the defect expansion path is predicted, specifically: P(t)=P0+(v0+β·K·t)·t Where, P(t) represents the spatial position of the defect at time t, P0 is the initial position of the defect, v0 is the initial expansion speed of the defect, β is the crack growth factor of TA22 titanium alloy, and K is the defect stress intensity factor; The prediction of the defect expansion size is as follows: Where S(t) is the defect size at time t, S0 is the initial defect size, k is the material expansion rate constant, and λ is the expansion saturation factor; The report generation unit generates a detection report according to the prediction results.

Images