Method and system for controlling layered cutting path of carbon fiber prepreg

Abstract

The invention discloses a carbon fiber prepreg layered cutting path control method and system, and relates to the technical field of carbon fiber prepreg cutting. The method comprises the following steps: acquiring geometric perception information and motion physical information in real time; carrying out anomaly analysis on the geometric perception information, and / or carrying out anomaly analysis on the motion physical information; correcting the geometric perception information according to an analysis result; and / or generating motion physical sign information according to the motion physical information analysis result; and adjusting the cutting of the carbon fiber prepreg according to the corrected geometric perception information and / or motion physical sign information. According to the method and system provided by the invention, the geometric perception information and the motion physical information in the cutting process are acquired in real time, and the anomaly analysis is performed, so that the problems of sensor data drift and / or physical state anomaly and the like can be found and identified in time, and the geometric perception information is corrected according to the anomaly analysis result; and / or motion physical sign information is generated to provide reliable data for subsequent cutting control.

Description

Technical Field

[0001] This application relates to the field of carbon fiber prepreg cutting technology, and more specifically, to a method and system for controlling the layered cutting path of carbon fiber prepreg. Background Technology

[0002] In modern industrial production, the precise cutting of carbon fiber prepreg is a crucial step in determining the quality and performance of the final product. To ensure cutting accuracy and efficiency, the factory has introduced a highly automated cutting system, the core of which is a path control optimization method based on machine learning. During the cutting process of carbon fiber prepreg, the high-speed friction and shearing action between the tool and the material generates a large number of fine particles. Due to the extremely small size of these particles and their electrostatic adsorption properties, a very small number of these particles escape the exhaust fan and remain suspended in the air around the cutting machine.

[0003] As production continues, fine carbon fiber and resin dust floating in the air gradually and slowly deposits on various surfaces inside the cutting system. For example, the lens surface of an optical sensor may be covered with a thin layer of dust, and similar deposits may appear on the transmitting or receiving windows of a laser displacement sensor. This deposition does not immediately cause the sensor to fail completely, nor does it trigger the system's preset "sensor malfunction" alarm. However, it is enough to have a subtle but continuous effect on the sensor readings.

[0004] When the path control optimization method receives sensor input data with slight deviations caused by dust deposition, it treats it as a genuine change in physical state. Because this method was trained with input data precisely corresponding to actual physical states, it wasn't trained to identify or distinguish these "pseudo-changes" caused by sensor contamination. Therefore, based on these contaminated inputs, the method logically infers the current tool state or material properties and generates an "optimal" cutting path accordingly.

[0005] Ultimately, this "optimization" based on flawed premises leads the system to generate a suboptimal cutting path. Although the system continues to optimize, its optimization objective has deviated from actual physical requirements. The resulting cutting results exhibit a subtle, gradual decline in quality. This chain reaction, caused by sensor data drift due to environmental factors, leading to misjudgments and the generation of suboptimal paths in the path control optimization method, creates systemic risks to the quality of produced components. This not only wastes materials and time but also poses a potential threat to the reliability of the final product, and the root cause of the problem is difficult to diagnose and correct directly by the existing system.

[0006] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0007] This application discloses a method for controlling the cutting path of carbon fiber prepreg in layers, which aims to solve the problem that sensor data drift caused by environmental factors during the cutting process of carbon fiber prepreg, which leads to misjudgment and generation of suboptimal paths by the path control optimization method, ultimately affecting the cutting quality and product reliability.

[0008] The technical solution of this application is as follows: In a first aspect, this application discloses a method for controlling the layered cutting path of carbon fiber prepreg, comprising the following steps: Real-time acquisition of geometric sensing information and motion physics information during the cutting process of carbon fiber prepreg; Anomaly analysis is performed on geometric perception information, and / or anomaly analysis is performed on motion physics information; Based on the anomaly analysis results of the geometric sensing information, the geometric sensing information is corrected; and / or, Based on the anomaly analysis results of the motion physics information, motion physics marker information is generated; Based on the corrected geometric perception information and / or the motion physical sign information, the cutting path, tool depth and cutting speed during the carbon fiber prepreg cutting process are adjusted to ensure the accuracy and quality of the cutting edge. The geometric sensing information includes the thickness value of the carbon fiber prepreg; the motion physics information includes the torque value of the servo motor used to cut the carbon fiber prepreg.

[0009] Through this technical solution, this application can monitor multi-source data in real time during the cutting process and perform intelligent analysis and correction on abnormal data, thereby effectively avoiding cutting path optimization deviations caused by sensor data drift or physical abnormalities, significantly improving cutting accuracy and product quality, and solving the systemic hidden dangers caused by environmental factors that are difficult to diagnose and correct in the prior art.

[0010] Furthermore, in some preferred embodiments, the step of performing anomaly analysis on the geometrically sensed information includes: Based on the preset normal correspondence between the reference thickness value and the reference torque value, calculate the first torque value of the servo motor that matches the thickness value; Calculate the abnormal deviation between this torque value and the first torque value; The method, based on the anomaly analysis results of the geometric sensing information, includes the following specific steps for correcting the geometric sensing information: Based on the abnormal deviation, calculate the confidence level of the thickness value and compare the confidence level with the preset confidence threshold. When the confidence level is lower than the preset confidence threshold, a first thickness value matching the torque value is calculated based on the normal correspondence. The thickness value of the carbon fiber prepreg is obtained by fusing the thickness value and the first thickness value according to the preset rules. When the confidence level is greater than or equal to the confidence threshold, the thickness value is defined as the corrected carbon fiber prepreg thickness value.

[0011] Through this technical solution, this application can utilize the normal correspondence between thickness and torque values, and through confidence assessment and fusion calculation, accurately correct thickness values ​​with abnormal deviations, effectively improve the accuracy of geometric perception information, and provide a more reliable data foundation for subsequent cutting control.

[0012] Based on the above, this application further proposes that the steps for anomaly analysis of geometric perception information and anomaly analysis of motion physics information specifically include: The thickness sliding average value of the thickness value is calculated in real time over a set forward shift time, and the torque sliding average value of the torque value is calculated in real time over a first set forward shift time. Based on the thickness sliding average and the torque sliding average, determine the type of anomaly in the carbon fiber prepreg cutting scenario; The abnormality types include: abnormal drift in the thickness value of carbon fiber prepreg; local hardening of carbon fiber prepreg; and abnormal drift in the thickness value of carbon fiber prepreg, with local hardening of the carbon fiber prepreg.

[0013] Through this technical solution, this application can determine the abnormal type of the cutting scenario by calculating the thickness sliding average value and torque sliding average value in real time and combining the two. This enables early identification and classification of various potential abnormal situations, providing clear guidance for subsequent targeted correction and control, and improving the system's ability to cope with complex abnormal scenarios.

[0014] More specifically, in some implementations, the step of determining the type of anomaly in a carbon fiber prepreg cutting scenario based on the thickness sliding average and the torque sliding average specifically includes: The thickness sliding change rate of the thickness sliding average is calculated based on the thickness sliding average and the preset reference thickness average. If the thickness slip change rate exceeding the preset change threshold lasts for more than the second set time, it indicates that the thickness value of the carbon fiber prepreg has a continuous unidirectional drift. The torque sliding average value is calculated and compared with the preset torque threshold. If the duration of the torque sliding average value exceeding the preset torque threshold is less than the third set time, it indicates that there is an instantaneous torque abnormality in the servo motor. When only the thickness value of carbon fiber prepreg shows a continuous unidirectional drift, the anomaly type is determined to be: the thickness value of carbon fiber prepreg shows a measurement anomaly drift. When only the servo motor exhibits instantaneous torque anomaly, the anomaly type is determined to be: localized hardening of the carbon fiber prepreg. If the thickness value of carbon fiber prepreg has a continuous unidirectional drift and the servo motor has an instantaneous torque abnormality, it is determined whether the two occur at the same time. When the two occur in different time periods, the anomaly type is determined to be: the thickness value of the carbon fiber prepreg shows abnormal drift and the carbon fiber prepreg exhibits local hardening.

[0015] Through this technical solution, this application can accurately distinguish and determine various abnormal types such as abnormal drift and local hardening in carbon fiber prepreg thickness measurement by refining the analysis of thickness slip change rate and torque slip average value, combined with duration judgment, thus providing a more detailed abnormal scenario identification capability for subsequent precise correction and control.

[0016] Preferably, the anomaly type is: the thickness value of the carbon fiber prepreg shows abnormal drift during measurement; The step of correcting the geometric perception information based on the anomaly analysis results specifically includes: Based on the first normal correspondence between the preset reference thickness value and the reference torque value, calculate the second torque value of the servo motor that matches the thickness value; Calculate the first abnormal deviation between the torque value and the second torque value, and calculate the first confidence level of the thickness value based on the first abnormal deviation; Compare the first confidence level with a preset first confidence threshold; When the first confidence level is lower than the first confidence threshold, the thickness compensation value is calculated based on the thickness sliding average and the preset reference thickness average. The thickness value is corrected based on the thickness compensation value to obtain the corrected thickness value of the carbon fiber prepreg.

[0017] This technical solution addresses the issue of abnormal drift in the thickness measurement of carbon fiber prepreg. By introducing a first normal correspondence and a first confidence level assessment, combined with a thickness compensation value, the drift data is effectively corrected, ensuring the accuracy of the thickness information and thus avoiding cutting path deviations caused by measurement errors.

[0018] In one implementation, when the anomaly type is: the thickness value of the carbon fiber prepreg shows abnormal drift during measurement, and the carbon fiber prepreg exhibits localized hardening, the step of correcting the geometric sensing information based on the anomaly analysis results specifically includes: Based on the second normal correspondence between the preset reference thickness value and the reference torque value, calculate the third torque value of the servo motor that matches the thickness value; Calculate the second abnormal deviation between the fourth torque value outside the time period when the servo motor experiences instantaneous torque abnormality and the third torque value, and calculate the second confidence level of the thickness value based on the second abnormal deviation; Compare this second confidence level with a preset second confidence threshold; When the second confidence level is lower than the second confidence threshold, the first thickness compensation value is calculated based on the thickness sliding average and the preset reference thickness average. The thickness value is corrected based on the first thickness compensation value to obtain the corrected carbon fiber prepreg thickness value. The steps for generating motion physical marker information based on the anomaly analysis results of motion physical information specifically include: Acquire the spatial position information of the tool used to cut carbon fiber prepreg during the period when the servo motor experiences an instantaneous torque anomaly; Based on the fifth torque value during the period when the servo motor experiences a momentary torque anomaly; The degree of hardening in the corresponding locally hardened region of the carbon fiber prepreg is calculated based on the fifth torque value. The spatial location information and the hardening degree value are fused into the motion physical indicator information.

[0019] Through this technical solution, this application can address complex anomaly scenarios where thickness measurement drift and local hardening coexist. By introducing a second normal correspondence and a second confidence level assessment, and combining the thickness compensation value to correct the thickness, it can generate motion physical marker information containing the spatial location and degree of hardening of the local hardened area. This enables comprehensive processing of multiple anomalies, provides more comprehensive and refined guidance for cutting control, and significantly improves the system's ability to cope with complex anomalies.

[0020] As an optional approach, the specific steps for anomaly analysis of motion physics information include: The first torque sliding average value of the servo motor torque value is calculated in real time over a fourth set time period. Calculate the magnitude of the first torque sliding average value and the preset first torque threshold value. If the first torque sliding average value exceeds the first torque threshold value and the duration is less than the fifth preset time, it indicates that the servo motor has an instantaneous torque abnormality. The step of generating motion physical marker information based on the anomaly analysis results of motion physical information specifically includes: Acquire the first spatial position information of the tool used to cut carbon fiber prepreg during the period when the servo motor experiences an instantaneous torque anomaly; Obtain the sixth torque value during the period when the servo motor experiences a momentary torque anomaly; Based on the sixth torque value, calculate the first degree of hardening in the corresponding locally hardened region of the carbon fiber prepreg. The first spatial location information and the first hardening degree value are fused into the motion physical marker information.

[0021] Through this technical solution, this application can detect instantaneous torque anomalies by real-time monitoring of the sliding average value of the servo motor torque value and combining it with the duration. This generates motion physical marker information containing the spatial location and degree of hardening of the local hardening area, enabling independent identification and marking of local hardening problems. This provides key information for local adjustment of the cutting path and effectively avoids the decline in cutting quality caused by local hardening.

[0022] In one embodiment, the motion physics information also includes vibration data of the tool used to cut the carbon fiber prepreg; Before the step of calculating the first torque value of the servo motor that matches the preset reference thickness value and reference torque value, the following steps are included: Obtain the batch identifier of the current batch of carbon fiber prepreg, and search for the normal correspondence that matches the current batch from the preset normal correspondence database based on the batch identifier. If no matching normal correspondence is found, the torque value is high-pass filtered to obtain the high-frequency component of the torque, and the root mean square value of the high-frequency component of the torque is calculated. Perform a real-time fast Fourier transform on the vibration data to extract the energy values ​​within a set frequency range; Compare the root mean square value with a preset root mean square threshold, and compare the energy value with a preset energy threshold; Based on the comparison results and preset modification rules, the normal correspondence is corrected to generate a corrected third normal correspondence with batch identifier, which is used to correct the thickness value of carbon fiber prepreg during the subsequent cutting process of the corresponding batch of carbon fiber prepreg.

[0023] Through this technical solution, this application can achieve adaptive adjustment of the characteristics of different batches of prepreg by introducing batch identification management and dynamic correction of normal correspondence, combined with multi-dimensional analysis of high-frequency torque components and tool vibration data. This effectively solves the problem of cutting parameter mismatch caused by batch differences and significantly improves the robustness and adaptability of the system.

[0024] Furthermore, after performing a high-pass filter on the torque value to obtain the high-frequency component of the torque, and calculating the root mean square value of the high-frequency component of the torque, the process also includes: Calculate the kurtosis coefficient of the high-frequency component of torque, and calculate the allowable torque deviation based on the magnitude of the kurtosis coefficient; The steps for correcting geometric perception information based on the anomaly analysis results specifically include: Based on the thickness value and the preset calculation rules, calculate the expected torque center value for the current batch; Calculate the normal torque range based on the allowable torque deviation and the expected torque center value; When the torque value is outside the normal range, the second thickness value of the carbon fiber prepreg that matches the torque value is calculated according to the corrected third normal correspondence. The thickness value of the carbon fiber prepreg is obtained by fusing the thickness value and the second thickness value according to the preset rules. When the torque value is within the normal range, the thickness value is defined as the corrected carbon fiber prepreg thickness value.

[0025] Through this technical solution, this application can dynamically calculate the normal range of torque by introducing the kurtosis coefficient and torque tolerance of the high-frequency component of torque, and correct the thickness value on this basis. This enables more refined identification and processing of transient anomalies during the cutting process, further improving the accuracy and real-time performance of thickness correction and ensuring the stability of the cutting process.

[0026] Secondly, this application also discloses a carbon fiber prepreg layer cutting path control system, comprising: The acquisition module is used to acquire geometric sensing information and motion physics information in real time during the cutting process of carbon fiber prepreg; The analysis module is used to perform anomaly analysis on geometric perception information and / or anomaly analysis on motion physics information; The correction module is used to correct the geometric sensing information based on the results of anomaly analysis; and / or, The generation module is used to generate motion physics marker information based on the results of anomaly analysis of motion physics information; The control module is used to adjust the cutting path, tool depth and cutting speed during the cutting process of carbon fiber prepreg based on the corrected geometric perception information and / or the motion physical sign information, so as to ensure the accuracy and quality of the cutting edge. The geometric sensing information includes the thickness value of the carbon fiber prepreg; the motion physics information includes the torque value of the servo motor used to cut the carbon fiber prepreg.

[0027] This application provides a system-level solution through this technical solution. Through modular design, it realizes real-time acquisition, anomaly analysis, data correction, and marker information generation of multi-source information during the cutting process of carbon fiber prepreg, and finally performs precise control. This effectively solves the cutting quality problems caused by sensor data drift and physical state abnormalities in the prior art, and improves the intelligence level and reliability of the automated cutting system. Beneficial effects

[0028] The carbon fiber prepreg layer cutting path control method disclosed in this application acquires geometric sensing information and motion physics information in real time during the cutting process, and performs anomaly analysis on this information to promptly detect and identify problems such as sensor data drift or abnormal physical states. Based on this, the method can correct the geometric sensing information according to the anomaly analysis results, or generate motion physics marker information, thereby providing an accurate and reliable data foundation for subsequent cutting control. Finally, the system precisely controls the cutting of the carbon fiber prepreg based on the corrected geometric sensing information and / or motion physics marker information.

[0029] Compared to existing technologies, the method in this application effectively solves the problem of sensor data drift caused by environmental factors (such as dust deposition), avoiding the drawbacks of path control optimization methods that generate suboptimal cutting paths based on erroneous premises. Through intelligent analysis and correction of multi-source information, this method can ensure the accuracy and optimization of the cutting path, thereby significantly improving the cutting quality of carbon fiber prepreg, reducing material waste, and increasing the reliability of the final product. This method overcomes the shortcomings of existing technologies in directly diagnosing and correcting systemic hazards caused by environmental factors, achieving refined and intelligent control of the cutting process, and has significant technological progress and practical value. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of a method for controlling the layered cutting path of carbon fiber prepreg provided in this application.

[0031] Figure 2 This application provides a schematic diagram of a carbon fiber prepreg layer cutting path control system.

[0032] Figure 2 In the diagram: 1 is the acquisition module; 2 is the analysis module; 3 is the calibration module; 4 is the generation module; and 5 is the control module. Detailed Implementation

[0033] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0034] See Figure 1 This application proposes a method for controlling the layered cutting path of carbon fiber prepreg, including: S10. Real-time acquisition of geometric sensing information and motion physics information during the cutting process of carbon fiber prepreg; S20. Perform anomaly analysis on geometric perception information, and / or perform anomaly analysis on motion physics information; S30. Based on the results of the anomaly analysis of the geometric perception information, the geometric perception information is corrected; And / or, S40. Generate motion physical indicator information based on the results of anomaly analysis of motion physical information; S50. Based on the corrected geometric perception information and / or the motion physical marker information, adjust the cutting path, cutting depth and cutting speed during the carbon fiber prepreg cutting process to ensure the accuracy and quality of the cutting edge. The geometric sensing information includes the thickness value of the carbon fiber prepreg; the motion physics information includes the torque value of the servo motor used to cut the carbon fiber prepreg.

[0035] "Geometric sensing information" refers to real-time data related to the geometric properties of carbon fiber prepreg, directly acquired by sensors during the prepreg cutting process. For example, this could be the prepreg thickness measured by a laser displacement sensor or optical sensor, or the surface profile data of the prepreg acquired by a vision system. This information forms the basis for assessing the physical state of the prepreg.

[0036] "Kinematic physical information" refers to physical quantities related to the motion state and force conditions of the cutting equipment during the cutting process. For example, this could be real-time position and speed data of the cutting tool obtained through a servo motor encoder, or the torque value of the servo motor measured by a torque sensor. This information reflects the dynamic response and material resistance during the cutting process.

[0037] "Anomaly analysis" refers to the processing of acquired geometric sensing information and / or motion physics information to identify deviations, drifts, or abrupt changes that deviate from normal operating conditions. This analysis aims to distinguish normal process fluctuations from genuine anomalies caused by sensor contamination, material defects, or equipment malfunctions.

[0038] "Correction" refers to the process of correcting geometric perception information by using algorithms or models after anomalies have been identified, making it closer to the true physical values. The purpose of correction is to eliminate or reduce the negative impact of anomalous data on subsequent control decisions.

[0039] "Motion physics marker information" refers to indicative data generated based on the type and severity of the anomaly after identifying anomalies in motion physics information. This marker information can be used for subsequent adjustments to cutting control strategies, such as indicating the location and degree of hardening in locally hardened areas, so that the cutting system can adjust cutting parameters accordingly.

[0040] The core of the carbon fiber prepreg layer cutting path control method proposed in this application lies in the intelligent perception, analysis and correction of multi-source information during the cutting process, so as to achieve precise control of the cutting path.

[0041] During the cutting process, it is necessary to acquire geometric sensing information and motion physics information of the carbon fiber prepreg in real time. Geometric sensing information can be acquired in various ways. For example, a laser displacement sensor can be used to continuously measure the thickness of the carbon fiber prepreg. After the laser beam emitted by the sensor is reflected from the surface of the prepreg, the thickness value is calculated by measuring the time or phase difference of the reflected light. Another method is to use a contact probe to directly measure the thickness of the prepreg through mechanical contact. Motion physics information can also be acquired in various ways. For example, a torque sensor mounted on a servo motor can monitor the torque value of the servo motor in real time, and this sensor converts the mechanical torque into an electrical signal output. In addition, the speed and position information of the servo motor can be obtained through an encoder, thereby calculating the motion state of the cutting tool.

[0042] Secondly, anomaly analysis is performed on the acquired geometric sensing information and / or on the motion physics information. For anomaly analysis of geometric sensing information, a preset normal range or threshold can be set. When the real-time acquired geometric sensing information (e.g., thickness value) exceeds this range, it is considered an anomaly. For example, an upper and lower limit for thickness can be set; when the measured thickness value is higher than the upper limit or lower than the lower limit, an anomaly is considered to exist. A similar method can be used for anomaly analysis of motion physics information. For example, a normal operating range for servo motor torque can be set; when the real-time monitored torque value continuously exceeds this range, an anomaly is considered to exist. Furthermore, anomaly patterns can be identified by comparing the trend of current data with historical data or by comparing it with a preset model.

[0043] Secondly, based on the anomaly analysis results of the geometric sensing information, the geometric sensing information is corrected; and / or, based on the anomaly analysis results of the motion physics information, motion physics marker information is generated. When the geometric sensing information is determined to be abnormal, it needs to be corrected. For example, if the thickness value measured by the laser displacement sensor shows a continuous drift due to dust deposition, a correction model can be established that predicts the true thickness value based on historical data and environmental parameters, and the predicted value is used to correct the current abnormal value. Another correction method is to introduce redundant sensors; when one sensor malfunctions, the readings of other normal sensors can be referenced for correction. When the motion physics information is determined to be abnormal, such as a sudden change in servo motor torque, this may indicate local hardening of the prepreg. In this case, motion physics marker information can be generated based on the degree and duration of the torque anomaly, combined with the real-time position information of the tool. This marker information can include the spatial coordinates of the locally hardened area and a quantified value of the hardening degree, used to guide subsequent cutting control.

[0044] Finally, the cutting of carbon fiber prepreg is controlled based on the corrected geometric sensing information and / or motion physics information. After acquiring the corrected geometric sensing information, the cutting system can adjust the cutting path, cutter depth, and cutting speed in real time based on this more accurate geometric data to ensure the precision and quality of the cut edges. For example, if the corrected thickness value indicates localized thinning of the prepreg, the system can correspondingly reduce the cutter depth to avoid cutting through or damaging the underlying material. Once motion physics information is generated, the cutting system can use this information for more precise control. For example, if the motion physics information indicates localized hardening in a certain area, the system can pre-adjust the cutting parameters of the tool, such as increasing the cutting force, decreasing the cutting speed, or changing the tool type, to smoothly pass through the hardened area and avoid tool damage or a decrease in cutting quality.

[0045] The carbon fiber prepreg layer cutting path control method proposed in this application can effectively solve the problem of reduced cutting accuracy caused by sensor data drift and material property changes in complex production environments by acquiring geometric sensing information and motion physical information in real time and performing anomaly analysis and correction.

[0046] Specifically, in the traditional carbon fiber prepreg cutting process, when the sensor's output geometric sensing information (such as thickness value) experiences a continuous, slight drift due to environmental contamination (such as dust deposition), existing path control systems often fail to recognize this "pseudo-anomaly" and instead treat it as a genuine change in physical state. For example, if the thickness sensor continuously reports a thickness slightly larger than the actual value due to dust deposition, the traditional system will adjust the cutting depth accordingly, resulting in insufficient actual cutting depth, potentially causing incomplete cutting or edge burrs. Similarly, when there are locally hardened areas inside the prepreg, the traditional system may fail to detect them in advance, causing the tool to suddenly encounter resistance when cutting into the hardened area, resulting in a momentary overload of the servo motor torque, which may even damage the tool or cause the cutting path to deviate.

[0047] This application introduces an anomaly analysis mechanism for geometric sensing information and motion physics information, effectively distinguishing between "pseudo-anomalies" caused by sensor contamination and "true anomalies" caused by material properties or equipment status. For example, by performing anomaly analysis on geometric sensing information, persistent drift in sensor readings can be identified, and the geometric sensing information can be corrected based on the analysis results to obtain a more accurate prepreg thickness value. This corrected information can guide the cutting system to make more accurate adjustments to the cutting depth, avoiding cutting defects caused by sensor errors.

[0048] Furthermore, by performing anomaly analysis on motion physics information, this application can promptly detect physical anomalies during the cutting process, such as instantaneous overload of the servo motor torque, and generate motion physics marker information accordingly. For example, when an instantaneous torque overload is detected, the system can combine the real-time position of the tool to accurately mark the location and degree of hardening of the locally hardened area in the prepreg. This motion physics marker information is then used to guide cutting control, enabling the cutting system to pre-adjust cutting parameters (such as reducing cutting speed, increasing cutting force, or switching cutting modes) before entering the hardened area, thereby smoothly passing through the hardened area, avoiding tool damage, and ensuring cutting quality.

[0049] In summary, this application constructs a more robust and accurate method for controlling the layered cutting path of carbon fiber prepreg by intelligently sensing, analyzing, and correcting multi-source information. Compared with existing technologies, this application can effectively address uncertainties in the production environment, significantly improve the cutting accuracy and efficiency of carbon fiber prepreg, reduce scrap rate, and extend tool life, thereby providing a more reliable and efficient solution for the manufacturing of high-performance composite materials.

[0050] The above-mentioned method for controlling the layered cutting path of carbon fiber prepreg includes the following steps for anomaly analysis using geometric sensing information: Based on the preset normal correspondence between the reference thickness value and the reference torque value, calculate the first torque value of the servo motor that matches the thickness value; Calculate the abnormal deviation between the torque value and the first torque value; The step of correcting the geometric perception information based on the anomaly analysis results specifically includes: Based on the abnormal deviation, calculate the confidence level of the thickness value and compare the confidence level with a preset confidence threshold. When the confidence level is lower than the confidence level threshold, a first thickness value matching the torque value is calculated based on the normal correspondence. The thickness value of the carbon fiber prepreg is calculated by fusing the thickness value and the first thickness value according to the preset rules. When the confidence level is greater than or equal to the confidence level threshold, the thickness value is defined as the corrected carbon fiber prepreg thickness value.

[0051] Specifically, geometric sensing information is defined as the thickness value of the carbon fiber prepreg, which can be acquired in real time by various sensors (such as laser thickness gauges, ultrasonic sensors, etc.). Motion physics information is defined as the torque value of the servo motor used to cut the carbon fiber prepreg, which is usually provided directly by the servo motor driver.

[0052] When performing anomaly analysis on geometric sensing information, the system first calculates the first torque value of the servo motor that matches the currently acquired thickness value, based on a preset normal correspondence between reference thickness and reference torque values. This normal correspondence can be a mapping model or lookup table between thickness and torque established through a large amount of experimental data under normal cutting conditions. Then, the system calculates the abnormal deviation between the currently acquired torque value and this first torque value. This abnormal deviation reflects the difference between the actual torque and the expected torque at the current thickness, and is an important basis for determining whether the thickness measurement is abnormal.

[0053] Furthermore, when correcting the geometric sensing information based on the anomaly analysis results, the confidence level of the current thickness value is calculated based on the anomaly deviation obtained above. The confidence level can be understood as a quantitative indicator of the reliability of the thickness value. Then, this confidence level is compared with a preset confidence threshold. When the calculated confidence level is lower than the preset confidence threshold, it indicates that the current thickness value may have a measurement anomaly and needs correction. At this time, the system will, conversely, calculate a first thickness value that matches the current torque value based on the normal correspondence. This means that when the thickness measurement is unreliable, the torque data is trusted instead, and a more reliable thickness value is calculated using the torque data. Finally, the original thickness value and the first thickness value calculated from the torque are fused according to preset rules, such as using weighted average, Kalman filtering, etc., to obtain the corrected carbon fiber prepreg thickness value. The purpose of the fusion calculation is to comprehensively utilize the two types of information to improve the accuracy of the thickness value. When the calculated confidence level is greater than or equal to the preset confidence threshold, the current thickness value is considered reliable and does not require correction; it is directly defined as the corrected carbon fiber prepreg thickness value.

[0054] This application's solution achieves comprehensive sensing of the prepreg's state during cutting by using the thickness of the carbon fiber prepreg as geometric sensing information and the torque of the servo motor as motion physical information. When an anomaly occurs in the thickness value, comparing it with the expected torque value calculated based on a normal correspondence can identify potential anomalies in the thickness measurement. The introduction of the torque value as an auxiliary judgment allows the system to physically verify the accuracy of the geometric sensing. When the confidence level of the thickness value is low, the system effectively avoids the impact of single-sensor measurement errors on cutting control by using the torque value to inversely calculate the thickness and then fusing the results, thereby improving the accuracy and robustness of the correction. This correction mechanism based on multi-source information cross-validation and fusion can more reliably obtain the true thickness of the prepreg, providing accurate input for subsequent cutting path control.

[0055] This application further proposes a more refined anomaly analysis method, which aims to accurately identify and classify specific anomaly types in a scenario through trend analysis of key parameters.

[0056] Specifically, in the above-mentioned method for controlling the layered cutting path of carbon fiber prepreg, the steps of performing anomaly analysis on geometric perception information and anomaly analysis on motion physics information specifically include: The thickness sliding average value of the thickness value is calculated in real time over a set forward shift time, and the torque sliding average value of the torque value is calculated in real time over a first set forward shift time. Based on the thickness sliding average value and the torque sliding average value, determine the abnormal type of the carbon fiber prepreg cutting scenario; The abnormality types include: abnormal drift in the thickness value of carbon fiber prepreg; local hardening of carbon fiber prepreg; and abnormal drift in the thickness value of carbon fiber prepreg, with local hardening of the carbon fiber prepreg.

[0057] In this context, geometric sensing information refers to sensing data related to the physical dimensions or shape of the carbon fiber prepreg; specifically, in this embodiment, it refers to the thickness value of the carbon fiber prepreg. This thickness value can be acquired in real time using various non-contact or contact sensors, such as laser rangefinders, ultrasonic sensors, or contact thickness gauges. Motion physics information refers to physical quantities related to the motion state or force conditions of the cutting equipment; specifically, in this embodiment, it refers to the torque value of the servo motor used to cut the carbon fiber prepreg. This torque value is typically monitored and output in real time by sensors within the servo motor driver.

[0058] Furthermore, to more accurately reflect the changing trends of the carbon fiber prepreg thickness and the servo motor torque, this application introduces the concept of a moving average. Specifically, "real-time calculation of the thickness moving average over a set time period forward" means calculating the average of all thickness values ​​within a set time period prior to the current sampling moment. This set time period can be adjusted according to the actual application scenario and data sampling frequency, with the aim of smoothing the thickness data, filtering out high-frequency noise, and thus better capturing the long-term or medium-term changing trends of the thickness value. Similarly, "real-time calculation of the torque moving average over a first set time period forward" means calculating the average of all torque values ​​within a first set time period prior to the current moment. This first set time period can also be set according to actual conditions, with the aim of smoothing the torque data and identifying continuous changes or anomalies in torque.

[0059] Therefore, by combining the thickness sliding average and the torque sliding average, the anomaly type in carbon fiber prepreg cutting scenarios can be determined. Specifically, this application defines three main anomaly types: The first type of anomaly is "abnormal drift in the thickness value of carbon fiber prepreg". This usually refers to a persistent, non-physical deviation in the thickness sensor reading, such as sensor aging, calibration errors, or environmental factors causing the measured value to deviate from the true value.

[0060] The second type of anomaly is "localized hardening of the carbon fiber prepreg." This typically refers to an abnormal increase in hardness in localized areas of the carbon fiber prepreg, which may be caused by uneven resin curing, impurity contamination, or material defects. Localized hardening leads to increased cutting resistance, which is reflected in the torque value of the servo motor.

[0061] The third type of anomaly is "abnormal drift in the thickness measurement of carbon fiber prepreg, and localized hardening of the carbon fiber prepreg." This indicates that both of the above anomalies have occurred, which may require more complex correction and control strategies.

[0062] This application's solution effectively overcomes the limitations of traditional instantaneous data analysis in anomaly detection by introducing thickness and torque moving averages. Instantaneous data is susceptible to random noise and short-term fluctuations, leading to misjudgments. By calculating moving averages, the data can be smoothed, and instantaneous noise can be filtered out, thus more accurately reflecting the true trends in thickness and torque values. For example, the thickness moving average can reveal persistent drift in thickness measurement, while the torque moving average can effectively capture short-term, persistent torque anomalies caused by changes in material properties (such as local hardening) during the cutting process. This trend-based analysis method enables the system to distinguish between instantaneous disturbances and actual process anomalies, improving the robustness of anomaly detection. Furthermore, by combining these two moving averages, the system can classify anomaly scenarios more finely, such as distinguishing between simple measurement drift, simple local material hardening, and complex situations involving both. This classification capability is the foundation for subsequent accurate correction and control, as different types of anomalies often require different response strategies.

[0063] Through the above technical solution, this application can significantly improve the accuracy and specificity of anomaly detection during the cutting process of carbon fiber prepreg. Specifically, by performing moving average processing on geometric sensing information (such as thickness values) and motion physics information (such as servo motor torque values), instantaneous noise and random fluctuations are effectively filtered out, enabling the system to more stably and reliably identify continuous abnormal trends. More importantly, this application can meticulously classify abnormal scenarios into various types, such as measurement anomaly drift, local hardening, or both, based on a comprehensive analysis of the thickness moving average and torque moving average. This refined anomaly classification allows for more targeted subsequent correction and control strategies, avoiding a "one-size-fits-all" approach, thereby improving the accuracy and efficiency of cutting path control, reducing scrap rate, and helping to extend tool life.

[0064] In some embodiments described above in this application, anomaly types in carbon fiber prepreg cutting scenarios are determined based on thickness sliding average and torque sliding average. Specifically, the steps for determining the anomaly type in carbon fiber prepreg cutting scenarios can be further refined as follows.

[0065] Based on the aforementioned thickness sliding average and torque sliding average, the specific steps for determining the anomaly type of the carbon fiber prepreg cutting scenario include: The thickness sliding change rate of the thickness sliding average is calculated based on the thickness sliding average and the preset reference thickness average. If the thickness slip change rate exceeding the preset change threshold lasts for more than the second set time, it indicates that the thickness value of the carbon fiber prepreg has a continuous unidirectional drift. The torque sliding average value is calculated and compared with the preset torque threshold. If the duration of the torque sliding average value exceeding the preset torque threshold is less than the third set time, it indicates that there is an instantaneous torque abnormality in the servo motor. When only the thickness value of carbon fiber prepreg shows a continuous unidirectional drift, the anomaly type is determined to be: the thickness value of carbon fiber prepreg shows a measurement anomaly drift. When only the servo motor exhibits instantaneous torque anomaly, the anomaly type is determined to be: localized hardening of the carbon fiber prepreg. If the thickness value of carbon fiber prepreg has a continuous unidirectional drift and the servo motor has an instantaneous torque abnormality, it is determined whether the two occur at the same time. When the two occur in different time periods, the abnormality type is determined to be: the thickness value of the carbon fiber prepreg shows abnormal drift and the carbon fiber prepreg exhibits local hardening.

[0066] Specifically, the thickness sliding average refers to the result of averaging the thickness values ​​of carbon fiber prepreg over a set time period during the cutting process, reflecting the short-term trend of the thickness value. The baseline thickness average can be understood as the expected or standard average thickness of the carbon fiber prepreg under normal cutting conditions. By comparing the thickness sliding average with the baseline thickness average, the thickness sliding rate of change can be calculated, which quantifies the deviation between the current thickness and the normal thickness. The preset change threshold is a pre-set critical value used to determine whether the thickness sliding rate of change has reached a level that warrants attention. The second set time is a time length used to confirm whether the anomaly in the thickness sliding rate of change is persistent, avoiding misjudgment due to instantaneous fluctuations. When the thickness sliding rate of change continuously exceeds the preset change threshold and continuously exceeds the second set time, it indicates that the thickness value of the carbon fiber prepreg may have a continuous unidirectional drift, which is usually caused by sensor drift or slow, continuous thickness changes in the material itself.

[0067] The torque sliding average is the result of averaging the torque values ​​of the servo motor used to cut carbon fiber prepreg over a first set time period during the cutting process. It reflects the short-term trend of the servo motor load. The preset torque threshold is a pre-set upper limit for torque, used to determine whether the servo motor is under excessive load. The third set time is a shorter time period used to determine whether the torque anomaly is instantaneous. When the torque sliding average exceeds the preset torque threshold but the duration is less than the third set time, it indicates that the servo motor has an instantaneous torque anomaly. This is usually due to the tool encountering a locally hardened area during the cutting process, causing a sudden increase in cutting resistance.

[0068] This application's solution, through refined analysis of the thickness sliding average and torque sliding average, can accurately distinguish different types of cutting anomalies. Specifically, by calculating the thickness sliding change rate and combining it with its duration, continuous unidirectional drift in the thickness value of carbon fiber prepreg can be effectively identified, which helps distinguish between sensor malfunctions or batch differences in materials. Simultaneously, by monitoring whether the torque sliding average instantaneously exceeds a preset threshold, locally hardened areas in the carbon fiber prepreg can be sensitively detected. Furthermore, by determining whether continuous unidirectional drift in the thickness value and instantaneous torque anomalies of the servo motor occur within the same time period, this application can more accurately classify abnormal scenarios, thereby avoiding confusion between two different types of anomalies and providing an accurate basis for subsequent correction and control.

[0069] The aforementioned technical solution enables precise classification of abnormal scenarios during the cutting process of carbon fiber prepreg, including identifying abnormal drift in the measurement of carbon fiber prepreg thickness, localized hardening of the carbon fiber prepreg, and situations where both exist simultaneously but may occur at different times. This detailed ability to identify anomaly types allows the system to adopt more precise response strategies for different types of anomalies. For example, for measurement drift, sensor calibration parameters can be adjusted; for localized hardening, the cutting path or cutting parameters can be adjusted, thereby significantly improving the stability and accuracy of the cutting process, reducing scrap rate, and optimizing material utilization efficiency.

[0070] This application further proposes a geometric perception information correction method for abnormal drift in the thickness value of carbon fiber prepreg. By introducing confidence judgment and thickness compensation mechanism, it aims to achieve more accurate and reliable thickness value correction.

[0071] According to the above-mentioned carbon fiber prepreg layer cutting path control method, when the anomaly type is: the thickness value of the carbon fiber prepreg shows abnormal drift in measurement; The step of correcting the geometric perception information based on the anomaly analysis results specifically includes: Based on the first normal correspondence between the preset reference thickness value and the reference torque value, calculate the second torque value of the servo motor that matches the thickness value; Calculate the first abnormal deviation between the torque value and the second torque value, and calculate the first confidence level of the thickness value based on the first abnormal deviation; Compare the first confidence level with a preset first confidence threshold. When the first confidence level is lower than the first confidence threshold, the thickness compensation value is calculated based on the thickness sliding average and the preset benchmark thickness average. The thickness value is corrected based on the thickness compensation value to obtain the corrected carbon fiber prepreg thickness value.

[0072] Specifically, when the system determines that the anomaly is a measurement anomaly drift in the thickness value of the carbon fiber prepreg, in order to correct the geometric sensing information (i.e., the thickness value of the carbon fiber prepreg), it first utilizes a preset normal correspondence between the reference thickness value and the reference torque value. This first normal correspondence can be understood as an empirical or model-based mapping relationship between the thickness of the carbon fiber prepreg and the torque of the servo motor under normal cutting conditions. Based on the currently acquired thickness value, a second torque value of the servo motor that matches this thickness value can be obtained through table lookup or calculation. This second torque value represents the torque performance that the servo motor should have under normal operating conditions at the current thickness.

[0073] Subsequently, the real-time acquired servo motor torque value is compared with the calculated second torque value to determine the first abnormal deviation between the two. This first abnormal deviation reflects the degree of difference between the actual torque and the theoretical normal torque. Based on this first abnormal deviation, the first confidence level of the current thickness value can be further calculated. The first confidence level is an indicator that measures the reliability or accuracy of the current thickness value; the higher the value, the more reliable the current thickness value.

[0074] Next, the calculated first confidence level is compared with a preset first confidence threshold. This threshold is a pre-defined standard used to determine whether the reliability of the current thickness value reaches an acceptable level.

[0075] When the first confidence level is determined to be lower than the preset first confidence threshold, it indicates that the reliability of the current thickness value is low and there may be a large measurement error, requiring correction. In this case, a thickness compensation value is calculated based on the thickness sliding average and a preset baseline thickness average. The thickness sliding average smooths out instantaneous measurement fluctuations, providing a more stable thickness trend, while the baseline thickness average represents the expected normal thickness of the carbon fiber prepreg in that batch or region. The difference between the two allows for estimation of the amount of compensation needed for the current thickness value. Finally, the current thickness value is corrected based on the calculated thickness compensation value, resulting in a more accurate and reliable corrected carbon fiber prepreg thickness value.

[0076] When the first confidence level is greater than or equal to the preset first confidence level threshold, it indicates that the reliability of the current thickness value is high and no compensation correction is required. At this time, the thickness value can be directly defined as the corrected carbon fiber prepreg thickness value.

[0077] This application's solution introduces a first anomaly deviation and a first confidence level based on the normal correspondence between torque and thickness values, enabling a more refined assessment of the reliability of current thickness measurements. Specifically, when the system detects an abnormal drift in the thickness value of the carbon fiber prepreg, it no longer blindly performs corrections. Instead, it first indirectly judges the accuracy of the thickness measurement by comparing the difference (first anomaly deviation) between the actual torque value and the theoretical normal torque value (second torque value) calculated based on the current thickness value. This indirect judgment mechanism utilizes the physical correlation between thickness and torque, making the assessment of thickness anomalies more comprehensive and reliable. Furthermore, by converting the first anomaly deviation into a first confidence level and comparing it with a preset first confidence level threshold, this solution can intelligently determine whether thickness compensation is needed. Only when the confidence level of the thickness value is lower than the preset threshold is a thickness compensation mechanism based on the thickness sliding average and the reference thickness average activated, thereby avoiding unnecessary corrections and ensuring the targetedness and effectiveness of the correction. This step-by-step judgment and conditional correction strategy makes the thickness value correction process more intelligent and robust.

[0078] This application further proposes a step for correcting geometric sensing information when the anomaly type is a measurement anomaly drift in the thickness value of carbon fiber prepreg and local hardening exists in the carbon fiber prepreg. The specific steps include: Based on the preset second normal correspondence between the reference thickness value and the reference torque value, calculate the third torque value of the servo motor that matches the thickness value; Calculate the second abnormal deviation between the fourth torque value outside the time period when the servo motor experiences instantaneous torque abnormality and the third torque value, and calculate the second confidence level of the thickness value based on the second abnormal deviation; Compare the second confidence level with a preset second confidence threshold. When the second confidence level is lower than the second confidence threshold, a first thickness compensation value is calculated based on the thickness sliding average and the preset reference thickness average. The thickness value is corrected according to the first thickness compensation value to obtain the corrected carbon fiber prepreg thickness value; The step of generating motion physical marker information based on the anomaly analysis results of motion physical information specifically includes: Acquire the spatial position information of the tool used to cut carbon fiber prepreg during the period when the servo motor experiences an instantaneous torque anomaly; Based on the fifth torque value during the period when the servo motor experiences a momentary torque anomaly; Calculate the degree of hardening in the corresponding local hardening region of the carbon fiber prepreg based on the fifth torque value; The spatial location information and the hardening degree value are fused into the motion physical indicator information.

[0079] Specifically, the second normal correspondence refers to the preset correlation between the thickness value of the carbon fiber prepreg and the torque value of the servo motor during the carbon fiber prepreg cutting process, when both are in a normal state. This relationship can be established based on historical data or experimental results and optimized or adjusted for situations where thickness measurement anomalies and local hardening coexist. The third torque value is the theoretical torque value calculated based on the currently acquired thickness value using this second normal correspondence. The fourth torque value refers to the actual measured torque value of the servo motor during the period when the servo motor does not experience instantaneous torque anomalies. Its purpose is to eliminate the influence of local hardening on the torque value when correcting the thickness value, thereby more accurately assessing the thickness measurement drift. The second anomaly deviation is the difference between the fourth and third torque values, used to quantify the degree of thickness measurement drift. The second confidence level is a reliability index of the thickness value calculated based on the second anomaly deviation. The preset second confidence level threshold is a critical value used to determine whether the thickness value needs correction. When the second confidence level is lower than the preset second confidence level threshold, it indicates that there is a significant measurement anomaly in the thickness value, requiring correction. The first thickness compensation value is a compensation amount calculated based on the thickness sliding average and the preset reference thickness average, used to correct the thickness value. Its purpose is to adjust the drifted thickness value to be closer to the true value.

[0080] In the step of generating motion physical marker information, the period of instantaneous torque anomaly of the servo motor refers to the time period during which local hardening occurs, as determined by the aforementioned anomaly analysis. The spatial position information of the cutting tool refers to the precise position coordinates of the cutting tool used to cut the carbon fiber prepreg within the cutting system during this time period, such as the X, Y, and Z axis coordinates. The fifth torque value is the actual torque value of the servo motor during the instantaneous torque anomaly, which is typically significantly higher than the normal torque. The hardening degree value of the locally hardened area is calculated based on the fifth torque value and is an indicator used to quantify the degree of local hardening of the carbon fiber prepreg; it can be calculated, for example, through torque peak value, torque duration, or torque integral. The motion physical marker information integrates the spatial position information of the cutting tool with the hardening degree value of the locally hardened area to form a comprehensive marker information that accurately indicates the location and severity of the locally hardened area.

[0081] This application's solution addresses the precise control problem when carbon fiber prepreg exhibits both thickness measurement anomaly drift and localized hardening by collaboratively analyzing and processing geometric sensing information and kinematic physics information. Specifically, when correcting geometric sensing information, a second anomaly deviation is calculated between the fourth torque value and the third torque value outside the time period of instantaneous torque anomaly in the servo motor. Based on this, a second confidence level for the thickness value is calculated, effectively eliminating the interference of localized hardening on the judgment of thickness measurement anomalies. When the confidence level of the thickness value is low, a first thickness compensation value is calculated based on the thickness sliding average and a preset benchmark thickness average, and the thickness value is corrected, thereby ensuring that thickness measurement drift outside the locally hardened area can be accurately identified and corrected. Simultaneously, regarding the localized hardening phenomenon, this application acquires the spatial position information of the tool and the fifth torque value during the period of instantaneous torque anomaly in the servo motor, and calculates the degree of hardening in the locally hardened area based on the fifth torque value, fusing the two to generate kinematic physics marker information. Thus, this solution can accurately locate the locally hardened area and quantify its degree of hardening, providing crucial and targeted information for subsequent cutting path adjustments.

[0082] In some preferred embodiments, it is assumed that during the cutting process of carbon fiber prepreg, the system analyzes the thickness sliding average and torque sliding average to determine that the current anomaly type is a measurement anomaly drift in the thickness value of the carbon fiber prepreg, and that the carbon fiber prepreg exhibits localized hardening. Specifically, the system detects a continuous unidirectional drift in the thickness value, and within a specific time period, the torque value of the servo motor experiences a momentary abnormal increase, indicating the presence of localized hardening.

[0083] At this point, to correct the geometric sensing information, the system first calculates the third torque value of the servo motor matching the current thickness value based on the preset second normal correspondence between the reference thickness value and the reference torque value. Subsequently, the system pays special attention to torque data outside the time period of instantaneous torque anomaly in the servo motor, extracts the fourth torque value, and calculates the second abnormal deviation between the fourth and third torque values. Based on this deviation, the system calculates the second confidence level of the current thickness value. If this second confidence level is lower than the preset second confidence level threshold, it indicates that thickness measurement drift does exist and needs correction. At this point, the system calculates a first thickness compensation value based on the thickness sliding average and the preset reference thickness average, and uses this compensation value to correct the original thickness value, obtaining the corrected carbon fiber prepreg thickness value.

[0084] Simultaneously, to generate motion physics information, the system precisely records the spatial position of the tool used to cut the carbon fiber prepreg during the instantaneous torque anomaly of the servo motor. For example, it records the tool's coordinates on the X, Y, and Z axes. Furthermore, the system acquires a fifth torque value during this instantaneous torque anomaly and calculates the hardening degree of the corresponding locally hardened area in the carbon fiber prepreg based on this value; for example, a higher peak torque indicates a greater hardening degree. Finally, this spatial position information and hardening degree value are fused to form motion physics information, which clearly indicates the specific location and severity of the hardened area. In this way, the cutting control system can simultaneously obtain accurate thickness information and detailed information about the locally hardened area, enabling more intelligent adjustments to cutting parameters and paths, such as reducing cutting speed or increasing cutting force in the hardened area to ensure cutting quality and protect the tool.

[0085] In some embodiments of this application described above, the step of performing anomaly analysis on motion physics information specifically includes: The first torque sliding average value of the servo motor torque value is calculated in real time over a fourth set time period. Calculate the magnitude of the first torque sliding average value and the preset first torque threshold value. If the first torque sliding average value exceeds the first torque threshold value and the duration is less than the fifth preset time, it indicates that the servo motor has an instantaneous torque abnormality. The step of generating motion physical marker information based on the anomaly analysis results of motion physical information specifically includes: Acquire the first spatial position information of the tool used to cut carbon fiber prepreg during the period when the servo motor experiences an instantaneous torque anomaly; Obtain the sixth torque value during the period when the servo motor experiences a momentary torque anomaly; Based on the sixth torque value, calculate the first degree of hardening value of the corresponding locally hardened region in the carbon fiber prepreg. The first spatial location information and the first hardening degree value are fused into the motion physical marker information.

[0086] Specifically, motion physics information can be understood as physical quantities directly related to the motion state of the cutting equipment during the cutting process of carbon fiber prepreg. In this embodiment, this motion physics information specifically refers to the torque value of the servo motor used to drive the cutting tool. The torque value of the servo motor can directly reflect the magnitude of the resistance encountered by the cutting tool during the cutting process, thereby indirectly reflecting the physical properties of the carbon fiber prepreg, such as its hardness or local defects.

[0087] When performing anomaly analysis on motion physics information, the first step is to calculate the first torque sliding average value of the servo motor's torque over a fourth predetermined time period. This fourth predetermined time period is a pre-set time window used to smooth out instantaneous fluctuations in torque values, thus better capturing persistent anomalies. The first torque sliding average value is obtained by averaging the torque values ​​within this time window and reflects the average load of the servo motor over a recent period.

[0088] Subsequently, the calculated first torque sliding average value is compared with a preset first torque threshold. When the first torque sliding average value exceeds the first torque threshold, and the duration of this exceedance is less than a fifth preset time, it can be determined that the servo motor has an instantaneous torque anomaly. The fifth preset time is used to distinguish between instantaneous anomalies and persistent anomalies; a shorter duration indicates that the anomaly is sudden and may be caused by factors such as local hardening. The first torque threshold is an upper limit determined empirically or experimentally based on the torque value range under normal cutting conditions.

[0089] When a transient torque anomaly is detected in the servo motor, motion physical marker information needs to be generated based on the anomaly analysis results. Specifically, firstly, the first spatial position information of the tool used to cut the carbon fiber prepreg is obtained during the period when the transient torque anomaly occurs. This first spatial position information records the precise position of the tool at the time of the anomaly, which helps to locate the abnormal area on the carbon fiber prepreg. Simultaneously, a sixth torque value is obtained during the period when the transient torque anomaly occurs. The sixth torque value is the actual torque value at the time of the anomaly, and it contains information about the intensity of the anomaly.

[0090] Furthermore, based on the sixth torque value, a first degree of hardening value for the corresponding locally hardened region in the carbon fiber prepreg can be calculated. The degree of hardening value can be calculated based on the mapping relationship between the torque value and the prepreg hardness; for example, the higher the torque value, the greater the degree of hardening. Finally, the first spatial location information and the first degree of hardening value are fused to form the motion physical marker information. This motion physical marker information integrates the location and intensity information of anomalies, providing crucial guidance for subsequent cutting path control.

[0091] The solution proposed in this application effectively filters out random noise during the cutting process by real-time monitoring of the servo motor's torque value and performing a sliding average process. This allows for more accurate detection of instantaneous torque anomalies caused by factors such as localized hardening of the carbon fiber prepreg. By setting a fourth preset time, it can be ensured that the sliding average reflects the torque trend within a certain time window, rather than an instantaneous value at a single moment. When the first torque sliding average exceeds a preset first torque threshold and the duration is less than a fifth preset time, it indicates that the cutting tool has encountered abnormal resistance in a localized area. However, this resistance is not continuous but occurs instantaneously, which is consistent with the characteristics of localized hardening of the carbon fiber prepreg.

[0092] Once an instantaneous torque anomaly is detected, the system immediately records the tool's first spatial position information and the actual sixth torque value at the time of the anomaly. The first spatial position information precisely indicates the physical location of the anomaly, while the sixth torque value provides a quantitative assessment of the anomaly's intensity. By converting the sixth torque value into a first hardening degree value, the severity of localized hardening can be intuitively evaluated. Finally, this position and hardness information is fused into motion physical marker information, enabling the cutting control system to obtain precise, quantitative information about the locally hardened area of ​​the prepreg, thus providing a basis for subsequent cutting path adjustments.

[0093] This application further proposes a method to dynamically correct the normal correspondence by introducing batch identification, vibration data, and high-pass filtering analysis of torque values, thereby improving the accuracy and adaptability of thickness correction.

[0094] In the above-mentioned method for controlling the layered cutting path of carbon fiber prepreg, the motion physical information also includes vibration data of the tool used to cut the carbon fiber prepreg.

[0095] Before the step of calculating the first torque value of the servo motor matching the thickness value based on the preset normal correspondence between the reference thickness value and the reference torque value, the method further includes: Obtain the batch identifier of the current batch of carbon fiber prepreg, and search for the matching normal correspondence in the preset normal correspondence database based on the batch identifier.

[0096] The batch identifier is information used to uniquely identify a specific production batch of carbon fiber prepreg, such as production date and batch number. The preset normal correspondence database is a database storing the normal correspondence between thickness and torque values ​​for different batches or types of carbon fiber prepreg. By searching, one can attempt to obtain a correspondence that better matches the characteristics of the current batch of prepreg, serving as an initial or reference normal correspondence.

[0097] If no matching normal correspondence is found, the torque value is high-pass filtered to obtain the high-frequency component of the torque, and the root mean square value of the high-frequency component of the torque is calculated.

[0098] High-pass filtering aims to remove low-frequency components (usually related to the average load of the cut) from the torque signal, retaining the high-frequency components. These high-frequency components are often related to transient impacts, vibrations, or localized material inhomogeneities during the cutting process. The root mean square (RMS) value is an indicator of signal strength or energy; the RMS value of the high-frequency components of the torque can reflect the severity of high-frequency vibrations or impacts during the cutting process.

[0099] The vibration data is subjected to real-time fast Fourier transform to extract energy values ​​within a set frequency range.

[0100] Fast Fourier Transform (FFT) is an algorithm that converts time-domain signals into frequency-domain signals. FFT can be used to analyze the energy distribution of tool vibration data at different frequencies. By extracting energy values ​​within a set frequency range, it's possible to focus on specific vibration modes related to the cutting process, tool condition, or material properties, such as vibration frequencies caused by tool wear or material hard spots.

[0101] The root mean square value is compared with a preset root mean square threshold, and the energy value is compared with a preset energy threshold.

[0102] By comparing the calculated root mean square value of the high-frequency torque component and the vibration energy value with preset thresholds, it is possible to determine whether the current cutting state is abnormal or deviates from the standard state. For example, a root mean square value or energy value exceeding the threshold may indicate increased material hardness, accelerated tool wear, or mismatched cutting parameters.

[0103] Based on the comparison results and preset modification rules, the normal correspondence is corrected to generate a corrected third normal correspondence carrying a batch identifier, which is used to correct the thickness value of carbon fiber prepreg during the subsequent cutting process of the corresponding batch of carbon fiber prepreg.

[0104] The preset modification rules are based on experience or models and guide how to adjust the normal correspondence based on the comparison results of the root mean square (RMS) value and energy value. For example, if both the RMS and energy values ​​are high, it may mean that the current batch of material is generally harder. In this case, the normal correspondence curve can be shifted upward, meaning that the expected torque value should be higher for the same thickness. The corrected third normal correspondence is stored in association with the current batch identifier so that it can be directly called when cutting prepregs from the same batch, avoiding repeated correction processes.

[0105] This application's solution effectively addresses the problem that traditional methods' preset normal correspondences cannot adapt to the differences in characteristics between different batches of carbon fiber prepreg by introducing batch identification, vibration data, and high-pass filtering analysis of torque values. Specifically, when the batch identification of the current batch is obtained, the system first attempts to find the existing normal correspondence for that batch from historical data. This mechanism allows the system to utilize past experience to quickly load the best-matching correspondence for known batches, thereby improving the initial accuracy of the correction. When no matching normal correspondence is found, the system no longer blindly uses general preset values, but dynamically adjusts by monitoring and analyzing the motion physics information during the cutting process in real time. High-pass filtering of the torque value and calculation of its high-frequency component root mean square value can effectively capture the high-frequency response caused by local material inhomogeneity, tool micro-wear, or transient impact during the cutting process. These high-frequency responses are important indicators reflecting the actual cutting resistance characteristics of the material. At the same time, performing fast Fourier transform on the tool vibration data and extracting energy values ​​within a specific frequency range can reveal the dynamic characteristics of the cutting process from another dimension, such as the interaction strength between the tool and the material, and the presence of abnormal resonance. By comprehensively comparing these real-time acquired root mean square (RMS) and energy values ​​with preset thresholds, the system can more accurately determine the deviation between the actual cutting characteristics of the current batch of carbon fiber prepreg and the baseline state. Based on these judgments, the system can refine the normal correspondence between the baseline thickness value and the baseline torque value according to preset modification rules. This corrected correspondence can more accurately reflect the actual physical characteristics of the current batch of prepreg, thus providing a more reliable baseline for subsequent thickness anomaly analysis and correction.

[0106] This application further proposes that after performing high-pass filtering on the above torque value to obtain the high-frequency component of the torque, and calculating the root mean square value of the high-frequency component of the torque, the following steps are also included: Calculate the kurtosis coefficient of the high-frequency component of torque, and calculate the torque tolerance based on the magnitude of the kurtosis coefficient; The step of correcting the geometric perception information based on the anomaly analysis results specifically includes: Based on the thickness value and the preset calculation rules, calculate the expected torque center value for the current batch; Calculate the normal torque range based on the torque tolerance and the expected torque center value; When the torque value is outside the normal torque range, the second thickness value of the carbon fiber prepreg that matches the torque value is calculated according to the corrected third normal correspondence. The thickness value and the second thickness value are fused and calculated according to a preset rule to obtain the corrected carbon fiber prepreg thickness value; When the torque value is within the normal torque range, the thickness value is defined as the corrected carbon fiber prepreg thickness value.

[0107] Specifically, after high-pass filtering the torque value to obtain the high-frequency components of the torque and calculating their root mean square (RMS) values, the kurtosis coefficient of these high-frequency components can be further calculated. The kurtosis coefficient is a statistical measure of the distribution pattern of data, used to characterize the "sharpness" or "flatness" of the data distribution, as well as the thickness of the tails. By analyzing the kurtosis coefficient, the frequency and extreme nature of outliers in the high-frequency components of the torque can be assessed more accurately. Based on the magnitude of the kurtosis coefficient, the torque tolerance can be dynamically calculated. For example, a high kurtosis coefficient indicates the presence of more extreme outliers; in this case, the torque tolerance can be appropriately increased to avoid misjudging normal fluctuations. Conversely, a low kurtosis coefficient can tighten the torque tolerance, improving the sensitivity of anomaly detection.

[0108] When correcting the geometric perception information, the expected torque center value for the current batch is first calculated based on the currently acquired thickness value and preset calculation rules. This expected torque center value can be understood as the theoretical or expected average value of the servo motor torque at the current thickness value, and its calculation can be based on historical data, physical models, or machine learning models. Subsequently, the torque tolerance obtained from the above calculation is combined with the expected torque center value to determine a normal torque range. This normal torque range defines the range within which the servo motor torque is considered to fluctuate normally at the current thickness value.

[0109] In practical applications, when the real-time acquired torque value is outside the normal torque range, it indicates that the torque value may be abnormal, and the thickness value needs to be corrected. Specifically, a second thickness value of the carbon fiber prepreg that matches the abnormal torque value is calculated based on the corrected third normal correspondence. This second thickness value is derived from the torque value and is considered a more reliable thickness estimate. Then, according to a preset fusion rule, the original thickness value and the second thickness value are fused to obtain the corrected carbon fiber prepreg thickness value. The fusion rule may include methods such as weighted averaging and Kalman filtering, aiming to integrate the information from both to improve the accuracy of the thickness value. Conversely, when the real-time acquired torque value is within the normal torque range, it is considered to be within the normal fluctuation range and no correction is required. In this case, the original thickness value is directly defined as the corrected carbon fiber prepreg thickness value.

[0110] Through the above technical solution, this application can significantly improve the accuracy and robustness of detecting anomalies in carbon fiber prepreg thickness values. Compared to methods that rely solely on anomaly deviations and confidence levels, the introduction of mechanisms for kurtosis coefficients and dynamic torque normal ranges allows the system to better adapt to complex changes in torque signals during cutting, especially in the presence of non-Gaussian distributions or occasional extreme values. This effectively reduces overcorrection or undercorrection caused by misjudgments, thereby ensuring the accuracy of carbon fiber prepreg cutting path control and ultimately improving cutting quality and production efficiency.

[0111] See Figure 2 This application also discloses a carbon fiber prepreg layered cutting path control system, comprising: an acquisition module 1, an analysis module 2, a correction module 3, a generation module 4, and a control module 5. The acquisition module 1 is used to acquire geometric perception information and motion physics information in real time during the carbon fiber prepreg cutting process; the analysis module 2 is used to perform anomaly analysis on the geometric perception information and / or on the motion physics information; the correction module 3 is used to correct the geometric perception information based on the anomaly analysis results; and / or the generation module 4 is used to generate motion physics marker information based on the anomaly analysis results; the control module 5 is used to adjust the cutting path, cutter depth, and cutting speed during the carbon fiber prepreg cutting process based on the corrected geometric perception information and / or the motion physics marker information to ensure the accuracy and quality of the cutting edge. The geometric sensing information includes the thickness value of the carbon fiber prepreg; the motion physics information includes the torque value of the servo motor used to cut the carbon fiber prepreg.

[0112] The system acquires multi-source data in real time during the cutting process via module 1, and analysis module 2 intelligently identifies anomalies in this data. Subsequently, correction module 3 corrects the abnormal geometric sensing information, while generation module 4 generates corresponding flag information based on the anomalies in the motion physics information. Finally, control module 5 precisely adjusts and controls the cutting path of the carbon fiber prepreg based on this accurate information after correction and marking, thereby effectively addressing issues such as sensor data drift and local material anomalies, ensuring cutting quality and product reliability.

[0113] To achieve the aforementioned control method, the system proposed in this application utilizes multiple functional modules working collaboratively. Specifically, the acquisition module 1 can be configured to include multiple sensor interfaces, such as a laser displacement sensor for obtaining prepreg thickness values ​​and a servo motor torque sensor for obtaining torque values. As one implementation, the acquisition module 1 can be an integrated data acquisition unit that converts the analog signals output by the sensors into digital signals via an analog-to-digital converter and transmits them to the system for processing. In some embodiments, the acquisition module 1 can also be a software driver responsible for communicating with external sensor devices and reading data according to a preset sampling frequency and data format. The specific methods for acquiring geometric sensing information and motion physics information have been described in the above embodiments and will not be repeated here.

[0114] Analysis module 2 is used to perform anomaly analysis on geometric sensing information and / or on motion physics information. Specifically, analysis module 2 can be a software program running on a general-purpose processor, with a built-in preset anomaly detection algorithm. For example, this module can use a statistical threshold-based method to identify anomalies by comparing real-time data with a preset normal range. In another implementation, analysis module 2 can be a dedicated signal processing unit that performs real-time filtering and feature extraction on sensor data through hardware acceleration to quickly identify potential anomaly patterns.

[0115] The correction module 3 is used to correct the geometric sensing information based on the anomaly analysis results. As one implementation, the correction module 3 can be a software algorithm library containing various correction models, such as a linear correction model based on historical data fitting or a simple average filtering algorithm. When the analysis module 2 identifies anomalies in the geometric sensing information, the correction module 3 is activated and corrects the abnormal data according to a preset correction strategy. In some embodiments, the correction module 3 can also compensate for specific types of sensor drift by consulting a preset correction parameter table.

[0116] The generation module 4 is used to generate motion physics flag information based on the anomaly analysis results of the motion physics information. Specifically, the generation module 4 can be a software component that, after receiving the anomaly determination from the analysis module, generates flag data containing specific information according to the type and degree of the anomaly. For example, when a momentary overload of servo motor torque is detected, the generation module 4 can combine the real-time position data of the tool to generate flag information indicating the spatial coordinates and degree of hardening of the locally hardened area. In some embodiments, the generation module 4 may also simply output a Boolean flag indicating whether a certain type of motion physics anomaly exists.

[0117] Control module 5 is used to control the cutting of carbon fiber prepreg based on the corrected geometric sensing information and / or the motion physical marker information. In one implementation, control module 5 can be a programmable logic controller (PLC) pre-programmed with various cutting strategies and parameter adjustment rules. Upon receiving the corrected geometric sensing information or motion physical marker information, the PLC adjusts the cutting depth, cutting speed, or cutting force of the cutting machine in real time based on this information. In another implementation, the control module can be an embedded system-based controller that manages the cutting machine's motion axes through a real-time operating system (RTOS) and dynamically adjusts the cutting path based on input data.

[0118] The carbon fiber prepreg layered cutting path control system proposed in this application provides a systematic solution to the problems of sensor data drift caused by environmental pollution and the difficulty in identifying and responding to local material anomalies in existing technologies. Traditional cutting systems often lack the ability to perform real-time anomaly analysis and intelligent correction on multi-source data, resulting in their inability to distinguish between "pseudo-anomalies" and "true anomalies" when receiving biased sensor data, thus generating suboptimal cutting paths. This application constructs a system architecture capable of comprehensively sensing, intelligently analyzing, accurately correcting, and adaptively controlling geometric perception information and motion physics information by introducing acquisition, analysis, correction, generation, and control modules. Compared with existing technologies, the system of this application can effectively identify and correct geometric perception information drift caused by sensor contamination, and can also promptly detect and mark motion physics anomalies such as local material hardening, thereby ensuring the accuracy of the cutting path and the stability of the cutting quality, significantly improving the robustness and reliability of the carbon fiber prepreg cutting process.

[0119] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for controlling the layered cutting path of carbon fiber prepreg, characterized in that, include: Real-time acquisition of geometric sensing information and motion physics information during the cutting process of carbon fiber prepreg; Anomaly analysis is performed on geometric perception information, and / or anomaly analysis is performed on motion physics information; Based on the anomaly analysis results of the geometric sensing information, the geometric sensing information is corrected; and / or, Based on the anomaly analysis results of the motion physics information, motion physics marker information is generated; Based on the corrected geometric perception information and / or the motion physical marker information, the cutting path, tool depth and cutting speed during the carbon fiber prepreg cutting process are adjusted to ensure the accuracy and quality of the cutting edge. The geometric sensing information includes the thickness value of the carbon fiber prepreg; the motion physics information includes the torque value of the servo motor used to cut the carbon fiber prepreg.

2. The method for controlling the layered cutting path of carbon fiber prepreg according to claim 1, characterized in that, The steps for anomaly analysis of geometric perception information include: Based on the preset normal correspondence between the reference thickness value and the reference torque value, calculate the first torque value of the servo motor that matches the thickness value; Calculate the abnormal deviation between the torque value and the first torque value; The step of correcting the geometric perception information based on the anomaly analysis results specifically includes: Based on the abnormal deviation, calculate the confidence level of the thickness value and compare the confidence level with a preset confidence threshold. When the confidence level is lower than the preset confidence threshold, a first thickness value matching the torque value is calculated based on the normal correspondence. The thickness value of the carbon fiber prepreg is calculated by fusing the thickness value and the first thickness value according to the preset rules. When the confidence level is greater than or equal to the confidence level threshold, the thickness value is defined as the corrected carbon fiber prepreg thickness value.

3. The method for controlling the layered cutting path of carbon fiber prepreg according to claim 1, characterized in that, The specific steps for anomaly analysis of geometric perception information and anomaly analysis of motion physics information include: The thickness sliding average value of the thickness value is calculated in real time over a set forward shift time, and the torque sliding average value of the torque value is calculated in real time over a first set forward shift time. Based on the thickness sliding average value and the torque sliding average value, determine the abnormal type of the carbon fiber prepreg cutting scenario; The abnormality types include: abnormal drift in the thickness value of carbon fiber prepreg; local hardening of carbon fiber prepreg; and abnormal drift in the thickness value of carbon fiber prepreg, with local hardening of the carbon fiber prepreg.

4. The method for controlling the layered cutting path of carbon fiber prepreg according to claim 3, characterized in that, The steps for determining the anomaly type of carbon fiber prepreg cutting scenario based on the thickness sliding average and the torque sliding average specifically include: The thickness sliding change rate of the thickness sliding average is calculated based on the thickness sliding average and the preset reference thickness average. If the thickness slip change rate exceeding the preset change threshold lasts for more than the second set time, it indicates that the thickness value of the carbon fiber prepreg has a continuous unidirectional drift. The torque sliding average value is calculated and compared with the preset torque threshold. If the duration of the torque sliding average value exceeding the preset torque threshold is less than the third set time, it indicates that there is an instantaneous torque abnormality in the servo motor. When only the thickness value of carbon fiber prepreg shows a continuous unidirectional drift, the anomaly type is determined to be: the thickness value of carbon fiber prepreg shows a measurement anomaly drift. When only the servo motor exhibits instantaneous torque anomaly, the anomaly type is determined to be: localized hardening of the carbon fiber prepreg. If the thickness value of carbon fiber prepreg has a continuous unidirectional drift and the servo motor has an instantaneous torque abnormality, it is determined whether the two occur at the same time. When the two occur in different time periods, the abnormality type is determined to be: the thickness value of the carbon fiber prepreg shows abnormal drift and the carbon fiber prepreg exhibits local hardening.

5. The method for controlling the layered cutting path of carbon fiber prepreg according to claim 3 or 4, characterized in that, When the anomaly type is: the thickness value of the carbon fiber prepreg shows abnormal drift during measurement; The step of correcting the geometric perception information based on the anomaly analysis results specifically includes: Based on the first normal correspondence between the preset reference thickness value and the reference torque value, calculate the second torque value of the servo motor that matches the thickness value; Calculate the first abnormal deviation between the torque value and the second torque value, and calculate the first confidence level of the thickness value based on the first abnormal deviation; Compare the first confidence level with a preset first confidence threshold. When the first confidence level is lower than the first confidence threshold, the thickness compensation value is calculated based on the thickness sliding average and the preset benchmark thickness average. The thickness value is corrected based on the thickness compensation value to obtain the corrected carbon fiber prepreg thickness value.

6. The method for controlling the layered cutting path of carbon fiber prepreg according to claim 3 or 4, characterized in that, When the anomaly type is: the thickness value of the carbon fiber prepreg shows abnormal drift during measurement, and the carbon fiber prepreg exhibits localized hardening, the specific steps for correcting the geometric sensing information based on the anomaly analysis results include: Based on the preset second normal correspondence between the reference thickness value and the reference torque value, calculate the third torque value of the servo motor that matches the thickness value; Calculate the second abnormal deviation between the fourth torque value outside the time period when the servo motor experiences instantaneous torque abnormality and the third torque value, and calculate the second confidence level of the thickness value based on the second abnormal deviation; Compare the second confidence level with a preset second confidence threshold. When the second confidence level is lower than the second confidence threshold, a first thickness compensation value is calculated based on the thickness sliding average and the preset reference thickness average. The thickness value is corrected according to the first thickness compensation value to obtain the corrected carbon fiber prepreg thickness value; The step of generating motion physical marker information based on the anomaly analysis results of motion physical information specifically includes: Acquire the spatial position information of the tool used to cut carbon fiber prepreg during the period when the servo motor experiences an instantaneous torque anomaly; Based on the fifth torque value during the period when the servo motor experiences a momentary torque anomaly; Calculate the degree of hardening in the corresponding local hardening region of the carbon fiber prepreg based on the fifth torque value; The spatial location information and the hardening degree value are fused into the motion physical indicator information.

7. The method for controlling the layered cutting path of carbon fiber prepreg according to claim 1, characterized in that, The specific steps for anomaly analysis of motion physics information include: The first torque sliding average value of the servo motor torque value is calculated in real time over a fourth set time period. Calculate the magnitude of the first torque sliding average value and the preset first torque threshold value. If the first torque sliding average value exceeds the first torque threshold value and the duration is less than the fifth preset time, it indicates that the servo motor has an instantaneous torque abnormality. The step of generating motion physical marker information based on the anomaly analysis results of motion physical information specifically includes: Acquire the first spatial position information of the tool used to cut carbon fiber prepreg during the period when the servo motor experiences an instantaneous torque anomaly; Obtain the sixth torque value during the period when the servo motor experiences a momentary torque anomaly; Based on the sixth torque value, calculate the first degree of hardening value of the corresponding locally hardened region in the carbon fiber prepreg. The first spatial location information and the first hardening degree value are fused into the motion physical marker information.

8. The method for controlling the layered cutting path of carbon fiber prepreg according to claim 2, characterized in that, The motion physics information also includes vibration data of the cutting tool used to cut the carbon fiber prepreg; Before the step of calculating the first torque value of the servo motor that matches the thickness value based on the preset normal correspondence between the reference thickness value and the reference torque value, the following steps are also included: Obtain the batch identifier of the current batch of carbon fiber prepreg, and search for the normal correspondence that matches the current batch from the preset normal correspondence database based on the batch identifier. If no matching normal correspondence is found, the torque value is high-pass filtered to obtain the high-frequency component of the torque, and the root mean square value of the high-frequency component of the torque is calculated. Perform a real-time fast Fourier transform on the vibration data to extract energy values ​​within a set frequency range; Compare the root mean square value with a preset root mean square threshold, and compare the energy value with a preset energy threshold; Based on the comparison results and preset modification rules, the normal correspondence is corrected to generate a corrected third normal correspondence carrying a batch identifier, which is used to correct the thickness value of carbon fiber prepreg during the subsequent cutting process of the corresponding batch of carbon fiber prepreg.

9. The method for controlling the layered cutting path of carbon fiber prepreg according to claim 8, characterized in that, The process of performing high-pass filtering on the torque value to obtain the high-frequency component of the torque, and calculating the root mean square value of the high-frequency component of the torque, further includes the following steps: Calculate the kurtosis coefficient of the high-frequency component of torque, and calculate the torque tolerance based on the magnitude of the kurtosis coefficient; The step of correcting the geometric perception information based on the anomaly analysis results specifically includes: Based on the thickness value and the preset calculation rules, calculate the expected torque center value for the current batch; Calculate the normal torque range based on the torque tolerance and the expected torque center value; When the torque value is outside the normal torque range, the second thickness value of the carbon fiber prepreg that matches the torque value is calculated according to the corrected third normal correspondence. The thickness value and the second thickness value are fused and calculated according to a preset rule to obtain the corrected carbon fiber prepreg thickness value; When the torque value is within the normal torque range, the thickness value is defined as the corrected carbon fiber prepreg thickness value.

10. A carbon fiber prepreg layer cutting path control system, characterized in that, include: The acquisition module is used to acquire geometric sensing information and motion physics information in real time during the cutting process of carbon fiber prepreg; The analysis module is used to perform anomaly analysis on geometric perception information and / or anomaly analysis on motion physics information; The correction module is used to correct the geometric sensing information based on the results of anomaly analysis; and / or, The generation module is used to generate motion physics marker information based on the results of anomaly analysis of motion physics information; The control module is used to adjust the cutting path, cutting depth and cutting speed of the carbon fiber prepreg during the cutting process based on the corrected geometric perception information and / or the motion physical marker information, so as to ensure the accuracy and quality of the cutting edge. The geometric sensing information includes the thickness value of the carbon fiber prepreg; the motion physics information includes the torque value of the servo motor used to cut the carbon fiber prepreg.

Images