Transportation hose defect ultrasonic detection method based on reciprocity and hybrid neural network

Abstract

The invention relates to the technical field of ultrasonic detection and noise reduction, and provides an external transmission hose defect ultrasonic detection method based on reciprocity and a hybrid neural network, and the method comprises the steps: carrying out the filtering processing of an initial signal, and obtaining a first signal; decomposing the first signal into an approximation coefficient and a first detail coefficient; performing soft threshold processing on the first detail coefficient to obtain a second detail coefficient, and reconstructing the second detail coefficient and the first approximation coefficient to obtain a second signal; performing normalization processing on the second signal to obtain a third signal; inputting the third signal into a convolutional neural network model to obtain a spatial feature vector; inputting the spatial feature vector into a recurrent neural network model, and extracting a time feature vector; fusing the spatial feature vector and the time feature vector to obtain a noise reduction signal; the convolutional neural network model and the recurrent neural network model are trained and then integrated with an ultrasonic detection system, defect characteristic signals are output, and the online detection precision and detection efficiency of the output hose are improved.

Description

Technical Field

[0001] The invention relates to the technical field of ultrasonic detection and noise reduction, and in particular to an ultrasonic detection method for defects of an external transmission hose based on reciprocity difference and a hybrid neural network. Background Art

[0002] Rubber hoses are widely used in the petroleum, chemical and other industries. Their quality and safety directly affect the stability and safety of production. Ultrasonic testing is one of the effective methods for detecting defects in rubber hoses. However, due to the complexity of rubber hose materials and the diversity of use environments, ultrasonic testing is often interfered by various noises, affecting the accuracy and reliability of detection. Therefore, how to effectively reduce the noise in ultrasonic testing has become an urgent problem to be solved.

[0003] When dealing with rubber hoses, existing ultrasonic testing methods are often subject to multiple interferences such as environmental noise, material noise, and testing equipment noise. The presence of these noises reduces the signal-to-noise ratio of the test signal, increases the uncertainty of the test results, and affects the accurate judgment of defects. In addition, existing noise reduction methods are not effective in dealing with complex noises and cannot meet the needs of high-precision testing. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the related art. To this end, the present invention provides an ultrasonic detection method for defects of an external transmission hose based on reciprocity difference and hybrid neural network, so as to improve the accuracy and efficiency of online detection of the external transmission hose.

[0005] The present invention provides an ultrasonic detection method for defects of an external transmission hose based on reciprocity difference and a hybrid neural network. Constructing a hybrid neural network model comprises the following steps: S1: The ultrasonic detection system detects and obtains an initial signal, and performs filtering on the initial signal to obtain a first signal; S2: decomposing the first signal into approximate coefficients and first detail coefficients; S3: performing soft threshold processing on the first detail coefficient to obtain a second detail coefficient, and reconstructing the second detail coefficient and the approximate coefficient to obtain a second signal; S4: performing normalization processing on the second signal to obtain a third signal; S5: Inputting the third signal into a convolutional neural network model to obtain a spatial feature vector of the third signal; S6: inputting the spatial feature vector into a recurrent neural network model to extract the temporal feature vector of the third signal; S7: fusing the spatial feature vector and the temporal feature vector to obtain a fused feature vector, and obtaining a noise reduction signal according to the fused feature vector; S8: A hybrid neural network model is obtained by jointly training the convolutional neural network model and the recurrent neural network model, the trained hybrid neural network model is integrated into the ultrasonic detection system, and a defect characteristic signal is output according to the noise reduction signal. According to the ultrasonic detection method for defects of an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, filtering the initial signal to obtain a first signal includes the following steps: The initial signal is processed by an adaptive filtering algorithm, wherein the adaptive filtering algorithm includes a minimum mean square error algorithm (LMS algorithm), and its formula is as follows: in, y ( t ) is the filter t The first signal of the output at time, w ( t ) is the filter t The filter coefficient vector at time , x ( t ) is the filter t The initial signal vector of the input at time , t =0, 1, 2, 3, ...; t The difference between the expected signal at the time and the first output signal is calculated as follows: in, d ( t ) is the filter t The expected signal of the moment, e ( t ) is the filter t The difference between the expected signal at the time instant and the first signal output; According to the calculated difference between the expected signal and the output first signal, the filter coefficient of the adaptive filtering algorithm is dynamically adjusted. The formula of the filter coefficient is as follows: in, w ( t +1) is the filter t The filter coefficient vector at time +1, μ is a learning factor that controls the speed at which the filter coefficients are updated.

[0006] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocal difference and hybrid neural network provided by the present invention, decomposing the first signal into approximate coefficients and first detail coefficients includes the following steps: performing continuous wavelet transform on the first signal, and decomposing the first signal into approximate coefficients and first detail coefficients of different scales through wavelet transform.

[0007] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, obtaining a second signal includes the following steps: The first detail coefficient is subjected to soft threshold processing to obtain the second detail coefficient, using the following formula: in, for t The first signal at the time The first detail coefficient of the first level is obtained after the soft threshold processing Level 2 detail coefficient, for t The first signal at the time Level 1 detail coefficient, is the threshold of the first detail coefficient; Reconstructing the second detail coefficient and the approximate coefficient comprises the following steps: in, for t The second signal of the moment, Indicates the first signal The approximate coefficient of the level.

[0008] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, obtaining the third signal includes the following steps: The second signal is normalized using the following formula: in, for t The third signal of the hour, is the minimum value of the second signal, is the maximum value of the second signal.

[0009] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, obtaining the spatial feature vector of the third signal includes the following steps: Inputting the third signal into a convolutional neural network model, wherein the convolution layer of the convolutional neural network model extracts local features of the third signal through multiple groups of convolution kernels; The pooling layer of the convolutional neural network model uses a maximum pooling or average pooling method to reduce the feature map dimension of the local feature; The fully connected layer of the convolutional neural network model performs feature fusion and outputs a spatial feature vector of the third signal.

[0010] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocal difference and hybrid neural network provided by the present invention, obtaining the time characteristic vector of the third signal includes the following steps: The spatial feature vector is input into the recurrent neural network model, and a long short-term memory network is used to capture the time dependency in the third signal to obtain the time feature vector of the third signal.

[0011] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocal difference and hybrid neural network provided by the present invention, fusing the spatial feature vector and the temporal feature vector comprises the following steps: The spatial feature vector and the temporal feature vector are concatenated into a fused feature vector, and the formula is as follows: in, is the spatial eigenvector, is the time feature vector, is the fused feature vector.

[0012] According to a method for ultrasonically detecting defects of an external transmission hose based on reciprocity difference and a hybrid neural network provided by the present invention, jointly training the convolutional neural network model and the recurrent neural network model to obtain a hybrid neural network model includes the following steps: Ultrasonic detection signals of a defect-free external transmission hose and a plurality of defective external transmission hoses are collected to form a data set, steps S1 to S7 are repeated for each group of detection signals in the data set, and the convolutional neural network model and the recurrent neural network model are jointly trained to obtain a hybrid neural network model.

[0013] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, the establishment of the data set includes the following steps: A defect-free external transmission hose and multiple defective external transmission hoses are provided, wherein the defect-free external transmission hose is provided with a first excitation probe and a first detection probe, and the defective external transmission hose is provided with a second detection probe and a second excitation probe, the position of the first excitation probe corresponds to the position of the second detection probe, the position of the first detection probe corresponds to the position of the second excitation probe, the first excitation probe sends out a first excitation signal, the first detection probe receives the first excitation signal and processes the first excitation signal to obtain a first ultrasonic detection signal, the second excitation probe sends out a second excitation signal, the second detection probe receives the second excitation signal and processes it to obtain a second ultrasonic detection signal, and the first ultrasonic detection signal and multiple second ultrasonic detection signals constitute the data set.

[0014] The above one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects: According to the ultrasonic detection method for defects of external transmission hoses based on reciprocity difference and hybrid neural network in an embodiment of the present invention, the problem of various types of noise affecting the accuracy of detection is solved by integrating the trained and optimized hybrid neural network model into the ultrasonic detection system, and real-time online noise reduction detection is realized, thereby improving the accuracy and efficiency of online detection of external transmission hoses.

[0015] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is a schematic flow chart of the method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention.

[0018] Figure 2 It is a structural schematic diagram of a defect-free external transport hose and a defective external transport hose provided by the present invention according to the method for ultrasonic detection of defects in external transport hoses based on reciprocity difference and hybrid neural network.

[0019] Figure 3 The output signals of the defect-free external transport hose and the defective external transport hose of the external transport hose defect ultrasonic detection method based on reciprocity difference and hybrid neural network provided by the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme in the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0021] In the description of the embodiments of the present invention, it should be noted that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the embodiments of the present invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limitations on the embodiments of the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0022] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiment of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0023] Figure 1 It is a schematic flow chart of the method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention.

[0024] The present invention provides an ultrasonic detection method for defects in an external transmission hose based on reciprocity difference and hybrid neural network. Figure 1 As shown, the following steps are included: S1: The ultrasonic detection system detects and obtains an initial signal, and performs filtering on the initial signal to obtain a first signal; S2: decomposing the first signal into approximate coefficients and first detail coefficients; S3: performing soft threshold processing on the first detail coefficient to obtain a second detail coefficient, and reconstructing the second detail coefficient and the approximate coefficient to obtain a second signal; S4: performing normalization processing on the second signal to obtain a third signal; S5: Inputting the third signal into a convolutional neural network model to obtain a spatial feature vector of the third signal; S6: inputting the spatial feature vector into a recurrent neural network model to extract the temporal feature vector of the third signal; S7: fusing the spatial feature vector and the temporal feature vector to obtain a fused feature vector, and obtaining a noise reduction signal according to the fused feature vector; S8: A hybrid neural network model is obtained by jointly training the convolutional neural network model and the recurrent neural network model, the trained hybrid neural network model is integrated into the ultrasonic detection system, and a defect characteristic signal is output according to the noise reduction signal.

[0025] In this embodiment, by integrating a trained and optimized hybrid neural network model into the ultrasonic detection system, the problem of various types of noise affecting the accuracy of detection is solved, and real-time online noise reduction detection is achieved, thereby improving the accuracy and efficiency of online detection of external transmission hoses.

[0026] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, filtering an initial signal to obtain a first signal comprises the following steps: The initial signal is processed by an adaptive filtering algorithm, wherein the adaptive filtering algorithm includes a minimum mean square error algorithm (LMS algorithm), and its formula is as follows: in, y ( t ) is the filter t The first signal of the output at time, w ( t ) is the filter t The filter coefficient vector at time , x ( t ) is the filter t The initial signal vector of the input at time , t =0, 1, 2, 3, ...; t The difference between the expected signal at the time and the first output signal is calculated as follows: in, d ( t ) is the filtert The expected signal of the moment, e ( t ) is the filter t The difference between the expected signal at the time instant and the first signal output; According to the calculated difference between the expected signal and the output first signal, the filter coefficient of the adaptive filtering algorithm is dynamically adjusted. The formula of the filter coefficient is as follows: in, w ( t +1) is the filter t The filter coefficient vector at time +1, μ is the learning factor, which is used to control the speed of updating the filter coefficients. Used to adjust the filter coefficients.

[0027] In this embodiment, the signal after the adaptive filtering process has a higher signal-to-noise ratio, providing a high-quality data basis for subsequent defect detection.

[0028] According to some embodiments of the present invention, the adaptive filtering process is to use an adaptive filtering algorithm to monitor the ultrasonic detection signal in real time and dynamically adjust the filter coefficients to suppress environmental noise and equipment noise.

[0029] According to some preferred embodiments of the present invention, the initial signal is filtered for the first time through a filter with set initial parameters to preliminarily suppress environmental noise and equipment noise and improve the signal-to-noise ratio of the signal. Then, the filtered signal is input into the LMS algorithm to determine the expected signal and calculate the difference between the expected signal and the output first signal. According to the calculated difference between the expected signal and the output first signal, the adaptive filtering algorithm dynamically adjusts the filter coefficients to minimize the difference between the expected signal and the output first signal. The adjustment process can use the gradient descent method to achieve iterative update of the filter coefficients. Repeat the above filtering and parameter adjustment process until the preset filtering effect is achieved or the difference between the expected signal and the output first signal is minimized. The signal after adaptive filtering has a high signal-to-noise ratio, providing a high-quality data basis for subsequent wavelet transform and defect detection.

[0030] According to some preferred embodiments of the present invention, the calculation formula of the signal-to-noise ratio is as follows: in, is the initial signal power, is the total number of signal sampling points, is the original signal, The signal after noise reduction is the noise signal power, is the noise signal, , SNR (dB) is the signal-to-noise ratio. The larger the SNR (dB), the better the noise reduction effect and the higher the signal quality.

[0031] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocal difference and hybrid neural network provided by the present invention, decomposing the first signal into approximate coefficients and first detail coefficients includes the following steps: performing continuous wavelet transform (CWT) on the first signal, and decomposing the first signal into approximate coefficients and first detail coefficients of different scales through wavelet transform.

[0032] In this embodiment, the first signal is decomposed into approximate coefficients and first detail coefficients of different scales to analyze the characteristics of the signal in different frequency and time ranges.

[0033] According to some preferred embodiments of the present invention, the approximation coefficient represents a feature of the first signal at a low-frequency scale, and the first detail coefficient represents a feature of the first signal at a high-frequency scale.

[0034] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, obtaining a second signal includes the following steps: The first detail coefficient is subjected to soft threshold processing to obtain the second detail coefficient, using the following formula: in, for t The first signal at the time The first detail coefficient of the first level is obtained after the soft threshold processing Level 2 detail coefficient, for t The first signal at the time Level 1 detail coefficient, is the threshold of the first detail coefficient; Reconstructing the second detail coefficient and the approximate coefficient comprises the following steps: in, for t The second signal of the moment, The first signal The approximate coefficient of the level.

[0035] In this embodiment, by performing soft threshold processing on the obtained first detail coefficients, the first detail coefficients greater than the threshold are smoothly reduced to suppress noise and retain important features in the signal, ensuring a natural transition of the coefficients near the threshold close to zero.

[0036] According to some embodiments of the present invention, a coefficient spectrum is obtained after CWT transformation, the time-frequency characteristics of the signal are analyzed, characteristic quantities related to the defect, such as energy, frequency distribution, instantaneous phase, etc., are extracted, and a defect recognition model is constructed to determine the defect location and nature in the signal.

[0037] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, obtaining the third signal includes the following steps: The second signal is normalized using the following formula: in, for t The third signal of the moment, is the minimum value of the second signal, is the maximum value of the second signal.

[0038] In this embodiment, the signal after wavelet transform is converted into a data format suitable for neural network processing through normalization processing, and the ultrasonic detection signal is normalized so that its amplitude range is scaled to [0, 1] to improve the training effect of the neural network.

[0039] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, obtaining the spatial feature vector of the third signal includes the following steps: Inputting the third signal into a convolutional neural network model (CNN), wherein the convolution layer of the convolutional neural network model extracts local features of the third signal through multiple groups of convolution kernels; The pooling layer of the convolutional neural network model uses a maximum pooling or average pooling method to reduce the feature map dimension of the local feature; The fully connected layer of the convolutional neural network model performs feature fusion and outputs a spatial feature vector of the third signal.

[0040] In this embodiment, the spatial feature vector of the third signal is extracted to subsequently identify the position of the defect.

[0041] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocal difference and hybrid neural network provided by the present invention, obtaining the time characteristic vector of the third signal includes the following steps: The spatial feature vector is input into the recurrent neural network model (RNN), and a long short-term memory network is used to capture the time dependency in the third signal to obtain a time feature vector of the third signal.

[0042] In this embodiment, by extracting the time feature vector of the third signal, the noise suppression effect is improved, which facilitates the detection of the nature and severity of the defect.

[0043] According to some embodiments of the present invention, the long short-term memory network solves the gradient vanishing problem, thereby improving the learning ability. The RNN processes dynamic changes through internal memory units, extracts time series features, and improves the noise suppression effect.

[0044] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocal difference and hybrid neural network provided by the present invention, fusing the spatial feature vector and the temporal feature vector comprises the following steps: The spatial feature vector and the temporal feature vector are concatenated into a fused feature vector, and the formula is as follows: in, is the spatial eigenvector, is the time feature vector, is the fused feature vector.

[0045] In this embodiment, the spatial feature vector and the temporal feature vector are spliced ​​into a fused feature vector to ensure that the final feature can effectively represent the characteristics of the signal.

[0046] According to some embodiments of the present invention, the output features of CNN and RNN are fused, and they are combined using a fully connected layer. The concatenated high-dimensional feature vectors are further mapped and fused to obtain a final denoised signal. The denoising effect is evaluated by calculating indicators such as the signal-to-noise ratio (SNR) to ensure that the quality of the final denoised signal is improved.

[0047] According to some embodiments of the present invention, obtaining the final noise reduction signal includes the following steps: The fused feature vector is passed to one or more fully connected layers for further feature fusion and dimensionality compression. The function of the fully connected layer is to linearly combine all input features, and then implement nonlinear mapping through the activation function to finally obtain the final noise reduction signal. The mathematical form of the fully connected layer can be expressed as: in, is the weight matrix of the fully connected layer, is the bias vector, is the activation function, is the final denoised signal, that is, the output feature after being processed by the fully connected layer.

[0048] According to a method for ultrasonically detecting defects of an external transmission hose based on reciprocity difference and a hybrid neural network provided by the present invention, jointly training the convolutional neural network model and the recurrent neural network model to obtain a hybrid neural network model includes the following steps: Ultrasonic detection signals of a defect-free external transmission hose and a plurality of defective external transmission hoses are collected to form a data set, steps S1 to S7 are repeated for each group of detection signals in the data set, and the convolutional neural network model and the recurrent neural network model are jointly trained to obtain a hybrid neural network model.

[0049] In this embodiment, a hybrid neural network model is obtained by jointly training the convolutional neural network model and the recurrent neural network model, thereby improving the performance and applicability of the hybrid neural network model in ultrasonic noise reduction detection of rubber hoses.

[0050] According to some preferred embodiments of the present invention, during the training process of the hybrid neural network model, the data set is divided into a training data set and a validation data set using a cross-validation method, and the back-propagation algorithm is used to optimize the model parameters and the mean square error to improve the performance of the hybrid neural network model. In order to prevent overfitting, a regularization strategy and an early stopping mechanism are implemented to ensure that the performance of the hybrid neural network model on the validation data set is consistent with that on the training data set.

[0051] According to some preferred embodiments of the present invention, the optimization of the hybrid neural network model is a continuous process. After preliminary verification, the hyperparameters of the hybrid neural network model are continuously adjusted according to the verification results, including learning rate, batch size, network structure, etc. The best hyperparameter combination is found through strategies such as grid search and random search to improve the noise reduction effect of the hybrid neural network model.

[0052] According to some preferred embodiments of the present invention, the trained hybrid neural network model will be integrated into the ultrasonic detection system to achieve real-time online noise reduction detection. The hybrid neural network model can continue to learn in actual detection, combined with user feedback, and continuously optimized and adjusted to adapt to the complex and changeable rubber hose detection environment. Through this dynamic learning mechanism, the system can maintain efficient noise reduction performance and ensure the accuracy and reliability of ultrasonic detection results.

[0053] Figure 2 It is a structural schematic diagram of a defect-free external transport hose and a defective external transport hose provided by the present invention according to the method for ultrasonic detection of defects in external transport hoses based on reciprocity difference and hybrid neural network.

[0054] According to a method for ultrasonic detection of defects in an external transmission hose based on reciprocity difference and hybrid neural network provided by the present invention, the establishment of the data set includes the following steps: Set up a defect-free external hose and multiple defect external hoses, such as Figure 2 As shown, the defect-free external transmission hose is provided with a first excitation probe and a first detection probe, and the defective external transmission hose is provided with a second detection probe and a second excitation probe, the position of the first excitation probe corresponds to the position of the second detection probe, the position of the first detection probe corresponds to the position of the second excitation probe, the first excitation probe sends out a first excitation signal, the first detection probe receives the first excitation signal and processes the first excitation signal to obtain a first ultrasonic detection signal, the second excitation probe sends out a second excitation signal, the second detection probe receives the second excitation signal and processes it to obtain a second ultrasonic detection signal, and the first ultrasonic detection signal and a plurality of the second ultrasonic detection signals constitute the data set.

[0055] In this embodiment, by arranging the first excitation probe and the first detection probe and the second detection probe and the second excitation probe at opposite positions, the specific position of the defect can be analyzed and determined by using the acoustic reciprocity difference principle.

[0056] According to some preferred embodiments of the present invention, for a defect-free external transmission hose, the first excitation probe sends a preset light signal, and the first detection probe receives the returned signal. In this process, the detection system needs to be initially calibrated to ensure the stability and accuracy of the signal transmission path. The calibration process includes adjusting the position and angle of the first excitation probe and the first detection probe to obtain the optimal signal reception effect, and recording the intensity and phase characteristics of the reference signal.

[0057] According to some embodiments of the present invention, for a defective external transmission hose, the second excitation probe sends a preset light signal, and the second detection probe receives the returned signal. At this time, recalibration is required to ensure the consistency of the signal transmission path under different excitation and detection conditions. The signal acquisition needs to be repeated multiple times at the same position to ensure the accuracy and stability of the data. After each signal acquisition, the detection system automatically records the signal intensity, phase and other related characteristics, and performs preliminary comparative analysis to identify possible abnormal signal characteristics.

[0058] Figure 3 The output signals of the defect-free external transport hose and the defective external transport hose of the external transport hose defect ultrasonic detection method based on reciprocity difference and hybrid neural network provided by the present invention.

[0059] According to some embodiments of the present invention, Figure 3 As shown, the trained hybrid neural network model is integrated into the ultrasonic detection system, and the first ultrasonic detection signal and the second ultrasonic detection signal are subjected to noise reduction processing by the hybrid neural network, and the defect position of the defective external transmission hose can be obtained by comparison.

[0060] According to some embodiments of the present invention, a large amount of ultrasonic inspection data of external transmission hoses is collected and annotated to construct a training data set and a verification data set. Then, the hybrid neural network model is trained using the training data set, and the third signal is input into the hybrid neural network for intelligent noise reduction processing. First, CNN is used to extract the spatial feature vector of the signal, and the signal is feature extracted through the convolution kernel. In this process, the ReLU activation function is used to accelerate the convergence of training. Then, the spatial feature vector extracted by CNN is input into RNN, which is used to process the time series characteristics of the signal, aiming to capture the dependency of the signal over time.

[0061] According to some embodiments of the present invention, during the training of the hybrid neural network model, the cross-validation method is used to divide the training data set and the validation data set, and the forward and reverse propagation algorithms are used to optimize the model parameters and the mean square error to improve the performance of the hybrid neural network model. In order to prevent overfitting, the L2 regularization strategy and early stopping mechanism are implemented, and the two methods are combined to ensure that the performance of the model on the validation data set is consistent with the training data set.

[0062] According to some embodiments of the present invention, the early stopping mechanism is a strategy for determining when to stop training by dynamically monitoring the performance of the validation data set during the model training process. When the model shows good fitting ability on the training data set, but the performance on the validation data set begins to deteriorate (i.e., the loss of the validation data set begins to rise), it means that the model has begun to overfit. During the training process, the loss function on the validation data set is monitored, and when the loss of the validation data set no longer decreases after several iterations (i.e., exceeds the set tolerance number), the training is stopped. This ensures that the model stops training when the best generalization performance is achieved, avoiding the risk of overfitting caused by continued training.

[0063] According to some embodiments of the present invention, the model is strictly controlled at each step of the training. L2 regularization limits the complexity of the model by adding a penalty term in the loss function, making the model more robust when facing new data. The early stopping mechanism dynamically determines when to stop training by detecting the changing trend of the loss of the validation data set to ensure that the model does not deteriorate on the validation data set. Through this combination, the performance of the model on the training data set and the validation data set is more consistent, significantly improving the generalization ability and reliability in practical applications.

[0064] According to some embodiments of the present invention, model optimization is a continuous process. After preliminary verification, the model's hyperparameters, including learning rate, batch size, network structure, etc., are continuously adjusted according to the verification results. The magnitude of the learning rate determines the magnitude of the weight update in each iteration. A learning rate that is too high will lead to unstable training, while a learning rate that is too low will lead to slow convergence. The learning rate is generally set at [10 −2 , 10−1 ], the system does not make adjustments, otherwise it makes adjustments. The batch size controls the number of samples used in each iteration of training. Small batch sizes help to better find the global optimum, but the training time is longer; large batches speed up training but may increase the risk of overfitting. Possible values ​​for batch size can be 16, 32, 64, 128, and 256. The depth (number of layers) and width (number of neurons in each layer) of the network in the network structure are also part of the hyperparameter combination. If the loss of the validation dataset is high and the loss of the training dataset is low, it means that the model is overfitting, and the number of network layers or the number of neurons in each layer can be reduced. If the losses of both the solicitation and training datasets are high, it means that the complexity of the model may not be enough, and the performance of the model can be improved by adding convolutional layers or fully connected layers and increasing the number of neurons.

[0065] According to some embodiments of the present invention, the set of multiple possible configurations formed by the values ​​of different hyperparameters during model training is called a hyperparameter combination. By selecting and optimizing these hyperparameter combinations, the noise reduction effect of the model can be improved, so that it performs well on both the validation data set and the training data set, thereby ensuring that the model has good generalization ability and application value. To improve the noise reduction effect of the model. After training and optimizing the model, the system has the function of high-precision detection of defects in ultrasonic signals, and can identify the specific location, type and severity of the defects while reducing noise.

[0066] According to some embodiments of the present invention, in the implementation of defect detection, the model first identifies the specific location of the abnormal signal based on the spatial feature vector of the third signal. For the identification of the defect location, the system uses the spatial feature vector extracted by CNN to enable it to accurately locate the area showing abnormality in the signal. In addition, the model further identifies the nature and evolution trend of the defect through the time-dependent features captured by RNN. With this feature fusion method, the system can comprehensively analyze the characteristics of the signal, effectively distinguish different defect types, such as cracks, wear, etc., and judge the severity of the defect based on the intensity and frequency changes of the defect signal.

[0067] According to some embodiments of the present invention, finally, the entire system completes signal acquisition under the reciprocity difference principle through the connection between the probe and the data acquisition device, and transmits the signal to the data preprocessing module for noise reduction. The trained hybrid neural network model further performs deep feature extraction and noise reduction on the signal to realize the real-time online defect recognition function of ultrasonic detection. After the model is integrated into the ultrasonic detection system, it can identify the specific location, type and severity of the defect during the detection process, and present the defect information to the operator to support real-time decision-making and operation. The system not only displays the location and nature of the defect, but also provides detailed detection data, including key data such as detection time, defect location, type, severity and signal-to-noise ratio (SNR) to support subsequent analysis and management of the data. At this point, the model can continue to learn in actual detection, combined with user feedback, and continuously optimize and adjust to adapt to the complex and changeable rubber hose detection environment. Through this dynamic learning mechanism, the system can maintain efficient noise reduction performance and ensure the accuracy and reliability of ultrasonic detection results.

[0068] Compared with the prior art, the ultrasonic detection method for defects of an external transmission hose based on reciprocity difference and hybrid neural network in the embodiment of the present invention has the following technical effects: The ultrasonic detection method for defects in external transmission hoses based on reciprocity difference and hybrid neural network proposed in the present invention realizes the accurate comparison of ultrasonic signals between defective samples and defect-free reference samples by introducing the principle of "reciprocity difference", laying a solid data foundation for the detection process. Combining the advantages of CNN and RNN, it not only deeply analyzes the spatial characteristics of ultrasonic signals, but also carefully captures its time series characteristics, enriching the dimension of defect identification. In the complex and changeable ultrasonic scattering wave field, the system can efficiently identify and accurately extract defect characteristic signals, improve the accuracy and reliability of external transmission hose defect detection, provide strong technical support for safe production and pipeline maintenance, and reduce potential safety risks and economic losses.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for ultrasonic detection of defects in external transmission hoses based on reciprocity difference and hybrid neural network, characterized in that: The following steps are involved: S1: The ultrasonic detection system detects and obtains an initial signal, and performs filtering on the initial signal to obtain a first signal; S2: decomposing the first signal into approximate coefficients and first detail coefficients; S3: performing soft threshold processing on the first detail coefficient to obtain a second detail coefficient, and reconstructing the second detail coefficient and the approximate coefficient to obtain a second signal; S4: performing normalization processing on the second signal to obtain a third signal; S5: Inputting the third signal into a convolutional neural network model to obtain a spatial feature vector of the third signal; S6: inputting the spatial feature vector into a recurrent neural network model to extract the temporal feature vector of the third signal; S7: fusing the spatial feature vector and the temporal feature vector to obtain a fused feature vector, and obtaining a noise reduction signal according to the fused feature vector; S8: A hybrid neural network model is obtained by jointly training the convolutional neural network model and the recurrent neural network model, the trained hybrid neural network model is integrated into the ultrasonic detection system, and a defect characteristic signal is output according to the noise reduction signal.

2. The ultrasonic detection method for defects of an external transmission hose based on reciprocity difference and hybrid neural network according to claim 1 is characterized in that: Filtering the initial signal to obtain the first signal comprises the following steps: The initial signal is processed by an adaptive filtering algorithm, wherein the adaptive filtering algorithm includes a minimum mean square error algorithm, and its formula is as follows: in, y ( t ) is the filter t The first signal of the output at time, w ( t ) is the filter t The filter coefficient vector at time , x ( t ) is the filter t The initial signal vector of the input at time , t =0, 1, 2, 3, ...; t The difference between the expected signal at the time and the first output signal is calculated as follows: in, d ( t ) is the filter t The expected signal of the moment, e ( t ) is the filter t The difference between the expected signal at the time instant and the first signal output; According to the calculated difference between the expected signal and the output first signal, the filter coefficient of the adaptive filtering algorithm is dynamically adjusted. The formula of the filter coefficient is as follows: in, w ( t +1) is the filter t The filter coefficient vector at time +1, μ is a learning factor that controls the speed at which the filter coefficients are updated.

3. The ultrasonic detection method for defects of an external transmission hose based on reciprocity difference and hybrid neural network according to claim 2 is characterized in that: Decomposing the first signal into approximate coefficients and first detail coefficients comprises the following steps: performing continuous wavelet transform on the first signal, and decomposing the first signal into approximate coefficients and first detail coefficients of different scales through wavelet transform.

4. The method for ultrasonic detection of defects in external transmission hoses based on reciprocity difference and hybrid neural network according to claim 3 is characterized in that: Obtaining the second signal comprises the following steps: The first detail coefficient is subjected to soft threshold processing to obtain the second detail coefficient, using the following formula: in, for t The first signal at the time The first detail coefficient of the first level is obtained after the soft threshold processing Level 2 detail coefficient, for t The first signal at the time Level 1 detail coefficient, is the threshold of the first detail coefficient, j =0, 1, 2, 3, ...; Reconstructing the second detail coefficient and the approximate coefficient comprises the following steps: in, for t The second signal of the moment, The first signal The approximate coefficient of the level.

5. The method for ultrasonic detection of defects in external transmission hoses based on reciprocity difference and hybrid neural network according to claim 4 is characterized in that: Obtaining the third signal comprises the following steps: The second signal is normalized using the following formula: in, for t The third signal of the hour, is the minimum value of the second signal, is the maximum value of the second signal.

6. The method for ultrasonic detection of defects in external transmission hoses based on reciprocity difference and hybrid neural network according to claim 5 is characterized in that: Obtaining the spatial feature vector of the third signal comprises the following steps: Inputting the third signal into a convolutional neural network model, wherein the convolution layer of the convolutional neural network model extracts local features of the third signal through multiple groups of convolution kernels; The pooling layer of the convolutional neural network model uses a maximum pooling or average pooling method to reduce the feature map dimension of the local feature; The fully connected layer of the convolutional neural network model performs feature fusion and outputs a spatial feature vector of the third signal.

7. The method for ultrasonic detection of defects in external transmission hoses based on reciprocity difference and hybrid neural network according to claim 6 is characterized in that: Obtaining the time characteristic vector of the third signal comprises the following steps: The spatial feature vector is input into the recurrent neural network model, and a long short-term memory network is used to capture the time dependency in the third signal to obtain the time feature vector of the third signal.

8. The method for ultrasonic detection of defects in external transmission hoses based on reciprocity difference and hybrid neural network according to claim 7 is characterized in that: The spatial feature vector and the temporal feature vector are merged The following steps are involved: The spatial feature vector and the temporal feature vector are concatenated into a fused feature vector, and the formula is as follows: in, is the spatial eigenvector, is the time feature vector, is the fused feature vector.

9. The ultrasonic detection method for defects of an external transmission hose based on reciprocity difference and hybrid neural network according to claim 1 is characterized in that: Jointly training the convolutional neural network model and the recurrent neural network model to obtain a hybrid neural network model includes the following steps: Ultrasonic detection signals of a defect-free external transmission hose and a plurality of defective external transmission hoses are collected to form a data set, steps S1 to S7 are repeated for each group of detection signals in the data set, and the convolutional neural network model and the recurrent neural network model are jointly trained to obtain a hybrid neural network model.

10. The method for ultrasonic detection of defects in external transmission hoses based on reciprocity difference and hybrid neural network according to claim 9 is characterized in that: The establishment of the data set includes the following steps: A defect-free external transmission hose and multiple defective external transmission hoses are provided, wherein the defect-free external transmission hose is provided with a first excitation probe and a first detection probe, and the defective external transmission hose is provided with a second detection probe and a second excitation probe, the position of the first excitation probe corresponds to the position of the second detection probe, the position of the first detection probe corresponds to the position of the second excitation probe, the first excitation probe sends out a first excitation signal, the first detection probe receives the first excitation signal and processes the first excitation signal to obtain a first ultrasonic detection signal, the second excitation probe sends out a second excitation signal, the second detection probe receives the second excitation signal and processes it to obtain a second ultrasonic detection signal, and the first ultrasonic detection signal and multiple second ultrasonic detection signals constitute the data set.

Images