Intelligent identification system and method for internal defects of plywood based on X-ray

Abstract

The invention belongs to the technical field of X-rays, and provides a plywood internal defect intelligent identification system and method based on X-rays, and the method comprises the steps: obtaining a defect positioning deviation value through a data collection and speed calculation module, and reversely deducing a relative movement speed; a relative movement speed-defect positioning deviation value sensing model is constructed through correlation analysis; combining a U-Net pixel level segmentation sub-module to extract an image defect positioning value, and generating a pre-calibration factor; after an accurate calibration factor is obtained through iterative verification, the relative movement speed is dynamically regulated and controlled to eliminate positioning deviation; according to the method, dual calibration of theoretical prediction and image actual measurement is realized, positioning distortion caused by motion blur is effectively counteracted, the defect positioning precision is improved, and the method is adaptive to industrial continuous detection requirements and has good practicability and expandability.

Description

Technical Field

[0001] This invention belongs to the field of X-ray technology, specifically an intelligent identification system and method for internal defects in plywood based on X-rays. Background Technology

[0002] In the field of X-ray-based intelligent identification of internal defects in plywood, plywood needs to maintain relative movement with the X-ray inspection system in online inspection scenarios to achieve continuous inspection. However, the control of this relative movement speed directly affects the inspection accuracy. When the relative movement speed is too fast, the plywood will inevitably be displaced during X-ray exposure. This displacement will directly cause motion blur and contour stretching in the X-ray image along the direction of motion.

[0003] The aforementioned imaging defects severely damage the edge and positional features of internal defects in plywood, causing defect feature distortion. This directly leads to insufficient defect positioning accuracy, causing the coordinate deviation between the actual defect location and the identification result to exceed the allowable range for industrial inspection. Simultaneously, motion blur and contour stretching significantly increase the quantification error of defect size, making it impossible to accurately determine the actual size of the defect. Ultimately, this renders the inspection results ineffective in guiding plywood quality grading and sorting, thus limiting the reliable application of X-ray inspection technology in high-speed online quality inspection scenarios for plywood.

[0004] To this end, the present invention provides an intelligent identification system and method for internal defects in plywood based on X-rays. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0006] The technical solution adopted by this invention to solve its technical problem is: On the one hand, the present invention provides an intelligent identification method for internal defects in plywood based on X-rays, including: Step S10: Obtain the internal defect positioning deviation value of the plywood, and calculate the relative moving speed between the X-ray and the plywood based on the internal defect positioning deviation value of the plywood. Step S20: Analyze the internal defect positioning deviation value and relative movement speed of the plywood, and establish a relative movement speed-defect positioning deviation value perception model based on the analysis results; Step S30: Based on the relative movement speed-defect location deviation value perception model, identify the defect location deviation value corresponding to the current relative movement speed, obtain the X-ray image corresponding to the current defect location deviation value, perform defect segmentation processing on the X-ray image, obtain the preliminary defect location value, and then generate a pre-calibration factor based on the preliminary defect location value and the current defect location deviation value. Step S40: Verify the pre-calibration factor to obtain the calibration factor, and control the current relative movement speed according to the calibration factor to eliminate the defect positioning deviation value.

[0007] As one embodiment, the specific calculation process for the relative moving speed is as follows: Obtain the known internal defect positioning deviation value of the plywood, and calculate the internal defect positioning deviation value of the plywood based on the known internal defect positioning deviation value of the plywood: Internal defect positioning deviation value of plywood = actual position coordinates of defect - identification position coordinates; The specific process for calculating the relative speed between the X-ray and the plywood based on the positioning deviation value of the internal defects of the plywood is as follows: Relative moving speed = Plywood internal defect positioning deviation value ÷ X-ray exposure time.

[0008] As one embodiment, the specific process for analyzing the positioning deviation value and relative moving speed of internal defects in plywood is as follows: The relative moving speeds are sorted in ascending order to obtain a relative moving speed sequence. Based on the relative moving speed sequence, the plywood internal defect positioning deviation value corresponding to each relative moving speed is matched one-to-one to obtain a plywood internal defect positioning deviation value sequence corresponding to the relative moving speed sequence. The relative moving speed sequence is denoted as the independent variable sequence, and the plywood internal defect positioning deviation value sequence is denoted as the dependent variable sequence. The Pearson correlation coefficient method is used to calculate the Pearson correlation coefficient between the independent variable sequence and the dependent variable sequence. If the calculated Pearson correlation coefficient is greater than or equal to the Pearson correlation coefficient threshold, it is determined that the relative moving speed and the positioning deviation value of the internal defect of the plywood have a linear correlation. If the calculated Pearson correlation coefficient is less than the Pearson correlation coefficient threshold, it is determined that the relative moving speed and the deviation value of the internal defect positioning of the plywood are not linearly related.

[0009] As one embodiment, the specific process of establishing the relative movement speed-defect location deviation value perception model is as follows: If it is determined that the relative moving speed and the internal defect positioning deviation of the plywood are linearly correlated, then the gated recurrent unit (GRU) time series model is adopted, with the independent variable sequence as the input feature and the dependent variable sequence as the output label, to perform time series feature learning and training fitting, and construct a relative moving speed-defect positioning deviation value perception model. If it is determined that the relative moving speed and the deviation value of the internal defect location of the plywood are not linearly related, then a nonlinear fitting method is used to perform nonlinear fitting on the sequence of independent variables and the sequence of dependent variables to establish a relative moving speed-defect location deviation value perception model.

[0010] As one embodiment, the specific process of identifying the defect location deviation value corresponding to the current relative moving speed is as follows: The current real-time relative movement speed is input into the relative movement speed-defect location deviation value perception model to obtain the corresponding plywood internal defect location deviation value under the current real-time relative movement speed.

[0011] As one embodiment, the specific process for obtaining the preliminary defect location value is as follows: Retrieve the plywood X-ray inspection image corresponding to the current defect location deviation value; process the X-ray inspection image using the U-Net segmentation model, and output the pixel-level position coordinates of the internal defects of the plywood in the X-ray inspection image, and use the pixel-level position coordinates as the preliminary defect location value.

[0012] As one embodiment, the specific process of generating the pre-calibration factor is as follows: The pre-calibration factor is obtained by comparing the deviation value of the plywood internal defect location corresponding to the current relative moving speed with the preliminary defect location value obtained after processing by the U-Net segmentation model. The ratio calculation process is as follows: Pre-calibration factor = Current internal defect location deviation value of plywood ÷ Preliminary defect location value.

[0013] As one embodiment, the specific process of verifying the pre-calibration factor to obtain the calibration factor is as follows: The plywood X-ray image corresponding to the current defect location deviation value is selected as the verification sample. The X-ray image of the verification sample is processed again using the U-Net segmentation model to extract the actual defect location value of the verification sample. The actual defect location deviation value of the verification sample is calculated as follows: Actual defect location deviation value = True defect location coordinates - Actual defect location value. The actual defect location deviation value is then multiplied by the pre-calibration factor to obtain the verification calibration value. The specific calculation formula is as follows: Verification calibration value = actual defect location deviation value × pre-calibration factor; The difference between the verification calibration value and the preliminary defect location value output by the U-Net segmentation model is verified using the following formula: Difference verification result = Preliminary defect location value - verification calibration value ; If the difference verification result is within the preset industrial allowable error range, it is determined that the pre-calibration factor can accurately compensate for the positioning deviation, and the pre-calibration factor is directly determined as the final calibration factor. If the difference verification result exceeds the preset industrial allowable error range, a correction factor is calculated. The correction factor = difference verification result ÷ actual defect location deviation value. The correction factor is used to make the first correction to the current pre-calibration factor. The corrected pre-calibration factor = current pre-calibration factor × (1 − correction factor). The first corrected pre-calibration factor is substituted into the difference verification until the difference verification result falls within the industrial allowable error range. The pre-calibration factor at this time is determined as the final calibration factor.

[0014] As one embodiment, the specific process for eliminating the defect location deviation value is as follows: Based on the calibration factor and combined with the exposure time of the X-ray, the current real-time relative movement speed between the X-ray detection system and the plywood is dynamically adapted and controlled. The control formula is: Optimal relative movement speed = Current relative movement speed ÷ Calibration factor.

[0015] On the other hand, the present invention provides an X-ray-based intelligent identification system for internal defects in plywood, comprising: Data acquisition and speed calculation module: acquires the positioning deviation value of internal defects in plywood, and calculates the relative moving speed between the X-ray and the plywood based on the positioning deviation value of internal defects in plywood. Association Analysis and Model Building Module: Analyzes the positioning deviation value and relative movement speed of internal defects in plywood, and establishes a relative movement speed-defect positioning deviation value perception model based on the analysis results; Precalibration factor generation module: Based on the relative movement speed-defect location deviation value perception model, identify the defect location deviation value corresponding to the current relative movement speed, obtain the X-ray image corresponding to the current defect location deviation value, perform defect segmentation processing on the X-ray image to obtain a preliminary defect location value, and then generate a precalibration factor based on the preliminary defect location value and the current defect location deviation value. Calibration factor verification module: Verifies the pre-calibration factor to obtain the calibration factor, and controls the current relative movement speed according to the calibration factor to eliminate defect positioning deviation value.

[0016] The beneficial effects of this invention are as follows: 1. By constructing a relative movement speed-defect location deviation perception model, accurate prediction of defect location deviation at the current speed is achieved. This specifically solves the problems of motion blur and contour stretching caused by excessive relative movement speed in X-ray detection, thereby improving defect location accuracy from the root. The U-Net pixel-level segmentation submodule is embedded to accurately extract the spatial location features of the defect. A pre-calibration factor is generated by calculating the ratio of the predicted deviation value to the preliminary location value. Combined with an iterative verification and correction mechanism, the reliability of the calibration factor is ensured, achieving dual accurate calibration through theoretical prediction and image measurement.

[0017] 2. Based on the final calibration factor, the relative moving speed is dynamically adjusted to accurately match the plywood conveying speed with the X-ray exposure time. This eliminates positioning deviations and ensures the efficiency of continuous online inspection, making it suitable for high-speed industrial quality inspection scenarios. The modular design ensures clear responsibilities for each module and a closed-loop data flow. The U-Net sub-modules can flexibly adapt to the segmentation requirements of different defect types. Further optimization of model parameters can further improve the inspection adaptability under different working conditions (such as plywood of different thicknesses and tree species). Attached Figure Description

[0018] The invention will now be further described with reference to the accompanying drawings.

[0019] Figure 1 This is a flowchart of the steps of the intelligent identification method for internal defects of plywood based on X-rays according to the present invention; Figure 2 This is a system module diagram of the X-ray-based intelligent identification system for internal defects in plywood according to the present invention. Detailed Implementation

[0020] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0021] Example 1 like Figure 1 As shown in the embodiment of the present invention, the intelligent identification method for internal defects of plywood based on X-rays includes: Step S10: Obtain the internal defect positioning deviation value of the plywood, and calculate the relative moving speed between the X-ray and the plywood based on the internal defect positioning deviation value of the plywood. In step S10, the specific process for obtaining the positioning deviation value of internal defects in plywood is as follows: In the historical plywood internal defect identification database, known plywood internal defect location deviation values ​​are obtained. Based on these known plywood internal defect location deviation values, the plywood internal defect location deviation value is calculated as follows: Plywood internal defect location deviation value = actual defect location coordinates - identification location coordinates; In step S10, the specific process of calculating the relative movement speed between the X-ray and the plywood based on the internal defect positioning deviation value of the plywood is as follows: The specific process for calculating the relative speed between the X-ray and the plywood based on the calculated deviation value of the internal defect location of the plywood is as follows: Relative moving speed = Plywood internal defect positioning deviation value ÷ X-ray exposure time; It should be noted that the historical plywood internal defect identification database is a database established during the historical plywood internal defect identification process. It records the plywood internal defect positioning deviation value for each instance, as well as the X-ray exposure time corresponding to the plywood internal defect positioning deviation value. Step S20: Analyze the internal defect positioning deviation value and relative movement speed of the plywood, and establish a relative movement speed-defect positioning deviation value perception model based on the analysis results; In step S20, the specific process of analyzing the positioning deviation value and relative movement speed of the internal defects in the plywood is as follows: Based on the relative movement speed, the relative movement speeds are sorted in ascending order to obtain a relative movement speed sequence; according to the relative movement speed sequence, a one-to-one matching process is performed on the plywood internal defect positioning deviation value corresponding to each relative movement speed to obtain a plywood internal defect positioning deviation value sequence corresponding to the relative movement speed sequence, wherein the plywood internal defect positioning deviation value is the difference between the actual position coordinates and the identified position coordinates of the defect; The relative moving speed sequence is denoted as the independent variable sequence, and the plywood internal defect positioning deviation value sequence is denoted as the dependent variable sequence. The Pearson correlation coefficient method is used to calculate the Pearson correlation coefficient between the independent variable sequence and the dependent variable sequence, and the calculated Pearson correlation coefficient is compared with the preset Pearson correlation coefficient threshold. If the calculated Pearson correlation coefficient is greater than or equal to the Pearson correlation coefficient threshold, it is determined that there is a linear correlation between the independent variable sequence and the dependent variable sequence, that is, the relative moving speed and the plywood internal defect positioning deviation value are linearly correlated. If the calculated Pearson correlation coefficient is less than the Pearson correlation coefficient threshold, it is determined that there is no linear relationship between the independent variable sequence and the dependent variable sequence, that is, the relative moving speed and the plywood internal defect positioning deviation value are not linearly related. In step S20, the specific process of establishing the relative movement speed-defect location deviation value perception model based on the analysis results is as follows: If it is determined that there is a linear correlation between the relative moving speed and the plywood internal defect location deviation value, a gated recurrent unit (GRU) time series model is adopted. The independent variable sequence (relative moving speed sequence) is used as the input feature and the dependent variable sequence (plywood internal defect location deviation value sequence) is used as the output label. The time series features are learned and trained for fitting. During the training process, the model parameters are optimized by minimizing the mean square error (MSE) between the predicted deviation value and the actual deviation value. After the model converges, a relative moving speed-defect location deviation value perception model that can accurately map the correlation between relative speed deviation and recognition accuracy deviation is constructed. If it is determined that the relative moving speed and the plywood internal defect positioning deviation value are not linearly related, a nonlinear fitting method is used to perform nonlinear fitting on the independent variable sequence and the dependent variable sequence. The nonlinear fitting method includes, but is not limited to: polynomial fitting, exponential function fitting, Gaussian function fitting, kernel function fitting and neural network fitting. The goodness of fit is calculated for the fitting results of each nonlinear fitting method, and the fitting result with the highest goodness of fit is selected (e.g., the coefficient of determination is used as the judgment index, the closer the coefficient of determination is to 1, the higher the goodness of fit), as the final relative moving speed-defect positioning deviation value perception model. Step S30: Based on the relative movement speed-defect location deviation value perception model, identify the defect location deviation value corresponding to the current relative movement speed, obtain the X-ray image corresponding to the current defect location deviation value, perform defect segmentation processing on the X-ray image, obtain the preliminary defect location value, and then generate a pre-calibration factor based on the preliminary defect location value and the current defect location deviation value. In step S30, the specific process of identifying the defect location deviation value corresponding to the current relative movement speed according to the relative movement speed-defect location deviation value perception model is as follows: The current real-time relative movement speed between the X-ray inspection system and the plywood is obtained. This real-time relative movement speed is then input into the relative movement speed-defect location deviation value perception model. Through feature mapping and output prediction of the relative movement speed-defect location deviation value perception model, the internal defect location deviation value of the plywood corresponding to the current real-time relative movement speed is obtained. The internal defect location deviation value of the plywood is the difference between the actual position coordinates of the internal defect and the defect identification position coordinates, representing the defect location offset caused by the relative movement speed.

[0022] In step S30, the X-ray image corresponding to the current defect location deviation value is obtained, and the X-ray image is processed to obtain the preliminary defect location value. The specific process is as follows: Based on the relative movement speed-defect location deviation value perception model outputting the corresponding internal defect location deviation value of the plywood at the current relative movement speed, the X-ray inspection image of the plywood corresponding to the current defect location deviation value is retrieved; the U-Net segmentation model is used to perform pixel-level defect segmentation processing on the X-ray inspection image. The U-Net segmentation model performs downsampling processing on the X-ray inspection image through the encoding layer to extract the deep feature information of the internal defects of the plywood in the image, and then performs upsampling processing through the decoding layer to restore the spatial position features of the defects. The feature information of the encoding layer and the decoding layer is fused through a skip connection structure to effectively preserve the edge detail features of the defects in the X-ray inspection image, accurately remove motion blur and contour stretching caused by excessive relative movement speed, and finally output the pixel-level position coordinates of the internal defects of the plywood in the X-ray inspection image. These pixel-level position coordinates are used as the preliminary defect location value. In step S30, the specific process of generating the pre-calibration factor based on the preliminary defect location value and the current defect location deviation value is as follows: The pre-calibration factor is obtained by comparing the plywood internal defect location deviation value corresponding to the current relative movement speed output by the relative movement speed-defect location deviation value perception model with the preliminary defect location value obtained after processing by the U-Net segmentation model. The ratio calculation process is as follows: Pre-calibration factor = Current internal defect location deviation value of plywood ÷ Preliminary defect location value; Step S40: Verify the pre-calibration factor to obtain the calibration factor, and control the current relative movement speed according to the calibration factor to eliminate the defect positioning deviation value; In step S40, the specific process of verifying the pre-calibration factor to obtain the calibration factor is as follows: The plywood X-ray image corresponding to the current defect location deviation value is selected as the verification sample. The U-Net segmentation model is used to perform pixel-level defect segmentation on the X-ray image of the verification sample again, and the actual defect location value of the verification sample is extracted. The measured location value of the defect in the verification sample is compared with the actual location coordinates of the defect inside the plywood to obtain the actual defect location deviation value of the verification sample (actual defect location deviation value = actual defect location coordinates − measured defect location value). The actual defect location deviation value is then multiplied by the pre-calibration factor. The core logic of this multiplication is to scale the actual defect location deviation value using the pre-calibration factor to obtain a verification calibration value with the same dimensions as the initial defect location value. The specific calculation formula is as follows: Verification calibration value = actual defect location deviation value × pre-calibration factor; The verification calibration value is then compared with the preliminary defect location value output by the U-Net segmentation model using the following formula: Difference verification result = Preliminary defect location value - verification calibration value ; If the difference verification result is within the preset industrial allowable error range (e.g., recommended industrial allowable error: defect location deviation ≤ ±0.3mm), then it is determined that the pre-calibration factor can accurately compensate for the location deviation, and the pre-calibration factor is directly determined as the final calibration factor. If the difference verification result exceeds the preset industrial allowable error range, an iterative correction process is initiated. The specific process is as follows: calculate the correction coefficient based on the difference verification result (correction coefficient = difference verification result ÷ actual defect location deviation value), and use the correction coefficient to perform the first correction on the current pre-calibration factor (corrected pre-calibration factor = current pre-calibration factor × (1 − correction coefficient)); substitute the pre-calibration factor after the first correction back into the above verification process, and use the U-Net segmentation model to re-segment the X-ray image of the verification sample to obtain a new actual defect location value, until the difference verification result falls within the industrial allowable error range, and then determine the pre-calibration factor at this time as the final calibration factor; In step S40, the specific process of controlling the current relative movement speed according to the calibration factor to eliminate the defect positioning deviation value is as follows: Based on the calibration factor, and combined with the exposure time of X-rays and the transport characteristics of plywood, the current real-time relative moving speed between the X-ray detection system and the plywood is dynamically adapted and controlled. The control formula is: Optimal relative moving speed = Current relative moving speed ÷ Calibration factor. By adjusting the conveying speed of the plywood to the optimal relative moving speed, the movement displacement of the plywood within the X-ray exposure time is precisely matched with the calibration factor, thereby fundamentally offsetting the imaging ghosting and pixel offset problems caused by the relative moving speed deviation, and ultimately eliminating the internal defect positioning deviation value of the plywood, achieving precise compensation and correction of defect positioning accuracy. Example 2 like Figure 2 As shown, according to the specific implementation process of Embodiment 1, the present invention provides an X-ray-based intelligent identification system for internal defects in plywood, comprising: Data acquisition and speed calculation module: acquires the positioning deviation value of internal defects in plywood, and calculates the relative moving speed between the X-ray and the plywood based on the positioning deviation value of internal defects in plywood. Association Analysis and Model Building Module: Analyzes the positioning deviation value and relative movement speed of internal defects in plywood, and establishes a relative movement speed-defect positioning deviation value perception model based on the analysis results; Precalibration factor generation module: Based on the relative movement speed-defect location deviation value perception model, identify the defect location deviation value corresponding to the current relative movement speed, obtain the X-ray image corresponding to the current defect location deviation value, perform defect segmentation processing on the X-ray image to obtain a preliminary defect location value, and then generate a precalibration factor based on the preliminary defect location value and the current defect location deviation value. Calibration factor verification module: Verifies the pre-calibration factor to obtain the calibration factor, and controls the current relative movement speed according to the calibration factor to eliminate defect positioning deviation value.

[0023] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present 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 the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent identification method for internal defects in plywood based on X-rays, characterized in that: include: Step S10: Obtain the internal defect positioning deviation value of the plywood, and calculate the relative moving speed between the X-ray and the plywood based on the internal defect positioning deviation value of the plywood. Step S20: Analyze the internal defect positioning deviation value and relative movement speed of the plywood, and establish a relative movement speed-defect positioning deviation value perception model based on the analysis results; Step S30: Based on the relative movement speed-defect location deviation value perception model, identify the defect location deviation value corresponding to the current relative movement speed, obtain the X-ray image corresponding to the current defect location deviation value, perform defect segmentation processing on the X-ray image, obtain the preliminary defect location value, and then generate a pre-calibration factor based on the preliminary defect location value and the current defect location deviation value. Step S40: Verify the pre-calibration factor to obtain the calibration factor, and control the current relative movement speed according to the calibration factor to eliminate the defect positioning deviation value.

2. The intelligent identification method for internal defects of plywood based on X-rays according to claim 1, characterized in that: The specific calculation process for the relative speed is as follows: Obtain the known internal defect positioning deviation value of the plywood, and calculate the internal defect positioning deviation value of the plywood based on the known internal defect positioning deviation value of the plywood: Internal defect positioning deviation value of plywood = actual position coordinates of defect - identification position coordinates; The specific process for calculating the relative speed between the X-ray and the plywood based on the positioning deviation value of the internal defects of the plywood is as follows: Relative moving speed = Plywood internal defect positioning deviation value ÷ X-ray exposure time.

3. The intelligent identification method for internal defects of plywood based on X-rays according to claim 1, characterized in that: The specific process for analyzing the positioning deviation value and relative movement speed of internal defects in plywood is as follows: The relative moving speeds are sorted in ascending order to obtain a relative moving speed sequence. Based on the relative moving speed sequence, the plywood internal defect positioning deviation value corresponding to each relative moving speed is matched one-to-one to obtain a plywood internal defect positioning deviation value sequence corresponding to the relative moving speed sequence. The relative moving speed sequence is denoted as the independent variable sequence, and the plywood internal defect positioning deviation value sequence is denoted as the dependent variable sequence. The Pearson correlation coefficient method is used to calculate the Pearson correlation coefficient between the independent variable sequence and the dependent variable sequence. If the calculated Pearson correlation coefficient is greater than or equal to the Pearson correlation coefficient threshold, it is determined that the relative moving speed and the positioning deviation value of the internal defect of the plywood have a linear correlation. If the calculated Pearson correlation coefficient is less than the Pearson correlation coefficient threshold, it is determined that the relative moving speed and the deviation value of the internal defect positioning of the plywood are not linearly related.

4. The intelligent identification method for internal defects of plywood based on X-rays according to claim 1, characterized in that: The specific process for establishing the relative movement speed-defect location deviation value perception model is as follows: If it is determined that the relative moving speed and the internal defect positioning deviation of the plywood are linearly correlated, then the gated recurrent unit (GRU) time series model is adopted, with the independent variable sequence as the input feature and the dependent variable sequence as the output label, to perform time series feature learning and training fitting, and construct a relative moving speed-defect positioning deviation value perception model. If it is determined that the relative moving speed and the deviation value of the internal defect location of the plywood are not linearly related, then a nonlinear fitting method is used to perform nonlinear fitting on the sequence of independent variables and the sequence of dependent variables to establish a relative moving speed-defect location deviation value perception model.

5. The intelligent identification method for internal defects of plywood based on X-rays according to claim 1, characterized in that: The specific process for identifying the defect location deviation value corresponding to the current relative moving speed is as follows: The current real-time relative movement speed is input into the relative movement speed-defect location deviation value perception model to obtain the corresponding plywood internal defect location deviation value under the current real-time relative movement speed.

6. The intelligent identification method for internal defects of plywood based on X-rays according to claim 1, characterized in that: The specific process for obtaining the preliminary defect location value is as follows: Retrieve the plywood X-ray inspection image corresponding to the current defect location deviation value; process the X-ray inspection image using the U-Net segmentation model, and output the pixel-level position coordinates of the internal defects of the plywood in the X-ray inspection image, and use the pixel-level position coordinates as the preliminary defect location value.

7. The intelligent identification method for internal defects of plywood based on X-rays according to claim 1, characterized in that: The specific process for generating the pre-calibration factor is as follows: The pre-calibration factor is obtained by comparing the deviation value of the plywood internal defect location corresponding to the current relative moving speed with the preliminary defect location value obtained after processing by the U-Net segmentation model. The ratio calculation process is as follows: Pre-calibration factor = Current internal defect location deviation value of plywood ÷ Preliminary defect location value.

8. The intelligent identification method for internal defects of plywood based on X-rays according to claim 1, characterized in that: The specific process for verifying the pre-calibration factor to obtain the calibration factor is as follows: The plywood X-ray image corresponding to the current defect location deviation value is selected as the verification sample. The X-ray image of the verification sample is processed again using the U-Net segmentation model to extract the actual defect location value of the verification sample. The actual defect location deviation value of the verification sample is calculated as follows: Actual defect location deviation value = True defect location coordinates - Actual defect location value. The actual defect location deviation value is then multiplied by the pre-calibration factor to obtain the verification calibration value. The specific calculation formula is as follows: Verification calibration value = actual defect location deviation value × pre-calibration factor; The difference between the verification calibration value and the preliminary defect location value output by the U-Net segmentation model is verified using the following formula: Difference verification result = Preliminary defect location value - verification calibration value ; If the difference verification result is within the preset industrial allowable error range, it is determined that the pre-calibration factor can accurately compensate for the positioning deviation, and the pre-calibration factor is directly determined as the final calibration factor. If the difference verification result exceeds the preset industrial allowable error range, a correction factor is calculated. The correction factor = difference verification result ÷ actual defect location deviation value. The correction factor is used to make the first correction to the current pre-calibration factor. The corrected pre-calibration factor = current pre-calibration factor × (1 − correction factor). The first corrected pre-calibration factor is substituted into the difference verification until the difference verification result falls within the industrial allowable error range. The pre-calibration factor at this time is determined as the final calibration factor.

9. The intelligent identification method for internal defects of plywood based on X-rays according to claim 1, characterized in that: The specific process for eliminating the defect location deviation value is as follows: Based on the calibration factor and combined with the exposure time of the X-ray, the current real-time relative movement speed between the X-ray detection system and the plywood is dynamically adapted and controlled. The control formula is: Optimal relative movement speed = Current relative movement speed ÷ Calibration factor.

10. An X-ray-based intelligent identification system for internal defects in plywood, used to perform any one of the methods described in claims 1-9, characterized in that: include: Data acquisition and speed calculation module: acquires the positioning deviation value of internal defects in plywood, and calculates the relative moving speed between the X-ray and the plywood based on the positioning deviation value of internal defects in plywood. Association Analysis and Model Building Module: Analyzes the positioning deviation value and relative movement speed of internal defects in plywood, and establishes a relative movement speed-defect positioning deviation value perception model based on the analysis results; Precalibration factor generation module: Based on the relative movement speed-defect location deviation value perception model, identify the defect location deviation value corresponding to the current relative movement speed, obtain the X-ray image corresponding to the current defect location deviation value, perform defect segmentation processing on the X-ray image to obtain a preliminary defect location value, and then generate a precalibration factor based on the preliminary defect location value and the current defect location deviation value. Calibration factor verification module: Verifies the pre-calibration factor to obtain the calibration factor, and controls the current relative movement speed according to the calibration factor to eliminate defect positioning deviation value.

Images