An ultrasonic rapid detection method and system for the vicat softening temperature of a cable protection pipe

Abstract

The present application relates to the technical field of nondestructive testing, and particularly relates to a cable protection pipe Vicat softening temperature ultrasonic rapid detection method and system, comprising: collecting ultrasonic original A scanning data through a longitudinal wave probe, extracting ultrasonic key parameters as characteristic quantities, setting up a training set with the Vicat softening temperature as an output quantity, carrying out Gaussian process regression on the data, taking the least square error as an evaluation quantity to carry out Bayesian optimization on the model, obtaining a Vicat softening temperature prediction model, collecting ultrasonic longitudinal wave signals of a cable protection pipe to be detected, extracting corresponding ultrasonic parameters and inputting the prediction model to obtain a predicted value of the Vicat softening temperature, when the predicted value and the actual value differ greatly, adding the measurement data to the training set to retrain and correct the model, improving the universality of the model, compared with a traditional detection method, the portability and detection efficiency are significantly improved, nondestructive detection is realized, the detection demand of an engineering site is better adapted, and the method has great market application prospect and popularization value.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing technology, and in particular to an ultrasonic rapid testing method and system for the Vicat softening temperature of cable protection pipes. Background Technology

[0002] Cable protection conduits are laid on the outer layer of cables to effectively prevent short circuits and ablation accidents caused by cable damage or breakage. They also provide some isolation against magnetic field interference from power lines, making them a crucial guarantee for the stable and safe operation of power supply units. The Vicat softening temperature of cable protection conduits directly affects their performance characteristics. It is an indicator of the material's heat resistance, reflecting the changes in the product's physical properties under heating conditions. It is also one of the core indicators for judging the quality of cable protection conduits, verifying whether the product meets usage requirements, and evaluating its ability to operate normally and stably under power grid service conditions.

[0003] Currently, the Vicat softening temperature is commonly tested using the following method: The sample to be tested is placed in a liquid heat transfer medium, and under a certain load and a constant rate of heating, the sample is tested for softening by 1 mm. 2 The temperature at which the indenter is pressed to a depth of 1 mm is the Vicat softening temperature of the sample. This method has strict requirements on the size and shape of the sample, resulting in high testing costs and low efficiency. It is only suitable for sampling testing. Furthermore, this method is a destructive test that requires sampling, and the testing instrument is bulky, which cannot meet the needs of non-destructive, rapid, and portable testing in engineering fields.

[0004] In conclusion, there is a need to find a new method for detecting Vicat softening temperature or to improve existing methods to meet the requirements of rapid and non-destructive testing in cable protection pipe engineering projects. This has significant research value and practical significance. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an ultrasonic rapid detection method and system for the Vicat softening temperature of cable protection pipes, which effectively solves the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: an ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes, comprising the following steps: S1: Use an ultrasonic longitudinal wave probe to collect raw data from the cable protection pipe. The raw data obtained is ultrasonic A-scan data along the pipe thickness direction. S2: Analyze and process the ultrasonic A-scan data obtained in S1, extract key parameters, use the ultrasonic parameters as feature values, and use the Vicat softening temperature of the corresponding cable protection pipe as the output value to establish a dataset. S3: Perform Gaussian process regression on the dataset obtained in S2, and use the least squares error as the evaluation value to perform Bayesian optimization on the model to obtain the optimized Vicat softening temperature prediction model. S4: In actual testing, ultrasonic A-scan data of the cable protection pipe under test is collected, the corresponding parameters are extracted, and the Vicat softening temperature prediction model in S3 is input to obtain the ultrasonic characterization value of the Vicat softening temperature. S5: Randomly verify the results using existing testing methods. If the model prediction results differ significantly from the actual values, the detection data can be added to the training set to retrain and optimize the model.

[0007] Furthermore, the raw data acquisition should be performed in a distributed manner along the axial and circumferential directions of the protective tube.

[0008] Furthermore, the horizontal axis of the ultrasonic A-scan data represents the signal propagation time, and the vertical axis represents the acoustic signal intensity. The data should include the primary and secondary reflected echoes from the inner wall of the cable protection pipe.

[0009] Furthermore, the extracted key parameters include ultrasonic velocity, ultrasonic attenuation, and frequency-attenuation variation characteristics calculated based on the ultrasonic spectrum.

[0010] Furthermore, the frequency-attenuation variation characteristic is obtained by the following steps: (1) The ultrasound A-scan signal is intercepted, and the primary and secondary reflected echoes of the inner wall of the tube are extracted; (2) Perform discrete Fourier transform on the first and second reflected echo regions of the inner wall of the pipe to obtain the corresponding regional spectrum; (3) Based on the amplitude of each frequency component in the regional spectrum, calculate the attenuation of ultrasonic components at different frequencies and obtain the ultrasonic attenuation characteristics that vary with frequency.

[0011] Furthermore, cross-validation is used during the training of Gaussian process regression in step 3.

[0012] Furthermore, the cross-validation method involves randomly dividing a portion of the training set as a validation set during the training process, and performing real-time prediction and error calculation on the trained model.

[0013] Furthermore, the training process of the Gaussian process regression specifically includes the following steps: Data preparation: Collect and prepare input features and corresponding output data; Choosing a covariance function: This measures the similarity between input features and determines the covariance function used to describe the similarity between samples. Constructing the covariance matrix: Calculate the covariance matrix between input features using the selected covariance function; Adding noise: A small noise variance is added to the covariance matrix to capture observation errors; Model training: Estimate the model's hyperparameters through Bayesian inference; Prediction: Using the trained model, make predictions on new input features.

[0014] Furthermore, the Bayesian optimization involves changing the model's hyperparameters and recording the corresponding least squared error during training, while simultaneously predicting the minimum value of the error function and iteratively approaching the minimum value.

[0015] Furthermore, in step S3, the specific steps of the Bayesian optimization include: S31. Define the hyperparameter space: Determine the hyperparameters to be optimized and their possible value ranges; S32. Initialize sample data: Randomly select a set of hyperparameters as initial values; S33. Training the model: Train the model using the training data and the current hyperparameter settings, and calculate the prediction results; S34. Evaluate the objective function: Compare the model's predictions with the actual observations, and calculate the least squared error as the value of the objective function; S35. Update model state: Record the current hyperparameters and the corresponding least squares error; S36. Fit a probabilistic model based on existing data: Based on historical hyperparameter-least squared error pairs, establish a probabilistic model to describe the relationship between hyperparameters and errors. S37. Parallel Sampling: Obtaining new candidate hyperparameter settings from the conditional probability distribution through sampling methods; S38. Select the optimal value of the objective function for the next iteration: Using the new hyperparameter settings, repeat steps S33 to S37 until the preset number of iterations is reached or the termination criterion is met. S39. Return the best hyperparameter: Based on the historical hyperparameter-least squared error pairs, select the hyperparameter with the smallest error as the best hyperparameter.

[0016] Furthermore, the random validation is based on existing Vicat softening temperature testing standards, and the results are recorded as actual values. These are compared with the predicted values ​​obtained by the model, and the error should be less than or equal to the overall variance of the model.

[0017] Furthermore, in step 5, retraining and optimizing the model involves adding data with large prediction errors to the training set and repeating the model training and optimization steps in step S3.

[0018] Furthermore, the longitudinal wave probe is placed perpendicular to the protective tube, and the returned echo signal is amplified with a high signal-to-noise ratio.

[0019] This invention also provides an ultrasonic rapid detection system for the Vicat softening temperature of cable protection pipes, comprising: The data acquisition module uses an ultrasonic longitudinal wave probe to acquire raw data from the cable protection pipe, obtaining ultrasonic A-scan data along the pipe thickness direction; The data processing and feature extraction module analyzes and processes the acquired ultrasonic A-scan data, extracts key parameters as feature values, and uses the ultrasonic parameters and the Vicat softening temperature of the cable protection pipe as output values ​​to establish a dataset. The model optimization module, based on the obtained dataset, adopts a Gaussian process regression model and performs Bayesian optimization using the least squares error as the evaluation criterion to obtain the optimized Vicat softening temperature prediction model. The temperature prediction module is used to collect ultrasonic A-scan data of the cable protection pipe under inspection, extract the corresponding parameters, and input these parameters into the optimized Vicat softening temperature prediction model to obtain the ultrasonic characterization value of the Vicat softening temperature. The results verification module verifies the model's prediction results using existing random verification methods. If the model's prediction results differ significantly from the actual values, these detection data can be added to the training set for model retraining and optimization.

[0020] The present invention also provides a computer device including a memory and a processor, wherein the memory stores steps that can be loaded by the processor and executed by the ultrasonic rapid detection method.

[0021] The present invention also provides a computer-readable storage medium storing a processor-executable program, which, when executed by a processor, is used to perform the ultrasonic rapid detection method.

[0022] The beneficial effects of this invention are as follows: This invention proposes an ultrasonic characteristic quantity associated with the Vicat softening temperature, and realizes rapid ultrasonic detection of the Vicat softening temperature through an optimizable Gaussian process. Compared with traditional detection methods, the portability and detection efficiency of the detection instrument are significantly improved, and non-destructive testing can be achieved. It can better adapt to the detection needs of engineering sites and has great market application prospects and promotion value. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the workflow of the ultrasonic rapid detection method in an embodiment of the present invention; Figure 2 This is a schematic diagram of the Bayesian optimization results in an embodiment of the present invention; Figure 3 This is a schematic diagram of the model cross-validation results in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the measurement error of the Vicat softening temperature prediction model in an embodiment of the present invention. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0026] It should be noted that when an element is referred to as being "fixed to" another element, it can be directly attached to the other element or there may be an intervening element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0027] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0028] like Figure 1 The ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes, as shown, includes the following steps: S1: Use an ultrasonic longitudinal wave probe to collect raw data from the cable protection pipe. The raw data obtained is ultrasonic A-scan data along the pipe thickness direction. S2: Analyze and process the ultrasonic A-scan data obtained in S1, extract key parameters, use the ultrasonic parameters as feature values, and use the Vicat softening temperature of the corresponding cable protection pipe as the output value to establish a dataset. S3: Perform Gaussian process regression on the dataset obtained in S2, and use the least squares error as the evaluation value to perform Bayesian optimization on the model to obtain the optimized Vicat softening temperature prediction model. S4: In actual testing, ultrasonic A-scan data of the cable protection pipe under test is collected, the corresponding parameters are extracted, and the Vicat softening temperature prediction model in S3 is input to obtain the ultrasonic characterization value of the Vicat softening temperature. S5: Randomly verify the results using existing testing methods. If the model prediction results differ significantly from the actual values, the detection data can be added to the training set to retrain and optimize the model.

[0029] In this invention, raw ultrasonic A-scan data is acquired using a longitudinal wave probe to obtain internal structural information of the cable protection pipe. Useful key ultrasonic parameters are extracted from the ultrasonic data as feature quantities. A dataset is established with Vicat softening temperature as the output quantity. Gaussian process regression is performed on the dataset to establish a model that can accurately predict and estimate the Vicat softening temperature of the cable protection pipe. During model training, a Bayesian optimizer is used to iteratively optimize the model multiple times to obtain the minimum squared error, further improving the model performance and thus obtaining the optimized Vicat softening temperature prediction model. When the cable protection pipe is actually inspected, the ultrasonic longitudinal wave signal of the cable protection pipe to be inspected is acquired, the corresponding ultrasonic parameters are extracted and input into the prediction model, and the predicted value of Vicat softening temperature can be obtained, realizing the ultrasonic characterization of Vicat softening temperature. The prediction model has a feedback function. When the predicted value differs greatly from the actual value, the measured data can be added to the training set to retrain and correct the model, improving the model's universality. This invention proposes an ultrasonic characteristic quantity associated with Vicat softening temperature, and realizes rapid ultrasonic detection of Vicat softening temperature through an optimizable Gaussian process. Compared with traditional detection methods, the portability and detection efficiency of the detection instrument are significantly improved, and non-destructive testing can be achieved. It can better meet the detection needs of engineering sites and has great market application prospects and promotion value.

[0030] In step S1, the raw data acquisition should involve distributed sampling in both the axial and circumferential directions of the protective pipe. By sampling in multiple directions, ultrasonic data from different directions is collected, comprehensively detecting the internal structure and defects of the pipe. Furthermore, the ultrasonic data from different directions provides the possibility of cross-validation, enabling a more accurate determination of the location of defects inside the pipe. Ultrasonic A-scan data is a technique used for non-destructive testing. In this solution, ultrasonic A-scan data refers to ultrasonic testing performed along the pipe wall thickness direction. By analyzing this data, changes in pipe wall thickness, potential defects, or anomalies can be identified. This is of great significance for pipeline safety and quality control in the industrial field. In this invention, the horizontal axis of the ultrasonic A-scan data represents signal propagation time, and the vertical axis represents acoustic signal intensity. The data should include primary and secondary reflected echoes from the inner wall of the cable protection pipe. The formation process of the primary and secondary reflected echoes is as follows: When ultrasonic waves are emitted from the ultrasonic probe, they pass through the medium (such as the cable protection pipe) and reach the interface of the medium (such as the pipe wall). A portion of the ultrasonic energy is reflected back from the interface, forming a primary reflected echo. This echo often corresponds to the contact surface of the inner wall of the cable protection pipe and can be used to determine the pipe wall thickness. When the primary reflected echo reaches the ultrasonic probe, it may be reflected back again, forming a secondary reflected echo. This echo corresponds to other interfaces, inhomogeneities, or defects inside the cable protection pipe, and can provide information about the internal structure of the pipe and potential problems. By analyzing the primary and secondary reflected echoes in the ultrasonic A-scan data, we can understand the internal structure of the cable protection pipe, changes in wall thickness, corrosion, cracks, and other conditions, and further assess the health status of the pipe.

[0031] In step S2, the extracted key parameters include ultrasonic velocity, ultrasonic attenuation, and frequency-attenuation variation characteristics calculated from the ultrasonic spectrum. Ultrasonic velocity refers to the speed at which ultrasound waves propagate in a medium. By analyzing the relationship between ultrasonic propagation time and distance, the ultrasonic velocity in the material can be calculated. Measuring the ultrasonic velocity allows for the assessment of uniformity and density variations. Ultrasonic waves experience energy loss during propagation in a medium, meaning the amplitude gradually decreases; this phenomenon is called ultrasonic attenuation. By analyzing the change in ultrasonic amplitude with propagation distance, the ultrasonic attenuation coefficient of the medium can be calculated. Ultrasonic attenuation can be used to assess the sound absorption performance, defects, and fatigue of materials. The ultrasonic spectrum is the frequency representation of the ultrasonic signal. By analyzing the ultrasonic spectrum, the frequency-attenuation variation characteristics can be obtained. This means we can observe the attenuation of ultrasound waves at different frequencies, usually represented by attenuation curves. Frequency-attenuation variation characteristics can be used to detect defects, foreign objects, or other inhomogeneities in materials and provide more detailed information for assessing material quality and reliability.

[0032] Specifically, the frequency-attenuation variation characteristic is obtained by the following steps: (1) The ultrasound A-scan signal is intercepted, and the primary and secondary reflected echoes of the inner wall of the tube are extracted; Extract the reflected echo signals of interest from the raw data. Primary reflected echoes typically correspond to the contact surfaces of the pipe's inner wall, while secondary reflected echoes may correspond to other interfaces, inhomogeneities, or defects within the pipe. By intercepting and extracting these reflected echoes, we can define the area of ​​interest and reduce the complexity of the analysis.

[0033] (2) Perform discrete Fourier transform on the first and second reflected echo regions of the inner wall of the pipe to obtain the corresponding regional spectrum; The Discrete Fourier Transform (DFT) is a mathematical tool for converting time-domain signals into frequency-domain signals. By performing a DFT on the first / secondary reflection echo region, we can obtain the corresponding spectrum, i.e., amplitude information at different frequencies. This makes signal analysis in the frequency domain more intuitive and convenient, and helps to observe the characteristics of different frequency components.

[0034] (3) Based on the amplitude of each frequency component in the regional spectrum, calculate the attenuation of ultrasonic components at different frequencies and obtain the ultrasonic attenuation characteristics that vary with frequency.

[0035] It can facilitate the prediction of signal intensity attenuation by analyzing ultrasonic attenuation characteristics, thereby improving the accuracy and precision of detection results.

[0036] In step 3 of this scheme, the Gaussian process regression training process adopts the cross-validation method, which can objectively evaluate the performance of the model. By dividing the dataset into training set and validation set, the performance of the model is evaluated on the validation set, which can more accurately estimate the generalization ability of the model. The cross-validation method randomly divides a part of the training set as the validation set during the training process, and performs prediction and error calculation on the trained model in real time.

[0037] During training, using different combinations of training and validation sets for multiple training and validation runs can effectively evaluate the model's performance on different data subsets, thereby identifying and avoiding overfitting in advance. By training and validating the model on different training and validation sets, the model performance under different hyperparameter settings can be compared, allowing the selection of the relatively optimal hyperparameter combination. Furthermore, repeated cross-validation experiments provide reliable verification of the model's stability and consistency, helping to improve the model's accuracy and enhance its effectiveness in practical applications.

[0038] In a preferred embodiment of the present invention, Gaussian Process Regression (GPR) is a nonparametric regression method based on a probabilistic model, used to model the relationship between input and output. The training process of the Gaussian Process Regression specifically includes the following steps: Data preparation: Collect and prepare input features and corresponding output data; Choosing a covariance function: This measures the similarity between input features and determines the covariance function used to describe the similarity between samples. Common covariance functions include linear, polynomial, and Gaussian kernels. Different covariance functions will affect the model's fitting ability and generalization ability.

[0039] Constructing the covariance matrix: Calculate the covariance matrix between input features using a selected covariance function; the covariance matrix describes the correlation and similarity between samples. Adding noise: A small noise variance is added to the covariance matrix to capture observation errors; Model training: The hyperparameters of the model are estimated through Bayesian inference; Bayesian inference takes into account the prior distribution of the parameters and estimates the hyperparameters of the model through the posterior probability; the hyperparameters are fixed parameters in the model, rather than being learned from the training data.

[0040] Prediction: Using the trained model, make predictions on new input features.

[0041] The Bayesian optimization described in this implementation involves changing the hyperparameters of the model and recording the corresponding least squared error during the training process, while simultaneously predicting the minimum value of the error function and iteratively approaching the minimum value.

[0042] Specifically, Bayesian optimization continuously changes the model's hyperparameters and, based on the evaluation results of the objective function (least squared error), uses Bayesian inference to establish a probabilistic model between hyperparameters and error. Then, it iteratively updates the hyperparameters through sampling and selection methods, approximating the minimum of the objective function. This process can iterate multiple times until the optimal set of hyperparameters is found. By selecting appropriate hyperparameters for training and evaluation in each iteration, it avoids trying a large number of ineffective hyperparameter combinations. Compared to exhaustive search or random search, Bayesian optimization utilizes computational resources more efficiently, making the hyperparameter search process faster. The iterative process of Bayesian optimization allows the model to adaptively adjust its exploration strategy in the hyperparameter space based on existing observation data. During the search, it performs denser sampling in known optimal regions, thus increasing the likelihood of discovering the optimal hyperparameter combination. Through continuous iteration, Bayesian optimization gradually converges to the hyperparameter combination corresponding to the minimum of the objective function. Since each iteration updates the Gaussian process model, it can more accurately predict the distribution of the objective function in the hyperparameter space, providing more reliable guidance for the next sampling.

[0043] In step S3, the specific steps of the Bayesian optimization include: S31. Define the hyperparameter space: Determine the hyperparameters to be optimized and their possible range of values; for example, in Gaussian process regression, you can choose to optimize hyperparameters such as kernel function parameters and noise variance.

[0044] S32. Initialize sample data: Randomly select a set of hyperparameters as initial values; S33. Training the model: Train the model using the training data and the current hyperparameter settings, and calculate the prediction results; S34. Evaluate the objective function: Compare the model's predictions with the actual observations, and calculate the least squared error as the value of the objective function; S35. Update model state: Record the current hyperparameters and the corresponding least squares error; S36. Fitting a probabilistic model based on existing data: Based on historical hyperparameter-least squared error pairs, establish a probabilistic model to describe the relationship between hyperparameters and errors; a common method is to use Gaussian processes for modeling.

[0045] S37. Parallel Sampling: Obtaining new candidate hyperparameter settings from the conditional probability distribution through sampling methods (such as Gaussian process sampling or Monte Carlo sampling); S38. Select the optimal value of the objective function for the next iteration: Use the new hyperparameter settings and repeat steps S33 to S37 until the preset number of iterations is reached or the termination criterion is met (e.g., least squares error convergence or exhaustion of the hyperparameter search space).

[0046] S39. Return the best hyperparameter: Based on the historical hyperparameter-least squared error pairs, select the hyperparameter with the smallest error as the best hyperparameter.

[0047] In this invention, the random verification is based on the existing Vicat softening temperature testing standard. The results are recorded as actual values ​​and compared with the predicted values ​​obtained by the model. The error should be less than or equal to the overall variance of the model.

[0048] By comparing actual values ​​with model predictions, we can evaluate the model's performance on new data. This helps determine the model's generalization ability and prediction accuracy. Comparing the error between actual and predicted values ​​can reveal biases in the model. If the error consistently deviates from zero error (i.e., perfect prediction) or tilts in a certain direction, it means that the model may have problems such as underfitting or overfitting.

[0049] In step 5, retraining and optimizing the model involves adding data with large prediction errors to the training set and repeating the model training and optimization steps in step S3.

[0050] When a model exhibits significant prediction errors on certain datasets, it indicates potential biases or errors. Retraining by adding these problematic samples to the training set allows for targeted correction of these biases or errors, thereby improving model performance. Further optimization of the model's parameters can help better adjust weights and biases, leading to improved performance across the entire training set. Data with large prediction errors often represent special or anomalous samples; focusing training on these samples can enhance the model's ability to handle similar data.

[0051] To achieve optimal signal transmission and reception, the longitudinal wave probe is placed perpendicular to the protective tube, and the returned echo signal is amplified with a high signal-to-noise ratio to improve flaw detection accuracy.

[0052] like Figures 2-4 As shown, MPP cable protection pipes were selected as the testing object, and the specific steps are as follows: Step 1: Use an ultrasonic longitudinal wave probe to collect raw data from the cable protection tube. The samples were randomly selected from 5 different batches of MPP protection tubes with known Vicat softening temperatures. Five samples were selected from each batch of protection tubes. During sampling, samples were collected at equidistant intervals along the circumference of the tube at 90° intervals, resulting in a total of 100 sets of raw ultrasonic A-scan data.

[0053] Step 2: Analyze and process the obtained ultrasonic A-scan data, extract ultrasonic velocity, ultrasonic attenuation, and frequency-attenuation variation characteristics, use the ultrasonic parameters as feature values, and use the Vicat softening temperature of the corresponding cable protection pipe as the output value to create a training set.

[0054] Step 3: Perform Gaussian process regression on the created training set. Based on the number of training set samples, use 5-fold cross-validation for training. Iteratively optimize the model using a Bayesian optimizer. Figure 2 As shown, the model achieves the least squared error in the 93rd iteration.

[0055] Step 4: Randomly select 5 more samples from each of the 5 batches of cable protection pipes. Collect data at equal intervals along the circumference of each sample to obtain 100 sets of ultrasonic A-scan signals. Extract ultrasonic velocity, ultrasonic attenuation and frequency-attenuation change characteristics to create a test set. Input the test set into the Vicat softening temperature prediction model.

[0056] Step 5: Compare the model prediction results with the Vicat softening temperature standard test results. The overall prediction variance of the model is consistent with the cross-validation results, indicating that the model has good reliability and there is no overfitting problem.

[0057] This invention also provides an ultrasonic rapid detection system for the Vicat softening temperature of cable protection pipes, comprising: The data acquisition module uses an ultrasonic longitudinal wave probe to acquire raw data from the cable protection pipe, obtaining ultrasonic A-scan data along the pipe thickness direction; The data processing and feature extraction module analyzes and processes the acquired ultrasonic A-scan data, extracts key parameters as feature values, and uses the ultrasonic parameters and the Vicat softening temperature of the cable protection pipe as output values ​​to establish a dataset. The model optimization module, based on the dataset obtained in S2, adopts a Gaussian process regression model and performs Bayesian optimization using the least squares error as the evaluation criterion to obtain the optimized Vicat softening temperature prediction model. The temperature prediction module is used to collect ultrasonic A-scan data of the cable protection pipe under inspection, extract the corresponding parameters, and input these parameters into the optimized Vicat softening temperature prediction model in S3 to obtain the ultrasonic characterization value of the Vicat softening temperature. The results verification module verifies the model's prediction results using existing random verification methods. If the model's prediction results differ significantly from the actual values, these detection data can be added to the training set for model retraining and optimization.

[0058] The present invention also provides a computer device including a memory and a processor, wherein the memory stores steps that can be loaded by the processor and executed by the ultrasonic rapid detection method.

[0059] The present invention also provides a computer-readable storage medium storing a processor-executable program, which, when executed by a processor, is used to perform the ultrasonic rapid detection method.

[0060] Those skilled in the art should understand that this invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to this invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A rapid ultrasonic detection method for the Vicat softening temperature of cable protection pipes, characterized in that, Includes the following steps: S1: Use an ultrasonic longitudinal wave probe to collect raw data from the cable protection pipe. The raw data obtained is ultrasonic A-scan data along the pipe thickness direction. S2: Analyze and process the ultrasonic A-scan data obtained in S1, extract key parameters, use the ultrasonic parameters as feature values, and use the Vicat softening temperature of the corresponding cable protection pipe as the output value to establish a dataset. S3: Perform Gaussian process regression on the dataset obtained in S2, and use the least squares error as the evaluation value to perform Bayesian optimization on the model to obtain the optimized Vicat softening temperature prediction model. S4: In actual testing, ultrasonic A-scan data of the cable protection pipe under test is collected, the corresponding parameters are extracted, and the Vicat softening temperature prediction model in S3 is input to obtain the ultrasonic characterization value of the Vicat softening temperature. S5: Perform random verification on the results. If the error between the model's prediction and the actual value exceeds the overall variance of the model, add the detection data to the training set to retrain and optimize the model. The extracted key parameters include ultrasonic velocity, ultrasonic attenuation, and frequency-attenuation variation characteristics calculated based on the ultrasonic spectrum. The frequency-attenuation variation characteristic is obtained by the following steps: (1) The ultrasound A-scan signal is intercepted, and the primary and secondary reflected echoes of the inner wall of the tube are extracted; (2) Perform discrete Fourier transform on the first and second reflected echo regions of the inner wall of the pipe to obtain the corresponding regional spectrum; (3) Based on the amplitude of each frequency component in the regional spectrum, calculate the attenuation of ultrasonic components at different frequencies and obtain the ultrasonic attenuation characteristics that vary with frequency.

2. The ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes according to claim 1, characterized in that, The raw data should be collected in a distributed manner along both the axial and circumferential directions of the protective tube.

3. The ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes according to claim 1, characterized in that, The horizontal axis of the ultrasonic A-scan data represents the signal propagation time, and the vertical axis represents the acoustic signal intensity. The data should include the primary and secondary reflected echoes from the inner wall of the cable protection pipe.

4. The ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes according to claim 1, characterized in that, In step S3, cross-validation is used during the training of Gaussian process regression.

5. The ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes according to claim 4, characterized in that, The cross-validation method involves randomly dividing a portion of the training set as a validation set during the training process, and performing real-time prediction and error calculation on the trained model.

6. The ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes according to claim 1, characterized in that, The training process of the Gaussian process regression specifically includes the following steps: Data preparation: Collect and prepare input features and corresponding output data; Choosing a covariance function: This measures the similarity between input features and determines the covariance function used to describe the similarity between samples. Constructing the covariance matrix: Calculate the covariance matrix between input features using the selected covariance function; Adding noise: A small noise variance is added to the covariance matrix to capture observation errors; Model training: Estimate the model's hyperparameters through Bayesian inference; Prediction: Using the trained model, make predictions on new input features.

7. The ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes according to claim 1, characterized in that, The Bayesian optimization involves changing the model's hyperparameters and recording the corresponding least squared error during training, while simultaneously predicting the minimum value of the error function and iteratively approaching that minimum value.

8. The ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes according to claim 1, characterized in that, In step S3, the specific steps of the Bayesian optimization include: S31. Define the hyperparameter space: Determine the hyperparameters to be optimized and their possible value ranges; S32. Initialize sample data: Randomly select a set of hyperparameters as initial values; S33. Training the model: Train the model using the training data and the current hyperparameter settings, and calculate the prediction results; S34. Evaluate the objective function: Compare the model's predictions with the actual observations, and calculate the least squared error as the value of the objective function; S35. Update model state: Record the current hyperparameters and the corresponding least squares error; S36. Fit a probabilistic model based on existing data: Based on historical hyperparameter-least squared error pairs, establish a probabilistic model to describe the relationship between hyperparameters and errors. S37. Parallel Sampling: Obtaining new candidate hyperparameter settings from the conditional probability distribution through sampling methods; S38. Select the optimal value of the objective function for the next iteration: Using the new hyperparameter settings, repeat steps S33 to S37 until the preset number of iterations is reached or the termination criterion is met. S39. Return the best hyperparameter: Based on the historical hyperparameter-least squared error pairs, select the hyperparameter with the smallest error as the best hyperparameter.

9. The ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes according to claim 1, characterized in that, The longitudinal wave probe is placed perpendicular to the protective tube, and the returned echo signal is amplified with a high signal-to-noise ratio.

10. An ultrasonic rapid detection system for the Vicat softening temperature of cable protection pipes, used in the ultrasonic rapid detection method for the Vicat softening temperature of cable protection pipes as described in claim 1, characterized in that, include: The data acquisition module uses an ultrasonic longitudinal wave probe to acquire raw data from the cable protection pipe, obtaining ultrasonic A-scan data along the pipe thickness direction; The data processing and feature extraction module analyzes and processes the acquired ultrasonic A-scan data, extracts key parameters as feature values, and uses the ultrasonic parameters and the Vicat softening temperature of the cable protection pipe as output values ​​to establish a dataset. The model optimization module, based on the obtained dataset, adopts a Gaussian process regression model and performs Bayesian optimization using the least squares error as the evaluation criterion to obtain the optimized Vicat softening temperature prediction model. The temperature prediction module is used to collect ultrasonic A-scan data of the cable protection pipe under inspection, extract the corresponding parameters, and input these parameters into the optimized Vicat softening temperature prediction model to obtain the ultrasonic characterization value of the Vicat softening temperature. The results verification module verifies the model's prediction results. If the error between the model's prediction results and the actual values ​​exceeds the overall variance of the model, the detection data is added to the training set for model retraining and optimization.

11. A computer device, characterized in that, It includes a memory and a processor, wherein the memory stores steps that can be loaded by the processor and executed as described in any one of claims 1-9 of the rapid ultrasonic detection method.

12. A computer-readable storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to perform the steps of the rapid ultrasonic detection method as described in any one of claims 1-9.

Images