A hollow plastic board surface defect detection apparatus and method

Abstract

This invention discloses a device and method for detecting surface defects in hollow plastic sheets, belonging to the field of plastic sheet inspection technology. The method obtains the spatiotemporal distortion coefficient based on the instantaneous linear velocity fluctuation rate of the plastic sheet, the dynamic warpage of the sheet surface, and the conveyor belt amplitude; it obtains the apparent state coefficient based on the ratio of PM1.0 to PM10 in airborne dust, the dyne value of the plastic sheet surface, and the electrostatic potential of the plastic sheet; and it obtains the defect risk coefficient based on the extruder die pressure fluctuation value, material moisture content, and the maximum temperature difference of the sheet surface. Furthermore, it obtains the steady-state fit by combining the internal stress and polarization fluctuation rate of the ultrasonic waves, and obtains an optimized line frequency trigger phase based on the steady-state fit, the apparent state coefficient, and the current line scan camera's line frequency trigger phase. This invention can adaptively compensate for imaging distortion and optical interference caused by speed fluctuations, sheet warpage, environmental dust, and process fluctuations, improving the stability and accuracy of defect detection.

Description

Technical Field

[0001] This invention belongs to the field of plastic sheet testing technology, and particularly relates to a device and method for detecting surface defects in hollow plastic sheets. Background Technology

[0002] Hollow plastic sheet surface defect detection is mainly used for high-speed, high-precision defect identification and location on the surface of hollow plastic sheets after extrusion molding.

[0003] Currently, surface defect detection in hollow plastic sheets mainly relies on manual visual inspection or traditional machine vision systems. Manual inspection is inefficient, highly subjective, and difficult to meet the needs of continuous production. Existing machine vision inspection systems typically use line-scan cameras with a constant line frequency triggering method, assuming a constant conveyor belt speed, a flat sheet surface, and stable ambient lighting. Based on this, they identify surface defects such as scratches, dents, and stains through image acquisition and image processing algorithms. Some systems also incorporate ultrasonic detection or electrostatic sensing as auxiliary methods for judgment.

[0004] However, numerous dynamic interference factors exist in actual production processes, leading to insufficient stability and accuracy of existing detection methods. For example, fluctuations in conveyor belt speed and dynamic warping of the sheet material during transport can cause geometric distortion in imaging; dust adhesion in the workshop air, static electricity accumulation on the sheet surface, and changes in surface energy can generate optical interference, affecting image clarity; process fluctuations such as extruder die pressure fluctuations, changes in material moisture content, and uneven sheet surface temperature can cause abnormal internal stress distribution. These internal state changes cannot be perceived by traditional vision systems but may induce surface defects or cause variations in defect characteristics. Existing technologies lack the ability to comprehensively model and dynamically compensate for the above-mentioned multi-source interferences, making it difficult to achieve adaptive optimization of the detection system under complex working conditions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a device and method for detecting surface defects in hollow plastic sheets, thus solving the aforementioned problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for detecting surface defects in hollow plastic sheets, comprising:

[0007] Based on the instantaneous linear velocity fluctuation rate (velocity standard deviation) of the plastic board, the dynamic warping degree of the board surface (radius of curvature), and the amplitude of the conveyor belt, the spatiotemporal distortion coefficient is obtained through a spatiotemporal distortion model.

[0008] Based on the ratio of PM1.0 / PM10 in airborne dust, the dyne value of the plastic board surface, and the electrostatic potential of the plastic board, the apparent state coefficient is obtained through the apparent state model.

[0009] Based on the pressure fluctuation value of the extruder die head (pressure standard deviation), material moisture content, and maximum temperature difference of the plate surface, the defect risk coefficient is obtained through a defect risk model.

[0010] Based on the spatiotemporal distortion coefficient and defect risk coefficient, the internal stress and polarization fluctuation rate of the ultrasonic wave in plastic sheet (the standard deviation of polarization in a continuous time series) are obtained by the steady-state fit model.

[0011] Based on the steady-state fit, apparent state coefficient, and the current line frequency trigger phase of the line scan camera, the optimized line frequency trigger phase is output through the phase optimization model.

[0012] Based on the above technical solutions, the present invention also provides the following optional technical solutions:

[0013] Further technical solution: The steps to obtain the optimized line frequency trigger phase are as follows:

[0014] Based on the apparent state coefficients, the transient response sensitivity and steady-state convergence factor are obtained and expressed as follows:

[0015]

[0016]

[0017] in, Indicates transient response sensitivity. Denotes the steady-state convergence factor. Indicates the basic transient response sensitivity. This represents the basic steady-state convergence factor. Indicates the gain attenuation coefficient and ;

[0018] Based on the steady-state fit, the current line frequency trigger phase of the line scan camera, and combined with the transient response sensitivity and steady-state convergence factor, an optimized line frequency trigger phase is output through a phase optimization model. The phase optimization model is expressed as follows:

[0019]

[0020] in, This indicates the optimized line frequency trigger phase. Indicates the current line frequency trigger phase of the line scan camera. Indicates the steady-state fit. This represents the target value for steady-state fit. This indicates the steady-state fit at historical moments. The instantaneous value, Indicates transient response sensitivity. This represents the steady-state convergence factor.

[0021] Further technical solution: The steps to obtain the steady-state fit are as follows:

[0022] Based on the spatiotemporal distortion coefficient and the defect risk coefficient, and combined with the ultrasonic internal stress of the plastic board and the polarization degree fluctuation rate (the standard deviation of polarization degree in a continuous time series), the expected values ​​of the ultrasonic internal stress and polarization degree fluctuation rate of the plastic board are obtained.

[0023] Based on the expected values ​​of the internal stress and polarization fluctuation rate of the ultrasonic wave in the plastic board, and combined with the corresponding internal stress and polarization fluctuation rate of the ultrasonic wave in the plastic board, the stress deviation metric and polarization fluctuation rate deviation metric are obtained.

[0024] Based on the stress deviation metric and the polarization degree fluctuation rate deviation metric, the steady-state fit is obtained through a steady-state fit model, which is expressed as follows:

[0025]

[0026] in, Indicates the steady-state fit. This represents a measure of stress deviation. This represents a measure of polarization degree volatility deviation. This represents the product of the stress deviation metric and the polarization degree fluctuation rate deviation metric. Indicates the attenuation coefficient and The Furthermore, the higher the value, the more stable the detection system.

[0027] Further technical solution: The method for obtaining the defect risk coefficient is as follows:

[0028] The pressure fluctuation value (pressure standard deviation), material moisture content, and maximum difference in plate surface temperature of the extruder die head are compared with the corresponding reference values ​​to obtain the pressure fluctuation index, material moisture content index, and temperature fluctuation index.

[0029] Based on the pressure fluctuation index, material moisture content index, and temperature fluctuation index, the defect risk coefficient is obtained through a defect risk model, which is expressed as follows:

[0030]

[0031] in, Indicates the defect risk coefficient. This indicates a pressure fluctuation index. Indicates the moisture content index of the material. Indicates the temperature fluctuation index. Indicates the weighted index and The Furthermore, the larger the value, the greater the precursor defects in the process.

[0032] Further technical solution: The method for obtaining the apparent state coefficients is as follows:

[0033] The ratio of PM1.0 / PM10 in airborne dust and the electrostatic potential of plastic boards are compared with the corresponding reference values ​​to obtain the dust impact index and electrostatic potential index.

[0034] The dyne value of the plastic board surface is inversely proportional to the dyne value reference value to obtain the dyne value index;

[0035] Based on the dust impact index, dyne value index, and electrostatic potential index, apparent state coefficients are obtained through an apparent state model, which is expressed as follows:

[0036]

[0037] in, Represents the apparent state coefficient. Indicates the dust impact index. Indicates the Dyne index, Indicates the electrostatic potential index. Represents the weight coefficient and The Furthermore, the larger the value, the easier it is to generate optical interference during detection.

[0038] Further technical solution: The steps for obtaining the spatiotemporal distortion coefficients are as follows:

[0039] The instantaneous linear velocity fluctuation rate (speed standard deviation) of the plastic board and the amplitude of the conveyor belt are compared with the corresponding reference values ​​to obtain the speed fluctuation rate index and amplitude index.

[0040] The curvature radius index is obtained by inversely proportionalizing the dynamic warp degree (radius of curvature) of the board surface to the reference value of the radius of curvature.

[0041] Based on the velocity volatility index, radius of curvature index, and amplitude index, the spatiotemporal distortion coefficients are obtained through a spatiotemporal distortion model, which is expressed as follows:

[0042]

[0043] in, Represents the spatiotemporal distortion coefficient. Indicates the velocity volatility index. Represents the radius of curvature exponent. The amplitude index is represented by the following. Furthermore, the larger the value, the more severe the geometric distortion of the image.

[0044] Further technical solution: The method for obtaining the expected values ​​of internal stress and polarization fluctuation rate of the plastic board via ultrasonic testing is as follows:

[0045] Based on the spatiotemporal distortion coefficient and defect risk coefficient, and combined with the internal stress of the PVC board during ultrasonic testing, the expected value of the internal stress of the PVC board during ultrasonic testing is obtained through a stress expectation model, which is expressed as follows:

[0046]

[0047] in, This represents the expected value of the internal stress of the plastic sheet under ultrasonic testing. This represents the internal stress of the plastic sheet under ideal ultrasonic conditions. Represents the spatiotemporal distortion coefficient. Indicates the defect risk coefficient. Represents the sensitivity coefficient and ;

[0048] Based on the spatiotemporal distortion coefficient and the defect risk coefficient, and combined with the polarization degree volatility (the standard deviation of polarization degree over a continuous time series), the expected value of polarization degree volatility is obtained through a polarization degree volatility expectation model, which is expressed as follows:

[0049]

[0050] in, This represents the expected value of the polarization degree volatility. This represents the polarization degree fluctuation rate under ideal operating conditions. Represents the spatiotemporal distortion coefficient. Indicates the defect risk coefficient. Represents the sensitivity coefficient and .

[0051] Further technical solution: The method for obtaining stress deviation measurement values ​​and polarization degree fluctuation rate deviation measurement values ​​is as follows:

[0052] Based on the expected value of the internal stress of the plastic sheet under ultrasonic testing and the internal stress of the plastic sheet under ultrasonic testing, the stress deviation metric value is obtained through a stress deviation metric model, which is expressed as follows:

[0053]

[0054] in, This represents a measure of stress deviation. This indicates the internal stress of the plastic sheet during ultrasonic testing. This represents the expected value of the internal stress of the plastic sheet under ultrasonic testing. Indicates a decimal excluding zero and whose unit is the same as the decimal. Consistent;

[0055] Based on the expected value and the actual polarization volatility, a polarization volatility deviation metric is obtained using a polarization volatility deviation metric model, which is expressed as follows:

[0056]

[0057] in, This represents a measure of polarization degree volatility deviation. This represents the expected value of the polarization degree volatility. Indicates the polarization degree fluctuation rate. Indicates protection against zero decimals.

[0058] A device for detecting surface defects in hollow plastic sheets, which uses the above-mentioned method for detecting surface defects in hollow plastic sheets to detect surface defects in hollow plastic sheets.

[0059] This invention provides a device and method for detecting surface defects in hollow plastic sheets, which has the following advantages compared with the prior art:

[0060] 1. This invention constructs a spatiotemporal distortion model, an apparent state model, and a defect risk model to quantitatively characterize multi-source disturbances such as conveyor speed fluctuations, dynamic warping of the sheet material, dust pollution, electrostatic interference, material moisture content, die pressure fluctuations, and uneven sheet surface temperature. This enables systematic modeling of the detection environment and process state, providing a reliable basis for the adaptive adjustment of subsequent detection parameters.

[0061] 2. This invention constructs a steady-state fit model based on the internal stress and polarization fluctuation rate of ultrasonic waves, and dynamically adjusts the line frequency trigger phase of the line scan camera in combination with the apparent state coefficient to form a closed-loop optimization mechanism. This can effectively suppress imaging geometric distortion and optical interference caused by equipment vibration, plate deformation and environmental changes, and significantly improve the image acquisition quality.

[0062] 3. Without increasing additional hardware costs, this invention achieves adaptive correction of the detection system under complex working conditions through multi-model fusion and phase dynamic optimization, effectively improving the stability, accuracy and online detection efficiency of hollow plastic board surface defect detection. Attached Figure Description

[0063] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0065] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0066] Please see Figure 1 The present invention provides a method for detecting surface defects in hollow plastic sheets, comprising:

[0067] Based on the instantaneous linear velocity fluctuation rate (velocity standard deviation) of the plastic board, the dynamic warping degree of the board surface (radius of curvature), and the amplitude of the conveyor belt, the spatiotemporal distortion coefficient is obtained through a spatiotemporal distortion model.

[0068] Based on the ratio of PM1.0 / PM10 in airborne dust, the dyne value of the plastic board surface, and the electrostatic potential of the plastic board, the apparent state coefficient is obtained through the apparent state model.

[0069] Based on the pressure fluctuation value of the extruder die head (pressure standard deviation), material moisture content, and maximum temperature difference of the plate surface, the defect risk coefficient is obtained through a defect risk model.

[0070] Based on the spatiotemporal distortion coefficient and defect risk coefficient, the internal stress and polarization fluctuation rate of the ultrasonic wave in plastic sheet (the standard deviation of polarization in a continuous time series) are obtained by the steady-state fit model.

[0071] Based on the steady-state fit, apparent state coefficient, and the current line frequency trigger phase of the line scan camera, the optimized line frequency trigger phase is output through the phase optimization model.

[0072] This embodiment achieves adaptive optimization of the hollow plastic sheet surface defect detection system by comprehensively modeling and dynamically compensating for multi-source dynamic interference factors. Compared with existing detection methods that rely on fixed line frequency triggering, this embodiment can perceive and quantify in real time the spatiotemporal distortion caused by instantaneous linear velocity fluctuations of the plastic sheet, dynamic warping of the sheet surface, and conveyor belt amplitude, as well as apparent state interference caused by air dust, dyne values ​​of the plastic sheet surface, and electrostatic potential. Furthermore, this embodiment can assess potential defect risks and production steady-state fit by combining process parameters such as extruder die pressure fluctuations, material moisture content, and maximum temperature difference of the sheet surface with the ultrasonic internal stress and polarization fluctuation rate of the plastic sheet. These internal state changes cannot be directly perceived by traditional vision systems. By integrating this multi-dimensional information and dynamically optimizing the line frequency triggering phase of the line scan camera, this embodiment effectively compensates for the impact of various interference factors on the accuracy of image acquisition. Therefore, this embodiment significantly improves the stability and accuracy of hollow plastic sheet surface defect detection in complex production environments, overcomes the limitations of existing technologies in terms of performance degradation under multi-source interference, and provides technical support for achieving high-precision and high-reliability industrial online detection.

[0073] Preferably, the steps for obtaining the spatiotemporal distortion coefficients are as follows:

[0074] The instantaneous linear velocity fluctuation rate (speed standard deviation) of the plastic board and the amplitude of the conveyor belt are compared with the corresponding reference values ​​to obtain the speed fluctuation rate index and amplitude index.

[0075] The curvature radius index is obtained by inversely proportionalizing the dynamic warp degree (radius of curvature) of the board surface to the reference value of the radius of curvature.

[0076] Based on the velocity volatility index, radius of curvature index, and amplitude index, the spatiotemporal distortion coefficients are obtained through a spatiotemporal distortion model, which is expressed as follows:

[0077]

[0078] in, Represents the spatiotemporal distortion coefficient. Indicates the velocity volatility index. Represents the radius of curvature exponent. The amplitude index is represented by the following. Furthermore, the larger the value, the more severe the geometric distortion of the image.

[0079] The instantaneous linear velocity fluctuation rate (velocity standard deviation) of the hollow plastic sheet refers to the dispersion of its instantaneous linear velocity over a period of time during the conveying process. It is typically quantified by calculating the standard deviation of the velocity. This parameter reflects the smoothness of the plastic sheet's movement; the greater the velocity fluctuation, the higher the risk of stretching or compression distortion in the image along the scanning direction. It can be obtained by collecting velocity data in real time using sensors such as encoders and laser velocimeters installed on the conveyor belt or plastic sheet, and then statistically analyzing the collected data. Conveyor belt amplitude refers to the vertical or horizontal vibration amplitude of the conveyor belt during operation. The vibration of the conveyor belt causes the plastic sheet to shift position perpendicular to the scanning direction, resulting in lateral distortion in the image. This parameter can be monitored and measured in real time by installing vibration sensors or displacement sensors at key locations on the conveyor belt. The dynamic warpage (radius of curvature) of a hollow plastic sheet refers to the degree of bending of its surface under dynamic conditions during transport. It is typically characterized by the radius of curvature. Surface warpage causes changes in the distance between the plastic sheet surface and the camera's focal plane during line scanning imaging, leading to image defocusing and geometric distortion. This parameter can be obtained by acquiring surface contour data using a laser triangulation sensor, a line scan camera, and image processing technology, and then calculating its radius of curvature. Reference values ​​refer to the baseline values ​​corresponding to the instantaneous linear velocity fluctuation rate of the plastic sheet, the conveyor belt amplitude, and the dynamic warpage (radius of curvature) of the plastic sheet under ideal or stable production conditions. These reference values ​​serve as the basis for measuring the degree to which the current operating conditions deviate from a stable state and can be determined based on historical production data, process specifications, or experimental calibration results. Ratio processing is a data standardization method that converts physical quantities of different dimensions into dimensionless exponents by comparing the current measured value with the corresponding reference value, facilitating unified processing and comparison in subsequent models. Inverse proportionality processing is a data transformation method. When a physical quantity is negatively correlated with the characteristic to be represented, its reciprocal or other inverse operations are used to convert it into an index that is positively correlated with the characteristic. The velocity volatility index is a dimensionless parameter obtained by comparing the instantaneous linear velocity volatility of the plastic sheet with a corresponding reference value. It is used to quantify the influence of the plastic sheet's motion stability on imaging geometric distortion. The amplitude index is a dimensionless parameter obtained by comparing the conveyor belt amplitude with a corresponding reference value. It is used to quantify the influence of conveyor belt vibration on imaging geometric distortion. The radius of curvature index is a dimensionless parameter obtained by inversely proportionalizing the dynamic warp (radius of curvature) of the board surface with a reference value for the radius of curvature. It is used to quantify the influence of board surface warp on imaging geometric distortion. The spatiotemporal distortion model is a mathematical function used to integrate the velocity volatility index, radius of curvature index, and amplitude index to quantify the degree of spatiotemporal distortion experienced by the hollow plastic sheet during production. This model integrates these independent distortion factors into a unified spatiotemporal distortion coefficient, providing a key input for subsequent defect detection and phase optimization.The spatiotemporal distortion coefficient is a dimensionless parameter calculated using a spatiotemporal distortion model, with a value between 0 and 1. This coefficient comprehensively reflects the degree of imaging geometric distortion caused by factors such as the movement of the plastic sheet, the shape of the sheet surface, and the vibration of the conveyor belt; a larger value indicates more severe imaging geometric distortion.

[0080] This application's solution systematically quantifies multiple spatiotemporal factors affecting the imaging quality of hollow plastic panels to obtain accurate spatiotemporal distortion coefficients. First, considering the panel's motion characteristics, the instantaneous linear velocity fluctuation rate of the panel and the conveyor belt amplitude are monitored in real time. These parameters are directly related to the geometric distortion of the image in the scanning direction and laterally. Simultaneously, considering the physical shape of the panel itself, the warpage of the panel surface is dynamically evaluated, as changes in its radius of curvature affect the image sharpness and geometric accuracy. To unify these different physical quantities, the instantaneous linear velocity fluctuation rate of the panel and the conveyor belt amplitude are compared with their respective reference values ​​to obtain the velocity fluctuation rate index and amplitude index, thus achieving parameter standardization. For the dynamic warpage of the panel surface, since its radius of curvature is inversely related to the degree of distortion, it is inversely proportional to the reference value to obtain the radius of curvature index, ensuring that all indices positively reflect the degree of distortion. Finally, these three standardized and transformed indices are used as inputs into a preset spatiotemporal distortion model. This model, through specific mathematical relationships, integrates these independent distortion indices into a single spatiotemporal distortion coefficient. This coefficient can comprehensively and quantitatively characterize the degree of geometric distortion faced by hollow plastic sheet imaging under current production conditions, providing accurate input for subsequent steady-state fit calculation and line frequency trigger phase optimization, thereby ensuring the accuracy of defect detection.

[0081] The following concrete example illustrates how to obtain the spatiotemporal distortion coefficient: Multiple laser velocimeters or encoders are installed along the plastic sheet conveyor path to collect the instantaneous linear velocity data of the plastic sheet in real time. The standard deviation of this instantaneous linear velocity over a continuous time series is calculated by a data processing unit, representing the instantaneous linear velocity fluctuation rate of the plastic sheet. Simultaneously, high-precision accelerometers or displacement sensors are installed on the support structure of the conveyor belt to continuously monitor its vibration and calculate its amplitude. Furthermore, a vision system consisting of a line laser scanner and an industrial camera is deployed above the plastic sheet. By projecting the laser line onto the plastic sheet surface and combining it with image processing algorithms, the surface contour of the plastic sheet is reconstructed in real time, and its dynamic radius of curvature is calculated, thus obtaining the dynamic warp of the sheet surface. The real-time instantaneous linear velocity fluctuation rate of the plastic sheet is compared with a pre-set velocity fluctuation rate reference value to obtain the velocity fluctuation rate index. The conveyor belt amplitude is compared with a preset amplitude reference value to obtain the amplitude index. For the dynamic warp of the sheet surface, the calculated radius of curvature is inversely proportional to a preset radius of curvature reference value to obtain the radius of curvature index. These reference values ​​can be calibrated based on typical production process parameters and historical stable operating data of hollow plastic sheets. Finally, the calculated velocity fluctuation rate index, radius of curvature index, and amplitude index are used as inputs and substituted into the spatiotemporal distortion model. The control system or dedicated processor performs the calculation, and the spatiotemporal distortion coefficient under the current operating conditions can be output in real time.

[0082] The above technical solution can accurately quantify the degree of imaging geometric distortion caused by factors such as speed fluctuations, board warping, and conveyor belt vibration during the production of hollow plastic sheets. This precise acquisition of spatiotemporal distortion coefficients provides reliable input for subsequent steady-state fit calculations, avoiding deviations in steady-state fit due to inaccurate distortion assessment. Therefore, when performing line frequency trigger phase optimization based on steady-state fit, the camera trigger timing can be adjusted more accurately, effectively compensating for image misalignment or blurring caused by spatiotemporal distortion. This significantly improves the accuracy and stability of hollow plastic sheet surface defect detection, reducing the probability of missed and false detections.

[0083] Preferably, the apparent state coefficients are obtained in the following way:

[0084] The ratio of PM1.0 / PM10 in airborne dust and the electrostatic potential of plastic boards are compared with the corresponding reference values ​​to obtain the dust impact index and electrostatic potential index.

[0085] The dyne value of the plastic board surface is inversely proportional to the dyne value reference value to obtain the dyne value index;

[0086] Based on the dust impact index, dyne value index, and electrostatic potential index, apparent state coefficients are obtained through an apparent state model, which is expressed as follows:

[0087]

[0088] in, Represents the apparent state coefficient. Indicates the dust impact index. Indicates the Dyne index, Indicates the electrostatic potential index. Represents the weight coefficient and The Furthermore, the larger the value, the easier it is to generate optical interference during detection.

[0089] The PM1.0 / PM10 ratio in airborne dust reflects the relative content of fine and coarse particulate matter in the air. It is an important indicator for measuring air cleanliness, especially its impact on optical imaging. PM1.0 particles, due to their small size, are more likely to form a thin layer or adhere to the surface of the plastic sheet, causing scattering or absorption in optical detection. PM10 particles, on the other hand, can create greater obstruction. This ratio can be monitored and acquired in real time using devices such as laser scattering particulate sensors or photoelectric counters. The electrostatic potential of the plastic sheet is an indicator of the degree of charge on its surface. A high electrostatic potential will attract dust particles from the air, making them adhere more firmly to the plastic sheet surface, thus exacerbating optical interference. The electrostatic potential of the plastic sheet can be measured using a non-contact electrostatic voltmeter or electrostatic field measuring instrument. Reference values ​​are benchmark values ​​used to compare and quantify the deviation of current measured values. These reference values ​​can be set based on ideal production environment conditions, historical data statistical averages, or optimal process parameters determined experimentally. Ratio processing mathematically compares the actual measured value with the preset reference value to quantify the deviation of the current state from the benchmark state. The Dust Impact Index (DII) is a ratio-based indicator that quantifies the impact of airborne dust on the optical detection of PVC panels. A higher DPI value indicates a greater potential interference from airborne dust. The Electrostatic Potential Index (ESP) is a ratio-based indicator that quantifies the impact of electrostatics on dust adsorption and optical detection on PVC panels. A higher ESP value indicates a higher risk of electrostatic dust adsorption on the PVC panel surface and a greater likelihood of optical interference.

[0090] The dyne value of a plastic sheet surface measures the surface energy or surface tension of a material, reflecting the wettability of the plastic sheet surface to liquids (such as water and oil). A low dyne value usually indicates the presence of oil, mold release agent residue, or other contaminants on the surface. These contaminants alter the optical properties of the plastic sheet surface, causing light scattering and uneven reflection, thus affecting the imaging quality of a line scan camera. The dyne value of a plastic sheet surface can be measured using a dyne tester or a contact angle meter. The dyne value reference value serves as a benchmark for measuring the cleanliness and wettability of the plastic sheet surface. This reference value can be determined based on the inherent characteristics of the plastic sheet material, production process requirements, or optimal surface treatment. Inverse comparison processing involves comparing the actual measured value with a preset reference value mathematically to quantify the deviation of the current state from the baseline state. The dyne value index, after inverse comparison processing, quantifies the impact of plastic sheet surface cleanliness on optical detection. A higher value indicates more severe surface contamination and a greater potential interference with detection. The apparent state model is a mathematical function used to calculate the apparent state coefficient by integrating the dust influence index, dyne value index, and electrostatic potential index. This model employs a sigmoid function, mapping multiple input variables to an output value between 0 and 1. The negative sign before the exponent and the exponential function ensure that as the input exponent value increases, the apparent state coefficient also increases, consistent with its physical meaning that "a larger value makes it easier to generate optical interference during detection." The apparent state coefficient is a quantitative indicator that comprehensively reflects the influence of the physicochemical state of the plastic sheet surface and environmental factors on optical detection. Its value ranges from 0 to 1; a larger value indicates that the plastic sheet surface is more prone to optical interference, thus affecting the accuracy of defect detection. Weighting coefficients are used to adjust the relative importance of each input index (dust influence index, dyne value index, and electrostatic potential index) in the apparent state model. These coefficients can be optimized and determined through expert experience, historical data analysis, or machine learning algorithms.

[0091] This application aims to accurately quantify the external environmental and surface condition factors affecting the optical detection of defects on hollow plastic sheets by introducing an apparent state coefficient. First, the PM1.0 / PM10 ratio and the electrostatic potential of the plastic sheet are monitored in real time and compared with preset reference values ​​to obtain the dust impact index and electrostatic potential index, respectively. These indices intuitively reflect the potential interference of environmental cleanliness and the surface charge of the plastic sheet on optical imaging. Simultaneously, the dyne value of the plastic sheet surface is measured and inversely compared with a dyne value reference value to obtain the dyne value index. This index characterizes the cleanliness and wettability of the plastic sheet surface, thus reflecting the impact of surface contaminants on optical detection. Subsequently, these three quantified indices are used as inputs into a preset apparent state model. This model performs a weighted summation of the indices using weighting coefficients and maps the results to a range of 0 to 1 using a Sigmoid function, ultimately outputting the apparent state coefficient. This coefficient comprehensively reflects the degree to which the current apparent state of the plastic sheet surface and the environment interferes with optical detection. In this way, this application can transform complex environmental and surface factors into a unified quantitative index, which is then introduced into the overall phase optimization model. This index works together with the steady-state fit and the line frequency trigger phase of the current line scan camera, so that the optimized line frequency trigger phase can more accurately adapt to optical interference caused by changes in appearance, ensuring the robustness and accuracy of defect detection.

[0092] The following is a specific example. To obtain the PM1.0 / PM10 ratio in airborne dust, a laser scattering particulate matter sensor, such as the PMS5003 series sensor, can be installed above or near the detection area of ​​the hollow plastic sheet production line to monitor the concentration of PM1.0 and PM10 in the air in real time and calculate their ratio. Simultaneously, to obtain the electrostatic potential of the plastic sheet, a non-contact electrostatic voltmeter, such as the Trek520D electrostatic field meter, can be installed along the plastic sheet transport path to measure the electrostatic potential of the plastic sheet surface in real time. These measured values ​​are then compared with pre-set reference values. For example, if the ideal PM1.0 / PM10 ratio is 0.4 and the actual measurement is 0.6, then the dust impact index is 0.6 / 0.4 = 1.5; if the ideal electrostatic potential reference value is 100V and the actual measurement is 200V, then the electrostatic potential index is 200V / 100V = 2. Secondly, to obtain the dyne value of the PVC sheet surface, it can be tested periodically or online using a dyne tester, or non-contactly measured using a contact angle meter. For example, if the reference dyne value of the PVC sheet is 40 dynes / cm, and the actual measured value is 30 dynes / cm, then the dyne index is 40 / 30 ≈ 1.33. Finally, these calculated dust impact index $X_{1}$ and dyne index... and electrostatic potential index Substituting this into the apparent state model, for example, if the weight coefficients... , , If the values ​​are 0.5, 0.3, and 0.2 respectively, then the apparent state coefficients can be calculated. In this way, the impact of the surface appearance of the plastic sheet on optical inspection can be assessed in real time and quantitatively.

[0093] Through the above technical solution, this application can accurately quantify the degree of optical interference on the surface of hollow plastic sheets caused by factors such as airborne dust, electrostatic adsorption, and surface cleanliness, and generate an apparent state coefficient. This coefficient serves as an important input to the phase optimization model, enabling the line frequency trigger phase of the line scan camera to adaptively adjust according to the actual apparent state of the plastic sheet surface. This effectively solves the problem of decreased detection accuracy due to changes in surface optical properties in complex production environments, significantly improves the accuracy and robustness of defect detection, avoids misjudgments or missed detections caused by optical interference, and thus enhances the overall performance and reliability of the hollow plastic sheet surface defect detection method.

[0094] Preferably, the defect risk coefficient is obtained in the following way:

[0095] The pressure fluctuation value (pressure standard deviation), material moisture content, and maximum difference in plate surface temperature of the extruder die head are compared with the corresponding reference values ​​to obtain the pressure fluctuation index, material moisture content index, and temperature fluctuation index.

[0096] Based on the pressure fluctuation index, material moisture content index, and temperature fluctuation index, the defect risk coefficient is obtained through a defect risk model, which is expressed as follows:

[0097]

[0098] in, Indicates the defect risk coefficient. This indicates a pressure fluctuation index. Indicates the moisture content index of the material. Indicates the temperature fluctuation index. Indicates the weighted index and The Furthermore, the larger the value, the greater the precursor defects in the process.

[0099] The pressure fluctuation value (pressure standard deviation) of the extruder die refers to the instantaneous change in melt pressure at the extruder die during the extrusion process. It is usually quantified by calculating the pressure standard deviation over a period of time. Large pressure fluctuations may lead to extrusion instability, thereby affecting the thickness uniformity or internal structure of the plastic sheet. This pressure fluctuation value can be monitored in real time by a pressure sensor installed at the extruder die and processed and calculated by a data acquisition system. Material moisture content refers to the proportion of water contained in the raw materials used to produce hollow plastic sheets. Excessive moisture content or its fluctuations can affect the plasticizing effect of the material, potentially leading to defects such as bubbles, surface roughness, or reduced strength in the product. Material moisture content can be continuously measured by an online humidity sensor or periodically tested using an offline sampling drying method. The maximum surface temperature difference refers to the highest temperature difference between different points on the surface of the hollow plastic sheet during molding or cooling. Uneven surface temperature may lead to inconsistent cooling rates, resulting in internal stress, warping, or surface defects. This maximum difference can be obtained by scanning the entire surface with an infrared thermal imager or by placing multiple non-contact temperature sensors at key locations on the surface for measurement and calculation.

[0100] The above-mentioned extruder die pressure fluctuation values, material moisture content, and maximum difference in plate surface temperature are compared with their corresponding reference values. The aim is to standardize these process parameters, which have different physical dimensions and numerical ranges, and transform them into a dimensionless risk index. For example, the pressure fluctuation index can be obtained by dividing the actual measured pressure fluctuation value by a preset ideal or acceptable pressure fluctuation reference value. This ratio processing allows different types of process parameters to be compared and comprehensively evaluated on a uniform scale, providing consistent input for subsequent defect risk models.

[0101] The defect risk model is a mathematical model used to comprehensively assess the potential impact of multiple process parameters on defects. This model employs a non-linear exponential function, incorporating the standardized pressure fluctuation index... Material moisture content index and temperature fluctuation index Through product terms and weighting indices The weighted combination is performed, and the final output is a defect risk coefficient between 0 and 1. The model is designed such that when any one or more process parameters deviate from the ideal state, the corresponding exponent increases, thereby increasing the defect risk coefficient. An increased weighting index can effectively reflect the severity of early-stage defects in the process. Optimization can be performed based on historical production data, expert experience, or through machine learning algorithms to accurately reflect the relative contribution of different process parameters to the risk of defects in the final product.

[0102] The proposed solution involves real-time monitoring of key process parameters such as extruder die pressure fluctuations, material moisture content, and maximum temperature difference on the extruder surface. These parameters are then compared with preset reference values ​​to obtain standardized pressure fluctuation indices, material moisture content indices, and temperature fluctuation indices. Subsequently, these indices are input into a defect risk model. This model, through its nonlinear mathematical structure, integrates these independent process risk indicators into a single, quantified defect risk coefficient. This model can capture the impact of multiple process parameters on defects, thus increasing the defect risk coefficient when any one or more key parameters in the production process become abnormal. It can respond sensitively and amplify, thus accurately reflecting the potential defect risks under the current process conditions. This quantitative assessment method provides accurate and forward-looking input for subsequent steady-state fit calculations, significantly improving the overall system stability and predictive ability of the hollow plastic sheet surface defect detection method.

[0103] The following is a concrete example to illustrate this. Suppose that on a hollow plastic sheet production line, an online sensor and data acquisition system obtains in real-time data showing that the extruder die pressure fluctuation (pressure standard deviation) is 0.4 MPa, the material moisture content is 0.025%, and the maximum temperature difference on the sheet surface is 4°C. Simultaneously, the system's preset ideal reference values ​​are: ideal pressure fluctuation 0.1 MPa, ideal material moisture content 0.01%, and ideal maximum temperature difference on the sheet surface 1°C. First, ratio processing is performed to obtain various indices: Pressure Fluctuation Index... =0.4 / 0.1 =4; Material moisture content index =0.025 / 0.01=2.5; Temperature fluctuation index =4 / 1=4. Next, substitute these indices into the defect risk model, assuming it has been trained with historical data, and the weighted indices are respectively... =0.7, =1.0, =0.8, then the defect risk coefficient is =0.75. The higher the defect risk coefficient value, the greater the risk of defects occurring under the current process conditions.

[0104] Through the aforementioned technical solution, this application can transform key process parameters such as extruder die pressure fluctuations, material moisture content, and maximum temperature difference on the board surface into standardized risk indices. These indices are then comprehensively evaluated using a defect risk model, enabling early, quantitative warnings of potential process defects in hollow plastic sheet production. This quantitative assessment method allows the defect risk coefficient to accurately reflect the health of the process state, providing reliable input for subsequent steady-state fit calculations. This significantly improves the sensitivity and predictive ability of the entire hollow plastic sheet surface defect detection method to potential defects. By accurately capturing and quantifying early-stage process defects, this application helps to adjust production parameters in a timely manner, reduce the occurrence of actual defects, and improve product quality and production efficiency.

[0105] Preferably, the step of obtaining the steady-state fit is as follows:

[0106] Based on the spatiotemporal distortion coefficient and defect risk coefficient, and combined with the internal stress of the PVC board during ultrasonic testing, the expected value of the internal stress of the PVC board during ultrasonic testing is obtained through a stress expectation model, which is expressed as follows:

[0107]

[0108] in, This represents the expected value of the internal stress of the plastic sheet under ultrasonic testing. This represents the internal stress of the plastic sheet under ideal ultrasonic conditions. Represents the spatiotemporal distortion coefficient. Indicates the defect risk coefficient. Represents the sensitivity coefficient and ;

[0109] Based on the spatiotemporal distortion coefficient and the defect risk coefficient, and combined with the polarization degree volatility (the standard deviation of polarization degree over a continuous time series), the expected value of polarization degree volatility is obtained through a polarization degree volatility expectation model, which is expressed as follows:

[0110]

[0111] in, This represents the expected value of the polarization degree volatility. This represents the polarization degree fluctuation rate under ideal operating conditions. Represents the spatiotemporal distortion coefficient. Indicates the defect risk coefficient. Represents the sensitivity coefficient and ;

[0112] Based on the expected value of the internal stress of the plastic sheet under ultrasonic testing and the internal stress of the plastic sheet under ultrasonic testing, the stress deviation metric value is obtained through a stress deviation metric model, which is expressed as follows:

[0113]

[0114] in, This represents a measure of stress deviation. This indicates the internal stress of the plastic sheet during ultrasonic testing. This represents the expected value of the internal stress of the plastic sheet under ultrasonic testing. Indicates a decimal excluding zero and whose unit is the same as the decimal. Consistent;

[0115] Based on the expected value and the actual polarization volatility, a polarization volatility deviation metric is obtained using a polarization volatility deviation metric model, which is expressed as follows:

[0116]

[0117] in, This represents a measure of polarization degree volatility deviation. This represents the expected value of the polarization degree volatility. Indicates the polarization degree fluctuation rate. Indicates protection against zero decimals;

[0118] Based on the stress deviation metric and the polarization degree fluctuation rate deviation metric, the steady-state fit is obtained through a steady-state fit model, which is expressed as follows:

[0119]

[0120] in, Indicates the steady-state fit. This represents a measure of stress deviation. This represents a measure of polarization degree volatility deviation. This represents the product of the stress deviation metric and the polarization degree fluctuation rate deviation metric. Indicates the attenuation coefficient and The Furthermore, the higher the value, the more stable the detection system.

[0121] The step of obtaining steady-state fit aims to provide the defect detection system with an accurate indicator reflecting the current production stability by comprehensively evaluating the internal physical state of the PVC board. This step is crucial for ensuring the effectiveness of subsequent phase optimization. The stress expectation model is used to predict the ideal value of the ultrasonic internal stress of the PVC board under specific spatiotemporal distortion coefficients and defect risk coefficients. This model can be trained based on historical data, for example, through regression analysis or machine learning algorithms to establish a mapping relationship between input parameters and expected stress; or it can be derived based on physical models of material mechanics and process parameters. The polarization fluctuation rate expectation model is similar to the stress expectation model, predicting the ideal value of the polarization fluctuation rate of the PVC board under current process conditions. This model can also be learned and predictive based on a large amount of production data using data-driven methods, such as statistical modeling or neural networks; or it can be modeled based on optical physics principles and material properties. The stress deviation measurement model is used to quantify the difference between the actual measured ultrasonic internal stress of the PVC board and its expected value. In addition to the above formula, this model can also use other statistical methods such as relative error and root mean square error to evaluate the deviation. The polarization degree volatility deviation measurement model is used to quantify the difference between the actual measured polarization degree volatility and its expected value. Similar to the stress deviation measurement model, it can also be implemented using various statistical measurement methods. The steady-state fit model is the core of the entire steady-state assessment. It integrates the stress deviation measurement value and the polarization degree volatility deviation measurement value to output a single steady-state fit index. In addition to the above formula, this model can also integrate multiple deviation indices into a comprehensive evaluation value through methods such as weighted averaging, fuzzy logic inference, or multivariate regression. The weight of each deviation term can be adjusted according to its impact on system stability.

[0122] Specifically, representing the attenuation coefficient Sensitivity coefficient , It can be determined through data fitting, experimental calibration, or empirical adjustment; for example, it can be optimized through offline simulation or online learning algorithms.

[0123] This application's solution dynamically establishes the expected values ​​of internal stress and polarization fluctuation rate of the PVC board using ultrasonic testing, and compares these with actual measured values ​​to obtain stress deviation and polarization fluctuation rate deviation metrics. These deviation metrics are then integrated into a steady-state fit model to generate a comprehensive steady-state fit index. This method can more comprehensively and accurately reflect the internal physical state and stability of the PVC board production process, rather than relying solely on external process parameters. In this way, the steady-state fit can more sensitively capture internal fluctuations that may affect the accuracy of defect detection, providing a more reliable and refined system state assessment for subsequent phase optimization models, thereby enabling the line frequency trigger phase adjustment of the line scan camera to more accurately adapt to actual production conditions.

[0124] The following is a concrete example. On a hollow plastic sheet production line, to accurately detect surface defects, it is necessary to assess the steady-state fit of the production process in real time. First, the system continuously monitors the current spatiotemporal distortion coefficient and defect risk coefficient of the plastic sheet. These coefficients are input into pre-trained stress expectation models and polarization fluctuation rate expectation models to dynamically calculate the expected values ​​of the ultrasonic internal stress and polarization fluctuation rate of the plastic sheet under the current operating conditions. For example, the stress expectation model may be a multinomial regression model based on historical data, while the polarization fluctuation rate expectation model may be a neural network-based model. Simultaneously, the actual ultrasonic internal stress and polarization fluctuation rate of the plastic sheet are measured in real time using online ultrasonic sensors and polarization light sensors. Then, the actual measured values ​​and expected values ​​are input into the stress deviation measurement model and polarization fluctuation rate deviation measurement model, respectively, to calculate the stress deviation metric value and polarization fluctuation rate deviation metric value. For example, if the actual stress differs significantly from the expected stress, the stress deviation metric value will increase accordingly. Finally, these deviation metrics are input into a steady-state fit model, which weights and combines these deviations to output a steady-state fit value between 0 and 1. For example, when both stress deviation and polarization fluctuation rate deviation are small, the steady-state fit value will be close to 1, indicating stable system operation; conversely, if the deviations are large, the steady-state fit value will decrease, suggesting potential instability in the system. This steady-state fit value is then passed to a phase optimization model to guide the adjustment of the line scan camera's line frequency trigger phase.

[0125] Through the above technical solution, this application can more accurately assess the steady-state condition of the hollow plastic sheet production process. Especially after considering key physical parameters such as internal stress and polarization fluctuation rate of the plastic sheet, the steady-state fit index can more realistically reflect the stability of the production line. This refined steady-state assessment provides a more reliable basis for subsequent line scan camera line frequency trigger phase optimization, thereby significantly improving the accuracy and robustness of surface defect detection, effectively avoiding false detections or missed detections caused by internal state fluctuations, and ultimately improving the overall performance of the detection system and production quality.

[0126] Preferably, the step of obtaining the optimized line frequency trigger phase is as follows:

[0127] Based on the apparent state coefficients, the transient response sensitivity and steady-state convergence factor are obtained and expressed as follows:

[0128]

[0129]

[0130] in, Indicates transient response sensitivity. Denotes the steady-state convergence factor. Indicates the basic transient response sensitivity. This represents the basic steady-state convergence factor. Indicates the gain attenuation coefficient and ;

[0131] Based on the steady-state fit, the current line frequency trigger phase of the line scan camera, and combined with the transient response sensitivity and steady-state convergence factor, an optimized line frequency trigger phase is output through a phase optimization model. The phase optimization model is expressed as follows:

[0132]

[0133] in, This indicates the optimized line frequency trigger phase. Indicates the current line frequency trigger phase of the line scan camera. Indicates the steady-state fit. This represents the target value for steady-state fit. This indicates the steady-state fit at historical moments. The instantaneous value, Indicates transient response sensitivity. This represents the steady-state convergence factor.

[0134] Transient response sensitivity This parameter measures a system's ability to respond quickly to changes in input (such as the deviation between steady-state fit and the target value). Higher transient response sensitivity generally means the system can adjust its output more quickly to reduce the deviation, but excessive sensitivity can lead to overshoot or oscillation. Steady-state convergence factor. This reflects the system's ability to eliminate long-term errors and eventually reach a steady state. It is usually associated with integral action and used to eliminate steady-state errors. These two parameters together determine the performance of the phase optimization model during dynamic adjustment. They can be dynamically adjusted according to the system's real-time state to adapt to different operating conditions. For example, in addition to adjustment through the apparent state coefficient, transient response sensitivity... It can also be adaptively adjusted based on the average absolute value of historical errors, or the steady-state convergence factor. It can be dynamically corrected based on the oscillation frequency output by the system. Basic transient response sensitivity. and the fundamental steady-state convergence factor These are the initial control parameters set under ideal or baseline operating conditions; they represent the default response characteristics of the system under no-disturbance or low-disturbance conditions. Gain attenuation coefficient. Used to adjust the apparent state coefficient Sensitivity to transient response and steady-state convergence factor The degree of influence of these parameters needs to be determined through data fitting, experimental calibration, or empirical adjustment to ensure that the system can achieve the best phase optimization effect under various operating conditions. For example, these basic parameters and attenuation coefficients can be optimized through offline simulation or online learning algorithms to adapt to the characteristic changes caused by different production batches or equipment wear.

[0135] This scheme aims to dynamically adjust the control parameters of the phase optimization model based on the apparent state of the plastic sheet. Apparent state coefficients. This reflects the degree of interference that the testing environment and the surface characteristics of the plastic sheet may cause to optical detection. When the apparent state coefficient... A higher value indicates a greater risk of optical interference, in which case it may be necessary to reduce the transient response sensitivity. and steady-state convergence factor This is to avoid unnecessary phase adjustments caused by the system's excessive sensitivity to noise or transient disturbances, thereby improving the system's robustness. Conversely, when the apparent state coefficients... When the parameters are low, these can be appropriately increased to accelerate system response and convergence. This adjustment mechanism ensures that the phase optimization model can adaptively optimize its performance according to actual operating conditions. The phase optimization model is used to calculate and output the optimized line frequency trigger phase. It combines the line frequency trigger phase of the current line scan camera. Steady-state fit Its target value The deviation, and through transient response sensitivity and steady-state convergence factor The deviation is processed by proportional and integral operations. Among them, The function maps the deviation value to a finite range to prevent excessive control input and provides nonlinear response characteristics, resulting in smooth control action when the deviation is small and enhanced control action when the deviation is large. The integral term is used to eliminate long-term steady-state errors and ensure steady-state fit. It can eventually converge to the target value. This model achieves precise, adaptive adjustment of the line frequency trigger phase through this variant of proportional-integral (PI) control. Besides using the hyperbolic tangent function, the sigmoid function or saturation function can be used to limit the output range of the bias term, or other forms of nonlinear mapping functions can be employed. Steady-state fit. This is an indicator that measures the overall stability and reliability of a detection system; a higher value indicates a more stable system. Steady-state fit target value. This represents the ideal steady-state level that the system hopes to achieve. The goal of the phase optimization model is to adjust the line frequency trigger phase to achieve a higher steady-state fit. as close as possible to or reach Steady-state fit at historical moments instantaneous value The calculation of the integral term reflects the trend of the system's steady-state changes over a period of time, helping to eliminate accumulated errors. For example, the steady-state fit target value... It can be set according to the specific requirements of the production line or the best historical operating data, or it can be dynamically adjusted according to the product type or the testing accuracy requirements.

[0136] The solution in this application introduces transient response sensitivity. and steady-state convergence factor And make it based on the apparent state coefficient of the plastic board. Adaptive adjustments are made to optimize the line scan camera's trigger phase. Specifically, the system first adjusts the apparent state coefficient based on the current environment and the surface characteristics of the plastic sheet. Dynamically calculate transient response sensitivity and steady-state convergence factor When the apparent state coefficient A higher level indicates an increased risk of optical interference, at which point the system will correspondingly reduce its speed. and The value of makes the phase optimization model fit the steady-state well. With target value The deviation response is smoother, avoiding over-adjustment caused by noise or transient disturbances. Conversely, when the apparent state coefficient... At lower levels, and The value will increase accordingly, enabling the system to adjust the line frequency trigger phase more quickly and accurately to cope with actual process fluctuations. Subsequently, these adaptively adjusted transient response sensitivities... and steady-state convergence factor The data is input into a phase optimization model, which triggers phase detection using the current line frequency of the line scan camera. Based on this, combined with steady-state fit Its target value The deviation is calculated using a control law that includes proportional and integral terms to determine the optimized line frequency trigger phase. The proportional term utilizes transient response sensitivity. It responds quickly to the current deviation, while the integral term utilizes the steady-state convergence factor. Historical biases are accumulated to eliminate long-standing steady-state errors. In this way, the phase optimization model can intelligently adjust its control strategy based on the apparent state of real-time operating conditions, ensuring that the line frequency trigger phase is always in the optimal state. This effectively compensates for the impact of various interference factors on image quality, improving the accuracy and stability of defect detection. This adaptive phase optimization mechanism enables the entire detection system to maintain excellent performance in complex and ever-changing production environments, significantly improving the reliability of defect detection.

[0137] As a specific implementation method, the apparent state coefficient is obtained. This can be achieved by integrating environmental sensors (such as PM2.5 / PM10 sensors) and surface characteristic detectors (such as dyne pen testers or electrostatic potentiometers). The data from these sensors is processed and then input into the apparent state model. Transient response sensitivity. and steady-state convergence factor The calculations can be performed on an embedded controller or industrial PC that receives the apparent state coefficients. As input, and dynamically adjusted according to preset formulas or lookup tables. and .For example, and The gain attenuation coefficient can be determined through the system's debugging results under standard operating conditions. and This can be obtained through regression analysis of experimental data on the system response under different apparent states. The core calculation of the phase optimization model is the optimized line frequency trigger phase. The generation of this can be implemented on a high-performance digital signal processor (DSP) or field-programmable gate array (FPGA) to ensure real-time performance and computational accuracy. The current line frequency trigger phase of the line scan camera. Steady-state fit can be obtained directly from the camera controller. and steady-state fit target value This is provided by the system's host control software. In the phase optimization model... Functions can be calculated quickly using lookup tables or floating-point units. Integral terms. Discretized numerical integration methods, such as the trapezoidal rule or Simpson's rule, can be used to integrate data over a past period. The values ​​are accumulated. The period of integration. The settings can be configured based on the system's dynamic characteristics and response requirements, for example, from several seconds to tens of seconds. The final calculated optimized line frequency trigger phase... It will be sent to the line scan camera controller to adjust the camera's trigger timing in real time, thereby ensuring that high-quality image data can be obtained under various operating conditions.

[0138] Through the above technical solution, this application can adaptively adjust the control parameters of the phase optimization model, i.e., the transient response sensitivity, according to the real-time apparent state of the plastic sheet. and steady-state convergence factor This dynamic adjustment mechanism enables the system to intelligently balance response speed and stability when facing different environmental disturbances and changes in the surface properties of the hollow plastic sheet. When the risk of optical interference is high, the system reduces response sensitivity to avoid overreacting to noise, thereby improving detection robustness; when the risk of optical interference is low, the system increases response speed to ensure rapid and accurate compensation for process fluctuations. This significantly improves the optimization accuracy and stability of the line frequency trigger phase of the line scan camera, effectively suppresses the decline in imaging quality caused by environmental or material changes, and thus improves the accuracy and reliability of hollow plastic sheet surface defect detection, reducing the occurrence of missed detections and false alarms.

[0139] A device for detecting surface defects in hollow plastic sheets, which uses the above-mentioned method for detecting surface defects in hollow plastic sheets to detect surface defects in hollow plastic sheets.

[0140] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0141] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for detecting surface defects in hollow plastic sheets, characterized in that, include: Based on the instantaneous linear velocity fluctuation rate of the plastic board, the dynamic warping of the board surface, and the amplitude of the conveyor belt, the spatiotemporal distortion coefficient is obtained through a spatiotemporal distortion model. Based on the ratio of PM1.0 / PM10 in airborne dust, the dyne value of the plastic board surface, and the electrostatic potential of the plastic board, the apparent state coefficient is obtained through the apparent state model. Based on the extruder die pressure fluctuation value, material moisture content and maximum difference in plate surface temperature, the defect risk coefficient is obtained through a defect risk model. Based on the spatiotemporal distortion coefficient and defect risk coefficient, the steady-state fit is obtained through a steady-state fit model, which is based on the internal stress and polarization fluctuation rate of ultrasonic waves in plastic sheet. Based on the steady-state fit, apparent state coefficient, and the current line frequency trigger phase of the line scan camera, the optimized line frequency trigger phase is output through the phase optimization model.

2. The method for detecting surface defects in hollow plastic sheets according to claim 1, characterized in that, The steps to obtain the optimized line frequency trigger phase are as follows: Based on the apparent state coefficients, the transient response sensitivity and steady-state convergence factor are obtained and expressed as follows: ； ； in, Indicates transient response sensitivity. Denotes the steady-state convergence factor. Indicates the basic transient response sensitivity. This represents the basic steady-state convergence factor. Indicates the gain attenuation coefficient and ; Based on the steady-state fit, the current line frequency trigger phase of the line scan camera, and combined with the transient response sensitivity and steady-state convergence factor, an optimized line frequency trigger phase is output through a phase optimization model. The phase optimization model is expressed as follows: ； in, This indicates the optimized line frequency trigger phase. Indicates the current line frequency trigger phase of the line scan camera. Indicates the steady-state fit. This represents the target value for steady-state fit. This indicates the steady-state fit at historical moments. The instantaneous value, Indicates transient response sensitivity. This represents the steady-state convergence factor.

3. The method for detecting surface defects in hollow plastic sheets according to claim 2, characterized in that, The steps to obtain the steady-state fit are as follows: Based on the spatiotemporal distortion coefficient and the defect risk coefficient, and combined with the ultrasonic internal stress and polarization fluctuation rate of the plastic board, the expected values ​​of the ultrasonic internal stress and polarization fluctuation rate of the plastic board are obtained. Based on the expected values ​​of the internal stress and polarization fluctuation rate of the ultrasonic wave in the plastic board, and combined with the corresponding internal stress and polarization fluctuation rate of the ultrasonic wave in the plastic board, the stress deviation metric and polarization fluctuation rate deviation metric are obtained. Based on the stress deviation metric and the polarization degree fluctuation rate deviation metric, the steady-state fit is obtained through a steady-state fit model, which is expressed as follows: ； in, Indicates the steady-state fit. This represents a measure of stress deviation. This represents a measure of polarization degree volatility deviation. This represents the product of the stress deviation metric and the polarization degree fluctuation rate deviation metric. Indicates the attenuation coefficient and The Furthermore, the higher the value, the more stable the detection system.

4. The method for detecting surface defects in hollow plastic sheets according to claim 3, characterized in that, The method for obtaining the defect risk coefficient is as follows: The pressure fluctuation value of the extruder die head, the material moisture content, and the maximum difference in plate surface temperature are compared with the corresponding reference values ​​to obtain the pressure fluctuation index, the material moisture content index, and the temperature fluctuation index. Based on the pressure fluctuation index, material moisture content index, and temperature fluctuation index, the defect risk coefficient is obtained through a defect risk model, which is expressed as follows: ； in, Indicates the defect risk coefficient. This indicates a pressure fluctuation index. Indicates the moisture content index of the material. Indicates the temperature fluctuation index. Indicates the weighted index and The Furthermore, the larger the value, the greater the precursor defects in the process.

5. The method for detecting surface defects in hollow plastic sheets according to claim 3, characterized in that, The apparent state coefficients are obtained as follows: The ratio of PM1.0 / PM10 in airborne dust and the electrostatic potential of plastic boards are compared with the corresponding reference values ​​to obtain the dust impact index and electrostatic potential index. The dyne value of the plastic board surface is inversely proportional to the dyne value reference value to obtain the dyne value index; Based on the dust impact index, dyne value index, and electrostatic potential index, apparent state coefficients are obtained through an apparent state model, which is expressed as follows: ； in, Represents the apparent state coefficient. Indicates the dust impact index. Indicates the Dyne index, Indicates the electrostatic potential index. Represents the weight coefficient and The Furthermore, the larger the value, the easier it is to generate optical interference during detection.

6. The method for detecting surface defects in hollow plastic sheets according to claim 3, characterized in that, The steps to obtain the spatiotemporal distortion coefficients are as follows: The instantaneous linear velocity fluctuation rate of the plastic board and the amplitude of the conveyor belt are compared with the corresponding reference values ​​to obtain the velocity fluctuation rate index and the amplitude index. The curvature radius index is obtained by inversely proportionalizing the dynamic warp of the board surface with the reference value of the curvature radius. Based on the velocity volatility index, radius of curvature index, and amplitude index, the spatiotemporal distortion coefficients are obtained through a spatiotemporal distortion model, which is expressed as follows: ； in, Represents the spatiotemporal distortion coefficient. Indicates the velocity volatility index. Represents the radius of curvature exponent. The amplitude index is represented by the following. Furthermore, the larger the value, the more severe the geometric distortion of the image.

7. The method for detecting surface defects in hollow plastic sheets according to claim 3, characterized in that, The method for obtaining the expected values ​​of internal stress and polarization fluctuation rate of the plastic sheet via ultrasonic testing is as follows: Based on the spatiotemporal distortion coefficient and defect risk coefficient, and combined with the internal stress of the PVC board during ultrasonic testing, the expected value of the internal stress of the PVC board during ultrasonic testing is obtained through a stress expectation model, which is expressed as follows: ； in, This represents the expected value of the internal stress of the plastic sheet under ultrasonic testing. This represents the internal stress of the plastic sheet under ideal ultrasonic conditions. Represents the spatiotemporal distortion coefficient. Indicates the defect risk coefficient. Represents the sensitivity coefficient and ; Based on the spatiotemporal distortion coefficient and the defect risk coefficient, and combined with the polarization degree volatility, the expected value of the polarization degree volatility is obtained through the polarization degree volatility expectation model, which is expressed as follows: ； in, This represents the expected value of the polarization degree volatility. This represents the polarization degree fluctuation rate under ideal operating conditions. Represents the spatiotemporal distortion coefficient. Indicates the defect risk coefficient. Represents the sensitivity coefficient and .

8. The method for detecting surface defects in hollow plastic sheets according to claim 3, characterized in that, The methods for obtaining stress deviation metrics and polarization degree fluctuation rate deviation metrics are as follows: Based on the expected value of the internal stress of the plastic sheet under ultrasonic testing and the internal stress of the plastic sheet under ultrasonic testing, the stress deviation metric value is obtained through a stress deviation metric model, which is expressed as follows: ； in, This represents a measure of stress deviation. This indicates the internal stress of the plastic sheet during ultrasonic testing. This represents the expected value of the internal stress of the plastic sheet under ultrasonic testing. Indicates a decimal excluding zero and whose unit is the same as the decimal. Consistent; Based on the expected value and the actual polarization volatility, a polarization volatility deviation metric is obtained using a polarization volatility deviation metric model, which is expressed as follows: ； in, This represents a measure of polarization degree volatility deviation. This represents the expected value of the polarization degree volatility. Indicates the polarization degree fluctuation rate. Indicates protection against zero decimals.

9. A device for detecting surface defects in hollow plastic sheets, characterized in that, The surface defects of hollow plastic boards are detected using the method for detecting surface defects of hollow plastic boards according to any one of claims 1-8.

Images