Carbon fiber prepreg performance intelligent evaluation method and system

Abstract

This invention discloses an intelligent evaluation method and system for carbon fiber prepreg performance, relating to the field of carbon fiber prepreg performance testing technology. The method includes: acquiring the dynamic process response parameter sequence of the target prepreg during the curing process; calling the curing degree-process response baseline corresponding to the resin system type; comparing the dynamic process response parameter sequence with the curing degree-process response baseline on the corresponding time axis, calculating the time-series exothermic deviation integral value as a process fluctuation characteristic quantity; based on the process fluctuation characteristic quantity and the fiber volume fraction of the target prepreg, combined with the curing degree-process response baseline, obtaining the curing reaction rate constant, obtaining the predicted curing degree distribution characteristic parameter, inputting it into the fatigue resistance performance mapping model, obtaining the fatigue resistance performance characterization value; calculating the curing degree deviation correction factor, correcting the fatigue resistance performance characterization value, and obtaining the fatigue resistance performance evaluation result. This invention achieves real-time evaluation of carbon fiber prepreg performance.

Description

Technical Field

[0001] This invention relates to the field of carbon fiber prepreg performance testing technology, specifically to an intelligent evaluation method and system for carbon fiber prepreg performance. Background Technology

[0002] In the curing and molding process of carbon fiber prepreg, the uniformity of the degree of curing is a key factor affecting the final fatigue resistance of the composite material. Currently, the traditional method for assessing the difference in fatigue resistance caused by inconsistent degrees of curing usually involves preparing multiple sets of laminate samples with different curing process parameters, conducting fatigue tests on each sample, and indirectly inferring the relationship between the degree of curing and fatigue performance by comparing the dynamic modulus decay rate or fatigue life of each sample.

[0003] However, traditional methods rely entirely on the preparation of numerous physical samples and multiple sets of destructive fatigue tests, resulting in complex testing procedures, lengthy testing cycles, and high raw material consumption and testing costs. Furthermore, this type of testing is a post-production sampling inspection model, unable to capture fluctuations in process parameters throughout the curing process online, unable to quantify the impact of dynamic process deviations on the time distribution of curing degree, and difficult to accurately characterize the differences in fatigue resistance caused by inconsistent curing degree. Overall, the timeliness and comprehensiveness of the assessment are insufficient. Summary of the Invention

[0004] This invention provides an intelligent evaluation method and system for the performance of carbon fiber prepregs, aiming to solve the technical problem that existing technologies cannot evaluate the impact of process fluctuations on the fatigue resistance of carbon fiber prepregs in real time online.

[0005] In view of the above problems, the present invention provides a method and system for intelligent evaluation of the performance of carbon fiber prepreg.

[0006] In a first aspect, the present invention provides an intelligent evaluation method for the performance of carbon fiber prepregs, comprising: Obtain the dynamic process response parameter sequence of the target prepreg during the curing process, wherein the dynamic process response parameter sequence includes the measured exothermic power of at least one characteristic temperature point distributed along the process time axis. Based on the resin system type of the target prepreg, the corresponding curing degree-process response baseline is invoked, wherein the curing degree-process response baseline defines the mapping relationship between exothermic power and curing degree under standard curing process; The dynamic process response parameter sequence is compared with the curing degree-process response baseline on the corresponding time axis, and the integral value of the time-series exothermic deviation is calculated as a process fluctuation characteristic quantity. Based on the process fluctuation characteristics and the fiber volume fraction of the target prepreg, combined with the curing degree-process response baseline, the curing reaction rate constant is obtained, and the distribution of curing degree along the process time axis is predicted to obtain the predicted curing degree distribution characteristic parameters, wherein the predicted curing degree distribution characteristic parameters include at least the curing degree mean and the curing degree standard deviation. The predicted curing degree distribution characteristic parameters are input into the pre-constructed fatigue resistance performance mapping model to obtain fatigue resistance performance characterization values; The curing degree deviation correction factor is calculated based on the average curing degree value. The fatigue resistance performance characterization value is then corrected using the curing degree deviation correction factor to obtain the final fatigue resistance performance evaluation result.

[0007] Secondly, the present invention provides an intelligent evaluation system for the performance of carbon fiber prepregs, comprising: The response parameter acquisition module is used to acquire the dynamic process response parameter sequence of the target prepreg during the curing process, wherein the dynamic process response parameter sequence includes the measured exothermic power of at least one characteristic temperature point distributed along the process time axis. The response baseline calling module is used to call the corresponding curing degree-process response baseline according to the resin system type of the target prepreg, wherein the curing degree-process response baseline defines the mapping relationship between exothermic power and curing degree under standard curing process; The fluctuation characteristic calculation module is used to compare the dynamic process response parameter sequence with the curing degree-process response baseline on the corresponding time axis, and calculate the integral value of the time-series exothermic deviation as a process fluctuation characteristic quantity. The curing degree distribution prediction module is used to obtain the curing reaction rate constant based on the process fluctuation characteristic quantity and the fiber volume fraction of the target prepreg, combined with the curing degree-process response baseline, and to predict the distribution of curing degree along the process time axis, thereby obtaining the predicted curing degree distribution characteristic parameters, wherein the predicted curing degree distribution characteristic parameters include at least the curing degree mean and the curing degree standard deviation. The fatigue performance mapping module is used to input the predicted curing degree distribution characteristic parameters into the pre-constructed fatigue resistance performance mapping model to obtain fatigue resistance performance characterization values. The performance deviation correction module is used to calculate the curing degree deviation correction factor based on the average curing degree, and use the curing degree deviation correction factor to correct the fatigue resistance performance characterization value to obtain the final fatigue resistance performance evaluation result.

[0008] One or more technical solutions provided in this invention have at least the following technical effects or advantages: This invention provides an intelligent evaluation method and system for the performance of carbon fiber prepregs. First, the dynamic process response parameter sequence of the target prepreg during the curing process is acquired. Combined with a pre-stored curing degree-process response baseline, the integral value of the time-series exothermic deviation is calculated as a process fluctuation characteristic. Then, based on the process fluctuation characteristic and fiber volume fraction, the distribution of the curing degree along the time axis is predicted, obtaining the mean and standard deviation of the curing degree. Finally, the predicted curing degree distribution characteristics are input into a fatigue resistance performance mapping model, and the output fatigue resistance performance characterization value is corrected using a curing degree deviation correction factor to obtain the final evaluation result. This invention eliminates the need for preparing a large number of samples and conducting destructive fatigue testing, shortening the evaluation cycle and reducing costs. Simultaneously, it can reflect the impact of process fluctuations on the curing degree in real time based on online process response data, achieving rapid, non-destructive intelligent evaluation of the fatigue resistance performance of prepregs, effectively meeting the practical needs of online quality monitoring. Attached Figure Description

[0009] Figure 1 A flowchart illustrating an intelligent performance evaluation method for carbon fiber prepreg provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of an intelligent evaluation system for carbon fiber prepreg performance provided in an embodiment of the present invention; The components represented by each number in the attached diagram are explained below: The module includes a response parameter acquisition module 11, a response baseline calling module 12, a fluctuation characteristic calculation module 13, a solidification degree distribution prediction module 14, a fatigue performance mapping module 15, and a performance deviation correction module 16. Detailed Implementation

[0010] This invention provides an intelligent evaluation method and system for the performance of carbon fiber prepregs, which addresses the technical problem that existing technologies cannot evaluate the impact of process fluctuations on the fatigue resistance of carbon fiber prepregs in real time online.

[0011] Example 1, as Figure 1 As shown, this invention provides an intelligent evaluation method for the performance of carbon fiber prepregs, the method comprising: S100: Obtain the dynamic process response parameter sequence of the target prepreg during the curing process, wherein the dynamic process response parameter sequence includes the measured exothermic power of at least one characteristic temperature point distributed along the process time axis.

[0012] In this embodiment of the invention, a dynamic process response parameter sequence of the target prepreg during the curing process is obtained. This dynamic process response parameter sequence includes the measured exothermic power at at least one characteristic temperature point distributed along the process time axis. The resin crosslinking chemical reaction during the curing stage of the carbon fiber prepreg continuously releases heat, and the dynamic change in exothermic power directly reflects the progress and rate of the curing reaction. Uneven resin flow and localized heat dissipation differences exist within the curing mold, making data from a single temperature detection point insufficiently representative. Temperature sensing devices are susceptible to electromagnetic interference and hardware errors, easily generating data jump noise. Furthermore, the specific heat capacity of the carbon fiber and resin components differs significantly, and the fiber volume fraction directly affects the overall heat capacity. Relying solely on temperature change data cannot accurately calculate the true exothermic power of the curing reaction. Therefore, a distributed multi-point temperature acquisition method is required, combined with the prepreg material ratio and thermophysical parameters, to accurately solve for the exothermic power. Abnormal noise data is then eliminated, and a standardized, continuous measured exothermic power time series is formed, providing accurate raw data support for subsequent curing degree analysis and performance evaluation.

[0013] Step S100 in the method provided in this embodiment of the invention includes: Multiple temperature sensing nodes are arranged along the resin flow direction on the inner surface of the curing mold of the target prepreg, and the position of each temperature sensing node corresponds to a characteristic temperature point. The measured temperature values ​​of each temperature sensing node in the curing heating section, isothermal section and cooling section are continuously collected at a fixed sampling frequency to generate a temperature-time series of each characteristic temperature point. Based on the mass and specific heat capacity of the target prepreg, and combined with the temperature-time series of each characteristic temperature point, the instantaneous heat release power at each sampling time of each characteristic temperature point is calculated to obtain the instantaneous heat release power series of each characteristic temperature point. For the instantaneous heat release power sequence at the same characteristic temperature point, based on the power change rate between adjacent sampling times, power jump points caused by sensor noise are removed to obtain the corrected instantaneous heat release power. The average instantaneous heat release power of each characteristic temperature point at the same sampling time is taken and arranged in chronological order to form a sequence of measured heat release power distributed along the process time axis.

[0014] First, multiple temperature sensing nodes are arranged along the resin flow direction on the inner surface of the curing mold of the target prepreg, and the position of each temperature sensing node corresponds to a characteristic temperature point. The characteristic temperature point refers to a preset detection point that covers the key heat exchange area of ​​the prepreg during curing, used to provide feedback on temperature changes throughout the curing process. The resin flow direction refers to the natural direction of fluid expansion during the prepreg laying and resin melting impregnation process.

[0015] Specifically, several sets of temperature sensing nodes are strictly arranged along the resin flow direction on the inner surface of the curing mold of the target prepreg, so that each sensing node corresponds to a preset characteristic temperature point, realizing distributed temperature monitoring of the entire curing area. For example, taking a certain type of epoxy carbon fiber prepreg as the detection object, the resin flow direction is determined according to the board laying direction, and three temperature sensing nodes are evenly arranged on the inner side of the curing mold. The three nodes correspond to three characteristic temperature points at the front end, middle end and rear end, respectively, completely covering the entire resin flow area.

[0016] Secondly, the measured temperature values ​​of each temperature sensing node during the curing heating, isothermal, and cooling stages are continuously collected at a fixed sampling frequency to generate a temperature-time series for each characteristic temperature point. The fixed sampling frequency refers to the fixed number of times the sensor collects temperature data per unit time. The complete curing process includes three stages: heating, isothermal, and cooling. The temperature-time series refers to the set of measured temperature values ​​recorded by the same sensing node in chronological order of sampling time.

[0017] Specifically, a fixed sampling frequency is set, and real-time measured temperature values ​​of all temperature sensing nodes in the heating, constant temperature and cooling stages are continuously collected during the complete curing cycle of the prepreg. The values ​​are arranged in the order of collection time to generate an independent temperature-time series for each characteristic temperature point.

[0018] For example, a uniform sampling interval of 5 seconds is set, meaning data acquisition is completed every 5 seconds. The entire curing process is continuously collected, generating temperature-time series for three characteristic temperature points: the front end, the middle end, and the back end. Each series includes temperature data for the heating, isothermal, and cooling phases. The total curing time is 1800 seconds. The heating phase (0-600 seconds) involves a uniform temperature increase from 25℃ to approximately 120℃; the isothermal phase (600-1500 seconds) maintains a stable temperature of approximately 120℃ throughout; and the cooling phase (1500-1800 seconds) gradually decreases the temperature to approximately 40℃. Data is organized according to the acquisition time sequence, generating independent temperature-time series for each characteristic temperature point. For example, the time series data of the front-end characteristic temperature points are [0s: 25.8℃, ..., 300s: 72.4℃, ..., 1800s: 40.2℃], the time series data of the middle characteristic temperature points are [0s: 26.0℃, ..., 300s: 72.9℃, ..., 1800s: 40.0℃], and the time series data of the back-end characteristic temperature points are [0s: 26.3℃, ..., 300s: 73.2℃, ..., 1800s: 39.8℃]. The three sets of sequences completely cover the temperature change pattern of the three stages of solidification.

[0019] Next, based on the mass and specific heat capacity of the target prepreg, and combined with the temperature-time series of each characteristic temperature point, the instantaneous heat release power at each sampling time of each characteristic temperature point is calculated to obtain the instantaneous heat release power series of each characteristic temperature point.

[0020] Specifically, based on the mass and specific heat capacity of the target prepreg, and combined with the temperature-time series of each characteristic temperature point, the instantaneous heat release power at each sampling time of each characteristic temperature point is calculated to obtain the instantaneous heat release power sequence of each characteristic temperature point, including: For each characteristic temperature point, the measured temperature value at the current sampling time is extracted from the temperature-time series of the corresponding characteristic temperature point; Extract the measured temperature value of the previous adjacent sampling time from the current sampling time, calculate the difference between the measured temperature value of the current sampling time and the measured temperature value of the previous adjacent sampling time, and combine it with the sampling time interval to obtain the forward temperature change rate at the sampling time. Extract the measured temperature value of the next adjacent sampling time from the sampling time, calculate the difference between the measured temperature value of the next adjacent sampling time and the measured temperature value, and combine it with the sampling time interval to obtain the backward temperature change rate at the sampling time. Calculate the difference between the forward temperature change rate and the backward temperature change rate, and combine it with the sampling time interval to obtain the temperature change acceleration at the sampling moment; The mass and specific heat capacity of the target prepreg are obtained, the heat capacity product is calculated, and the basic exothermic power is obtained by combining the forward temperature change rate. The basic heat release power is inertially corrected based on the heat capacity product and the temperature change acceleration to obtain the instantaneous heat release power at the sampling time; Iterate through all sampling times to obtain the instantaneous heat release power of the current characteristic temperature point at each sampling time; By iterating through all characteristic temperature points, the instantaneous heat release power sequence of each characteristic temperature point is obtained.

[0021] First, for each characteristic temperature point, the measured temperature value at the current sampling moment is extracted from the temperature-time series of the corresponding characteristic temperature point. For each characteristic temperature point, the corresponding temperature-time series is retrieved individually, and the independent measured temperature values ​​at the current sampling moment, the previous adjacent sampling moment, and the next adjacent sampling moment are extracted for single-point thermal parameter calculation. For example, selecting the sampling moment at the 300th second of curing, three independent temperatures are extracted: 72.4℃ at the front end, 72.9℃ at the middle end, and 73.2℃ at the back end. Each point uses its own temperature for subsequent calculations.

[0022] Secondly, the measured temperature value of the previous adjacent sampling time is extracted from the current sampling time. The difference between the measured temperature value of the current sampling time and the measured temperature value of the previous adjacent sampling time is calculated. Combined with the sampling time interval, the forward temperature change rate at the sampling time is obtained. The forward temperature change rate refers to a physical quantity that characterizes the speed of positive temperature change based on the current time and the temperature data of the previous adjacent time. For a single characteristic temperature point, the forward temperature change rate is calculated according to the formula: Forward temperature change rate = (single-point temperature at the current sampling time - single-point temperature at the previous adjacent sampling time) / sampling time interval.

[0023] For example, with a sampling interval of 5 seconds, taking 300 seconds of data from each point as an example, the previous adjacent sampling time is the 295th second. The temperatures at each point at 295 seconds are: front end 71.60℃, middle end 72.10℃, and back end 72.45℃. The forward temperature change rate of the front end is (72.4-71.60) / 5=0.16℃ / s, the forward temperature change rate of the middle end is (72.9-72.10) / 5=0.16℃ / s, and the forward temperature change rate of the back end is (73.2-72.45) / 5=0.15℃ / s.

[0024] Next, the measured temperature value of the next adjacent sampling time is extracted from the sampling time. The difference between the measured temperature value of the next adjacent sampling time and the measured temperature value is calculated. Combined with the sampling time interval, the backward temperature change rate of the sampling time is obtained. The backward temperature change rate refers to a physical quantity that characterizes the rate of temperature lag change based on the current time and the temperature data of the next adjacent time. For a single characteristic temperature point, the backward temperature change rate is calculated according to the formula: Backward temperature change rate = (single-point temperature of the next adjacent sampling time - single-point temperature of the current sampling time) / sampling time interval.

[0025] For example, the next adjacent sampling time is the 305th second. The temperatures at each point in 305 seconds are: front end 73.30℃, middle end 73.75℃, and rear end 74.05℃. The backward temperature change rate of the front end is (73.30-72.4) / 5=0.18℃ / s, the backward temperature change rate of the middle end is (73.75-72.9) / 5=0.17℃ / s, and the backward temperature change rate of the rear end is (74.05-73.2) / 5=0.17℃ / s.

[0026] Then, the difference between the forward temperature change rate and the backward temperature change rate is calculated, and combined with the sampling time interval, the temperature change acceleration at the sampling moment is obtained. Temperature change acceleration is a correction parameter characterizing the rate of temperature change, used to compensate for detection bias caused by temperature lag. The forward and backward temperature change rates are retrieved, and the temperature change acceleration is calculated according to the formula: Temperature change acceleration = (Forward temperature change rate - Backward temperature change rate) / Sampling time interval. For example, the temperature change acceleration at each point at 300 seconds is: Front-end temperature change acceleration. Acceleration of mid-range temperature change Acceleration of temperature change at the back end .

[0027] Subsequently, the mass and specific heat capacity of the target prepreg are obtained, the heat capacity product is calculated, and the basic exothermic power is obtained by combining it with the forward temperature change rate.

[0028] The process of obtaining the mass and specific heat capacity of the target prepreg and calculating the heat capacity product includes: Before curing, the target prepreg is placed on a weighing device, and three consecutive weighing readings are taken. The arithmetic mean of the three readings is calculated as the mass of the target prepreg. Based on the resin system type of the target prepreg, the corresponding fiber specific heat capacity and resin specific heat capacity are extracted from the pre-stored material property database. Obtain the fiber volume fraction of the target prepreg and calculate the product of the fiber volume fraction and the fiber specific heat capacity as the fiber heat capacity contribution value; Based on the fiber volume fraction and the resin specific heat capacity, the resin heat capacity contribution value is calculated, and combined with the fiber heat capacity contribution value, the specific heat capacity of the target prepreg is obtained. The heat capacity product is determined based on the mass of the target prepreg and the specific heat capacity.

[0029] First, the target prepreg is placed on a weighing device before curing, and three consecutive weighing readings are taken. The arithmetic mean of the three readings is calculated as the mass of the target prepreg. The three-weighing average is used to reduce random errors in the weighing equipment and ensure the accuracy of the mass parameters. Before curing, the complete target prepreg is placed on a standard weighing device, and three independent weighing values ​​are taken consecutively. The arithmetic mean of the three values ​​is calculated as the final mass of the prepreg. For example, the epoxy system prepreg is weighed three times, with readings of 205g, 203g, and 204g respectively. The average value is calculated to give the prepreg mass as 204g.

[0030] Secondly, based on the resin system type of the target prepreg, the corresponding fiber specific heat capacity and resin specific heat capacity values ​​are extracted from a pre-stored material property database. The material property database refers to a pre-entered library of thermophysical parameters corresponding to different resin systems and carbon fiber types. Specific heat capacity refers to the amount of heat required to raise the temperature per unit mass of material by one unit; it is an inherent property of the material.

[0031] Specifically, the resin system type of the target prepreg is identified, and the local pre-stored material property database is accessed to extract the corresponding specific heat capacity of carbon fiber and specific heat capacity of resin matrix. For example, in this case, the prepreg is an epoxy resin system, and the parameters retrieved from the database are: specific heat capacity of carbon fiber 0.75 J / (g·℃), and specific heat capacity of epoxy resin 1.45 J / (g·℃).

[0032] Next, the fiber volume fraction of the target prepreg is obtained, and the product of the fiber volume fraction and the fiber specific heat capacity is calculated as the fiber heat capacity contribution value. Based on the fiber volume fraction and the resin specific heat capacity, the resin heat capacity contribution value is calculated, and combined with the fiber heat capacity contribution value, the specific heat capacity of the target prepreg is obtained. The fiber volume fraction refers to the proportion of carbon fiber volume in the prepreg to the total material volume; this parameter is directly specified in the prepreg supplier's product specifications. The heat capacity contribution value refers to the proportionate contribution of a single component material to the overall heat capacity of the prepreg. The overall specific heat capacity of the target prepreg is obtained by superimposing the fiber heat capacity contribution value and the resin heat capacity contribution value.

[0033] Specifically, consult the product specification sheet to obtain the nominal fiber volume fraction, and calculate according to the formula: Fiber heat capacity contribution value = fiber volume fraction × fiber specific heat capacity; Resin heat capacity contribution value = (1 − fiber volume fraction) × resin specific heat capacity; Specific heat capacity of the target prepreg = fiber heat capacity contribution value + resin heat capacity contribution value.

[0034] For example, the epoxy prepreg has a nominal fiber volume fraction of 65%, a fiber heat capacity contribution of 0.65 × 0.75 = 0.4875 J / (g·℃), a resin heat capacity contribution of (1 - 0.65) × 1.45 = 0.5075 J / (g·℃), and a target prepreg specific heat capacity of 0.4875 + 0.5075 = 0.995 J / (g·℃).

[0035] Finally, the heat capacity product is determined based on the mass of the target prepreg and its specific heat capacity. The heat capacity product is the product of the prepreg mass and the overall specific heat capacity, and is a fundamental parameter for calculating the exothermic power. Calculating the heat capacity product using the prepreg mass: Heat capacity product = Mass of target prepreg × Specific heat capacity of target prepreg. For example, heat capacity product = 204g × 0.995J / (g·℃) = 202.98J / ℃.

[0036] Furthermore, the basic heat release power is obtained by combining the aforementioned forward temperature change rate. Basic heat release power refers to the initial heat release parameter calculated solely based on the temperature change rate, reflecting the basic heat release effect brought about by temperature changes. Basic heat release power = heat capacity product × forward temperature change rate. The heat capacity product reflects the overall heat storage capacity of the prepreg; combined with the temperature change rate, it yields the basic heat release intensity without considering inertial deviation. Using a single characteristic temperature point as the calculation unit, the heat capacity product of the prepreg is multiplied by the forward temperature change rate at the corresponding point at the current sampling time to obtain the basic heat release power at that characteristic temperature point.

[0037] For example, the calculation is performed at the 300th second sampling time of the characteristic temperature point at the front end. The known heat capacity product is 202.98 J / ℃, the forward temperature change rate at this moment is 0.16℃ / s, and the basic heat release power... The basic heat release power of the mid- and rear characteristic temperature points is calculated separately according to their corresponding forward temperature change rates and using the same calculation method.

[0038] Subsequently, the basic exothermic power is corrected for inertia based on the heat capacity product and the temperature change acceleration to obtain the instantaneous exothermic power at the sampling moment. Inertia correction refers to compensating for calculation deviations caused by mold heat transfer and material thermal inertia by incorporating temperature change acceleration, making the exothermic power more closely reflect the actual state of the curing reaction. The instantaneous exothermic power, after inertia correction, accurately characterizes the true exothermic intensity of the curing reaction at the current sampling moment. Instantaneous exothermic power = basic exothermic power + heat capacity product × temperature change acceleration. Temperature change acceleration reflects the fluctuation of the temperature rate; combined with the heat capacity product, it compensates for calculation deviations caused by inertia. Positive values ​​indicate accelerated heat release, and negative values ​​indicate decelerated heat release.

[0039] Specifically, the temperature change acceleration corresponding to a single point is introduced, and an inertial correction term is constructed by combining the heat capacity product. This inertial correction term is used to compensate for and correct the basic heat release power, eliminating the calculation deviation caused by material thermal inertia and local heat dissipation fluctuations. Finally, the instantaneous heat release power of a single characteristic temperature point at the current sampling moment is obtained. For example, the temperature change acceleration of the front-end characteristic temperature point at the 300th second is... Instantaneous heat release power Similarly, for the mid-range and rear-end characteristic temperature points, substitute their respective temperature change accelerations to complete the inertial correction and obtain the instantaneous heat release power at the corresponding sampling time.

[0040] Finally, by iterating through all sampling times, the instantaneous heat release power of the current characteristic temperature point at each sampling time is obtained; by iterating through all characteristic temperature points, the instantaneous heat release power sequence of each characteristic temperature point is obtained. Substituting the forward temperature change rate and temperature change acceleration of each sampling time one by one, the above two steps are repeated to obtain the instantaneous heat release power sequence of each sampling time throughout the entire time period. For example, by iterating through all sampling times from 0 to 1800 seconds using the above method, avoiding the calculation blind spot of missing adjacent data at the beginning and end, the complete power sequence of each point is obtained: the instantaneous heat release power sequence at the front end: [5s: 12.56J / s, ..., 300s: 31.67J / s, ..., 1795s: 31.15J / s]; the middle and back ends are synchronously generated with their own independent instantaneous heat release power sequences.

[0041] Furthermore, for the instantaneous exothermic power sequence at the same characteristic temperature point, based on the power change rate between adjacent sampling times, power jump points caused by sensor noise are eliminated to obtain the corrected instantaneous exothermic power. The power change rate refers to the amplitude of the change in instantaneous exothermic power between two adjacent sampling times, used to determine whether the data is abnormal. A power jump point refers to an irregular power change caused by sensor noise or signal interference, where the change rate exceeds a reasonable threshold, and is inconsistent with the exothermic behavior of the curing reaction itself.

[0042] Specifically, the power change rate is calculated as follows: Power change rate = |Instantaneous heat release power at the current moment - Instantaneous heat release power at the previous adjacent moment| / Instantaneous heat release power at the previous adjacent moment × 100%. A sudden change threshold is set: based on the heat release characteristics of epoxy prepreg curing, for example, a power sudden change threshold of 20% is set, meaning a change rate exceeding 20% ​​is considered a jump point. The power change rates at adjacent sampling moments are compared segment by segment, and abnormal data points exceeding the 20% threshold are removed, retaining only the corrected instantaneous heat release power with a change rate < 20% that conforms to the heat release characteristics.

[0043] For example, taking the instantaneous heat release power data segments at the characteristic temperature points of the front end as an example: 300s: 31.67 J / s, 305s: 30.92 J / s, 310s: 39.15 J / s, 315s: 31.32 J / s. Calculate the power change rate at each adjacent time point: 305s relative to 300s: Change rate. This is normal data; 310s relative to 305s: rate of change The data at 315s was identified as a power jump point. The change rate from 310s to 315s was 20%. Since 310s was the jump point, the data at 315s was affected and simultaneously identified as abnormal. The abnormal data at 310s and 315s were removed. Linear interpolation was used between 305s (30.92 J / s) and 320s (31.08 J / s) to supplement and correct the data: 310s was corrected to 30.98 J / s, and 315s to 31.04 J / s. After correction, the power data for this period were: 300s: 31.67 J / s, 305s: 30.92 J / s, 310s: 30.98 J / s, 315s: 31.04 J / s, and 320s: 31.08 J / s, with all change rates < 20%, consistent with the heat release law of curing. Similarly, for the mid-range and rear-end characteristic temperature points, noise is removed from their respective instantaneous heat release power sequences to obtain the corrected instantaneous heat release power at the corresponding points.

[0044] Finally, the average of the corrected instantaneous heat release power at each characteristic temperature point at the same sampling time is taken and arranged in chronological order to form a measured heat release power sequence distributed along the process time axis. The time-series fusion mean refers to the arithmetic mean of the corrected power at multiple characteristic temperature points at the same time node, which is used to weaken the error caused by local curing differences and make the final sequence more representative.

[0045] Specifically, for each sampling time, the corrected instantaneous heat release power at each characteristic temperature point is extracted. At the same sampling time, the arithmetic mean of the power data at each characteristic temperature point is taken as the measured heat release power at that time. The measured heat release power at all sampling times is arranged strictly in chronological order according to the curing process timeline to form a complete measured heat release power sequence.

[0046] For example, the corrected instantaneous heat release power at the 300th second is: front-end characteristic point: 31.67 J / s, middle-end characteristic point: 31.82 J / s, and rear-end characteristic point: 31.97 J / s; the measured heat release power... The mean of all sampling times is calculated using this method and arranged in chronological order to form a complete measured heat release power sequence, such as: [5s: 10.23J / s, ..., 300s: 31.82J / s, ..., 1795s: 29.87J / s], which fully covers the entire process of heating, isothermal, and cooling.

[0047] In this embodiment of the invention, multiple sensing nodes are deployed along the resin flow direction, and each characteristic temperature point uses an independent temperature sequence to calculate thermodynamic parameters separately, accurately distinguishing the local differences in heat release during curing in different areas of the mold. Based on the modeling of thermophysical parameters within the compliant range of carbon fiber and epoxy resin, and combined with the specific temperature change rate and acceleration at each point for inertial correction, the accuracy of heat release power calculation is higher. Simultaneously, noise is independently eliminated at each point to suppress single-point sensor acquisition errors, and finally, the local operating condition deviation is weakened by fusing the power averages from multiple points. The final output is a time-continuous, interference-resistant, full-domain dynamic heat release power sequence, completely restoring the full-cycle curing reaction state of the carbon fiber prepreg, providing reliable underlying data for subsequent baseline comparison of curing degree, quantification of process fluctuations, and intelligent evaluation of fatigue resistance.

[0048] S200: Based on the resin system type of the target prepreg, call the corresponding curing degree-process response baseline, wherein the curing degree-process response baseline defines the mapping relationship between exothermic power and curing degree under standard curing process.

[0049] In this embodiment of the invention, based on the resin system type of the target prepreg, a corresponding curing degree-process response baseline is invoked. This baseline defines the mapping relationship between exothermic power and curing degree under a standard curing process. Different resin systems exhibit variations in crosslinking reaction exothermic rates and exothermic heat, and the fiber volume fraction directly constrains the resin proportion and overall exothermic characteristics. Therefore, the correlation between exothermic power and curing degree for different prepreg formulations lacks universality. Under on-site conditions, only measured exothermic power time-series data can be collected, making it impossible to directly quantify the curing degree value, and a standard reference is lacking. Therefore, it is necessary to conduct calibration tests in advance on standard samples with matching resin systems and fiber volume fractions to construct a standardized curing degree-process response baseline, establishing a one-to-one mapping relationship between exothermic power and curing degree under a standard curing process. This provides a unified reference standard for subsequent comparison of measured parameters, quantification of process fluctuations, and curing degree inversion calculations.

[0050] Step S200 in the method provided in this embodiment of the invention includes: The steps for constructing the curing degree-process response baseline include: Prepare a standard sample with the same resin system type and the same fiber volume fraction as the target prepreg, place the standard sample in a differential scanning calorimeter, and heat it according to the standard curing process. During the curing process, the real-time exothermic power of the standard sample is continuously collected, and the measured value of the degree of curing is extracted from the differential scanning calorimeter at fixed time intervals to obtain the discrete point pair of exothermic power-degree of curing. The discrete points of the exothermic power-degree of cure were sorted, and a scatter plot was drawn with exothermic power as the abscissa and degree of cure as the ordinate. From the scatter plot, the curing degree value before the heat release power starts to rise continuously is selected as the curing degree zero point, and the curing degree value corresponding to the point where the heat release power reaches the peak value and then drops to half of the peak value is selected as the gel point curing degree. Adjacent discrete point pairs are connected using a linear interpolation method to form a continuous curve, which serves as the baseline for the degree of curing-process response.

[0051] First, a standard sample with the same resin system type and fiber volume fraction as the target prepreg is prepared. The standard sample is then placed in a differential scanning calorimeter and heated according to a standard curing process. The standard sample refers to a parallel control sample that is completely identical to the target prepreg in terms of resin system, fiber volume fraction, and raw material specifications. The standard curing process refers to a temperature control procedure with the same curing parameters as in on-site production, including a heating stage, a isothermal stage, and a cooling stage.

[0052] Specifically, a standard epoxy resin raw material consistent with the target prepreg was selected to prepare standard samples of the same specifications, ensuring that the fiber volume fraction was consistent with that of the test piece. The standard sample was placed inside a differential scanning calorimeter, and the temperature rise was programmed to replicate the complete on-site curing process parameters. For example, the fiber volume fraction was fixed at 65%. The differential scanning calorimeter was loaded with a synchronous curing program: from 0 to 600 seconds, the temperature was increased from 25°C to 120°C; from 600 to 1500 seconds, the temperature was kept constant at 120°C; and from 1500 to 1800 seconds, the temperature was decreased to 40°C, completely replicating the on-site curing temperature control curve.

[0053] Secondly, during the curing process, the real-time exothermic power of the standard sample is continuously collected, and the measured degree of cure is extracted from the differential scanning calorimeter at fixed time intervals to obtain discrete point pairs of exothermic power and degree of cure. The discrete point pairs refer to the real-time exothermic power value and the measured degree of cure recorded synchronously at a single sampling moment, serving as the basic raw data for constructing the baseline. The fixed time interval adopts the same acquisition cycle as the on-site temperature sampling to ensure the consistency of the data time series benchmark.

[0054] Specifically, during the entire curing process of the standard sample, the differential scanning calorimeter continuously collects the reaction exothermic power in real time; at a fixed time interval of 5 seconds, the instrument's built-in calculated measured value of the degree of curing is extracted synchronously, the time sequence data is recorded one by one, and multiple sets of exothermic power-degree of curing discrete point pairs are generated in batches.

[0055] For example, synchronous sampling is performed with a collection cycle of 5 seconds to obtain multiple sets of matching data in sequence: 5s (10.18J / s, curing degree 2.3%), 300s (31.65J / s, curing degree 41.5%), 600s (33.85J / s, curing degree 76.2%), and 1795s (29.85J / s, curing degree 94.8%), forming a complete discrete dataset.

[0056] Next, the discrete points of the exothermic power-degree of cure are sorted, and a scatter plot is drawn with exothermic power as the abscissa and degree of cure as the ordinate. The mapping rules for the abscissa and ordinate are as follows: exothermic power is used as the abscissa to represent the exothermic intensity of the curing reaction; degree of cure is used as the ordinate to represent the degree of completion of the resin crosslinking reaction. All discrete point pairs are sorted sequentially according to the order of the curing reaction, and a unified coordinate axis definition is used to map all discrete data points to the coordinate system, generating a complete scatter plot that visually reflects the change in exothermic power with increasing degree of cure.

[0057] For example, all the time sequence points such as 5s, 300s, 600s, and 1795s are uniformly sorted, and the points of all test data points are plotted with the exothermic power as the horizontal axis and the degree of curing as the vertical axis, showing the distribution characteristics of the epoxy system curing exothermic heat rising first, stabilizing at the peak, and slowly decaying in the later stage.

[0058] Furthermore, the degree of cure before the exothermic power begins to rise continuously is selected from the scatter plot as the zero point of cure, and the degree of cure corresponding to the point where the exothermic power reaches its peak and then drops to half of its peak value is selected as the gel point degree of cure. The zero point of cure refers to the initial degree of cure where the resin has not yet undergone effective cross-linking reaction and the exothermic power has not continued to rise. The gel point degree of cure refers to the critical degree of cure where the resin reaches the gel state and the system changes from liquid to solid, corresponding to the characteristic node where the peak exothermic power decays to half.

[0059] Specifically, observe the distribution pattern of scattered points, select the stable interval before the stage of continuous increase in exothermic power, and take the corresponding value as the zero point of curing degree; locate the peak value of global exothermic power, calculate the power value corresponding to half of the peak value, and match the corresponding curing degree in reverse, which is calibrated as the gel point curing degree.

[0060] For example, the stable low-amplitude stage of exothermic power corresponds to a degree of curing of 2.0%, which is set as the degree of curing zero; the peak value of exothermic power during the curing stage is 34.12 J / s, and half of the peak value is 17.06 J / s. This power corresponds to a degree of curing of 58.5%, which is calibrated as the degree of curing at the gel point.

[0061] Finally, a linear interpolation method is used to connect adjacent discrete point pairs to form a continuous curve, which serves as the curing degree-process response baseline. Linear interpolation refers to a fitting method that uses two adjacent sets of discrete test points as a reference and completes the continuous values ​​within the interval using a linear function, thereby realizing the continuity of discrete data. Taking the zero point of curing degree and the gel point of curing degree as key dividing nodes, linear interpolation is used to connect all adjacent exothermic power-curing degree discrete point pairs one by one to eliminate data breakpoints and form a continuous and smooth change curve throughout the entire process, which is finally determined as the curing degree-process response baseline specific to this resin system.

[0062] For example, using two sets of discrete points at 300s and 600s as interval benchmarks, linear interpolation is used to complete the degree of curing value corresponding to any exothermic power between 300s and 600s; after uniform fitting of the entire interval, a continuous baseline suitable for epoxy prepreg with 65% fiber volume fraction is obtained, which can realize the rapid matching and query of any exothermic power and degree of curing.

[0063] Based on this, according to the resin system type of the target prepreg, the corresponding curing degree-process response baseline is invoked. The resin system type and fiber volume fraction of the target prepreg are identified, and the curing degree-process response baseline calibrated with the same formulation and process is retrieved from the database for subsequent time-series comparison and deviation calculation of the measured exothermic power sequence. For example, if the test piece is a certain type of epoxy-based carbon fiber prepreg with a fiber volume fraction of 65%, the specific curing degree-process response baseline constructed for this experiment is accurately retrieved to complete parameter matching.

[0064] In this embodiment of the invention, standard samples with equal proportions are prepared, and calibration tests are conducted by replicating the on-site curing process to ensure a high degree of matching between the baseline and the material properties and process environment of the prepreg under test. Using a differential scanning calorimeter to simultaneously acquire dual-core parameters, combined with key feature node calibration and linear interpolation fitting, discrete experimental data are transformed into a continuous mapping baseline. This effectively solves the problem of inconsistent exothermic-curing degree correlation patterns among different resin systems, achieving a quantitative correlation between exothermic power and curing degree, and providing a standardized and high-precision reference for subsequent calculation of time-series exothermic deviation integrals, curing reaction rate solutions, and curing degree distribution prediction.

[0065] S300: Compare the dynamic process response parameter sequence with the curing degree-process response baseline on the corresponding time axis, and calculate the integral value of the time-series exothermic deviation as a process fluctuation characteristic quantity.

[0066] In this embodiment of the invention, the dynamic process response parameter sequence is compared with the curing degree-process response baseline on the corresponding time axis, and the integral value of the time-series exothermic deviation is calculated as a characteristic quantity of process fluctuation. The actual curing process is affected by factors such as uneven heat dissipation from the mold, differences in prepreg placement, and environmental disturbances, resulting in a real-time deviation between the measured exothermic power and the baseline exothermic power of the standard curing process. The power deviation at a single moment can only reflect local instantaneous differences and cannot characterize the overall fluctuation level of the entire process cycle. Therefore, using the curing degree-process response baseline as a unified reference, the normalized instantaneous exothermic deviation is calculated moment by moment, integrated and accumulated along the complete process time axis to obtain the integral value of the time-series exothermic deviation. This can quantitatively characterize the overall deviation of the actual curing conditions from the standard process, achieving a digital and quantitative evaluation of curing process fluctuations.

[0067] Step S300 in the method provided in this embodiment of the invention includes: The measured heat release power at each sampling moment is extracted from the dynamic process response parameter sequence to form a set of measured heat release power time series values; Based on the curing degree-process response baseline, the standard exothermic power corresponding to each sampling time under the standard curing process is obtained, forming a set of standard exothermic power time series values. Based on the measured heat release power and the standard heat release power at the same sampling time, the instantaneous heat release deviation value at each sampling time is determined; The instantaneous heat release deviation value is integrated and accumulated along the process time axis to obtain the time-series heat release deviation integral value, which is used as the process fluctuation characteristic quantity.

[0068] First, the measured exothermic power at each sampling moment is extracted from the dynamic process response parameter sequence to form a set of measured exothermic power time series values. The set of measured exothermic power time series values ​​refers to the collection of measured exothermic power data after correction, arranged sequentially along the curing time axis and fully covering the heating, isothermal, and cooling phases. From the measured exothermic power sequence finally generated by S100, the measured exothermic power corresponding to all valid sampling moments is extracted one by one according to the sampling time sequence, and integrated and summarized to form a continuous and complete set of measured exothermic power time series values. For example, the set of measured exothermic power time series values ​​is [5s: 10.23J / s, ..., 300s: 31.82J / s, ..., 1795s: 29.87J / s].

[0069] Secondly, based on the curing degree-process response baseline, the standard exothermic power corresponding to each sampling time under the standard curing process is obtained, forming a set of standard exothermic power time series values. Standard exothermic power refers to the theoretical exothermic power value obtained by mapping the curing degree-process response baseline to the standard sample under ideal process conditions without interference at the same curing time. Using the process time axis as the matching benchmark, for each sampling time, the curing degree-process response baseline constructed by S200 is retrieved, and the standard exothermic power matching at each time is obtained by querying according to the curing reaction process, and arranged in chronological order to form a set of standard exothermic power time series values. For example, the set of standard exothermic power time series values ​​is [5s: 9.85J / s, ..., 300s: 30.56J / s, ..., 1795s: 28.96J / s].

[0070] Next, based on the measured heat release power and the standard heat release power at the same sampling time, the instantaneous heat release deviation value at each sampling time is determined. The instantaneous heat release deviation value refers to the deviation ratio of the measured heat release power relative to the standard heat release power at a single moment, using a normalized calculation method, thus eliminating evaluation errors caused by differences in power magnitude. The calculation formula is: Instantaneous heat release deviation value = (Measured heat release power - Standard heat release power) / Standard heat release power. By substituting the data one by one at the same sampling time, the calculation is completed point by point to obtain the instantaneous heat release deviation value at each moment throughout the entire time period.

[0071] For example, taking a sampling time of 300s as an example: the measured heat release power is 31.82 J / s, and the standard heat release power is 30.56 J / s; the instantaneous heat release deviation value... The same principle applies to the other sampling times. By substituting the corresponding measured and standard power, the instantaneous heat release deviation value at each time point can be calculated.

[0072] Finally, the instantaneous heat release deviation values ​​are integrated and accumulated along the process time axis to obtain the time-series heat release deviation integral value, which serves as the process fluctuation characteristic quantity. The time-series heat release deviation integral value refers to the summation of all instantaneous heat release deviation values ​​along the time axis throughout the entire curing cycle, equivalent to an integral operation under discrete time sequence. The process fluctuation characteristic quantity is a quantitative index used to uniformly characterize the overall process fluctuation degree. The larger the integral value, the more obvious the deviation of the actual curing process from the standard process, and the more severe the fluctuation.

[0073] Specifically, using a 5-second sampling interval as the accumulation step, the instantaneous heat release deviation values ​​at all valid sampling moments in the heating, isothermal, and cooling stages are sequentially summed to complete the deviation integral statistics for the entire process cycle. The final time-series heat release deviation integral value is then defined as the process fluctuation characteristic quantity. For example, by summing all instantaneous heat release deviation values ​​over the entire time period and continuously accumulating them, the 5-second deviation is 0.0386, the 300-second deviation is 0.0412, and the 600-second deviation is -0.0099. After accumulating all nodes, the time-series heat release deviation integral value for this curing process is 2.16, which is the process fluctuation characteristic quantity for this prepreg curing process.

[0074] In this embodiment of the invention, the cure degree-process response baseline is used as the standard reference to achieve precise temporal matching between measured exothermic data and standard theoretical data. By calculating the normalized instantaneous exothermic deviation, interference from differences in the magnitude of exothermic power at different reaction stages is avoided, resulting in a more objective deviation evaluation. The full-cycle time-axis integral accumulation method is adopted to expand the single-point instantaneous deviation into a quantitative index of overall fluctuation across the entire domain. This allows for intuitive and quantitative differentiation of the stability of the curing process under different batches and operating conditions, providing reliable quantitative parameters for subsequent cure degree correction and compensation, and accurate evaluation of prepreg curing performance.

[0075] S400: Based on the process fluctuation characteristic quantity and the fiber volume fraction of the target prepreg, combined with the curing degree-process response baseline, obtain the curing reaction rate constant, and predict the distribution of curing degree along the process time axis to obtain the predicted curing degree distribution characteristic parameters, wherein the predicted curing degree distribution characteristic parameters include at least the curing degree mean and the curing degree standard deviation.

[0076] In this embodiment of the invention, based on the process fluctuation characteristic quantity and the fiber volume fraction of the target prepreg, combined with the curing degree-process response baseline, the curing reaction rate constant is obtained, and the distribution of curing degree along the process time axis is predicted to obtain the predicted curing degree distribution characteristic parameters. These predicted curing degree distribution characteristic parameters include at least the mean curing degree and the standard deviation of curing degree. The resin crosslinking reaction rate of carbon fiber prepreg is simultaneously affected by both process fluctuations and material ratios: process fluctuations during curing cause the exothermic reaction to deviate from standard conditions, altering the speed of the resin crosslinking reaction; a higher fiber volume fraction results in a lower resin content, and the fiber skeleton inhibits resin molecule movement, producing a volume constraint effect and reducing the curing reaction rate. A single standard reaction rate constant cannot adapt to the varying chemical conditions in actual production; therefore, a dual correction factor is needed to correct the standard rate constant. Combined with gel point characteristic rate fusion optimization, the actual curing degree increment is solved interval by interval using a piecewise time-series recursive method, fully reproducing the curing degree change pattern throughout the entire cycle. Finally, the mean and standard deviation of curing degree are obtained through statistical calculation, quantitatively characterizing the overall curing level and curing uniformity, providing quantitative parameters for evaluating the curing performance of the prepreg.

[0077] Step S400 in the method provided in this embodiment of the invention includes: Specifically, based on the process fluctuation characteristics and the fiber volume fraction of the target prepreg, and combined with the degree of curing-process response baseline, the curing reaction rate constant is obtained, including: Extracting the standard curing reaction rate constant corresponding to the standard curing process from the curing degree-process response baseline includes: Obtain the exothermic power sequence and degree of cure sequence at each sampling time under the standard curing process; Based on the change in curing degree at adjacent sampling times and the sampling time interval in the curing degree sequence, the rate of change in curing degree at each sampling time is determined; By combining the exothermic power and the rate of change of degree of cure at the same sampling time, the apparent reaction rate constant corresponding to each sampling time is calculated; Extract the apparent reaction rate constants corresponding to all sampling times within the curing degree interval from zero point to gel point, and calculate the arithmetic mean as the standard curing reaction rate constant; A rate fluctuation correction factor is determined based on the process fluctuation characteristic quantity, wherein the magnitude of the rate fluctuation correction factor increases as the process fluctuation characteristic quantity increases; A volume effect correction factor is determined based on the fiber volume fraction, wherein the magnitude of the volume effect correction factor decreases as the fiber volume fraction increases; The standard curing reaction rate constant is corrected using the rate fluctuation correction factor and the volume effect correction factor to obtain the preliminary curing reaction rate constant; By combining the rate value corresponding to the gel point curing degree obtained from the curing degree-process response baseline, the initial curing reaction rate constant is fused to obtain the curing reaction rate constant.

[0078] First, the exothermic power sequence and cure degree sequence at each sampling time under the standard curing process are obtained. The standard exothermic power sequence and standard cure degree sequence refer to the set of reference exothermic parameters and cure degree parameters of the standard sample under the standard curing process, which are completely synchronized with the measured time axis. The cure degree-process response baseline constructed by S200 is retrieved, and the complete effective process time period is matched to extract the standard exothermic power sequence and standard cure degree sequence corresponding to each sampling time. For example, based on the standard parameters of the epoxy system mentioned above, the key nodes are extracted: 300s standard cure degree 41.2%, 600s standard cure degree 76.2%, and 1795s standard cure degree 94.8%, forming a continuous standard cure degree time sequence.

[0079] Secondly, based on the change in cure degree at adjacent sampling times and the sampling time interval in the cure degree sequence, the cure degree change rate at each sampling time is determined. The cure degree change rate refers to the increment of the resin's curing degree per unit time, reflecting the speed of the crosslinking reaction under standard process conditions. The calculation formula is: Curing degree change rate = (Curing degree at the current sampling time - Curing degree at the previous adjacent sampling time) / Sampling time interval. For example, if the sampling time interval is fixed at 5 seconds, and the standard operating condition at 300 seconds is selected, the standard cure degree at the previous adjacent 295 seconds is 39.6%; the cure degree change rate... .

[0080] Next, combining the exothermic power and the rate of change of curing degree at the same sampling time, the apparent reaction rate constant corresponding to each sampling time is calculated. The apparent reaction rate constant is a fundamental parameter characterizing the inherent kinetic properties of the resin crosslinking reaction, eliminating the influence of the current curing degree. The calculation formula is: Apparent reaction rate constant = Rate of change of curing degree / (1 − Curing degree at the current sampling time), calculated moment by moment. For example, if the current standard curing degree is 41.2% after 300 seconds, the apparent reaction rate constant is... .

[0081] Furthermore, the apparent reaction rate constants corresponding to all sampling times within the curing degree interval from zero to gel point are extracted, and their arithmetic mean is calculated as the standard curing reaction rate constant. The standard curing reaction rate constant is the average kinetic parameter within the stable curing reaction range, serving as the benchmark value for rate correction. The curing degree interval from zero to gel point represents the core reaction range from the resin's initial reaction to the critical gelation state. The apparent reaction rate constants at all sampling times from zero to gel point are selected, and their arithmetic mean is calculated and defined as the standard curing reaction rate constant. For example, with a curing degree of 2.0% at zero and a gel point of 58.5%, the average of multiple apparent rate constants within this interval yields the standard curing reaction rate constant. .

[0082] Subsequently, a rate fluctuation correction factor is determined based on the process fluctuation characteristic quantity, wherein the magnitude of the rate fluctuation correction factor increases with the increase of the process fluctuation characteristic quantity. The reference fluctuation baseline value refers to the fixed baseline integral value obtained by self-comparison and integration of the standard exothermic power sequence under standard process conditions. The rate fluctuation correction factor is used to compensate for reaction rate deviations caused by process disturbances, and its value increases synchronously with the increase of the process fluctuation characteristic quantity. The calculation formula is: Rate fluctuation correction factor = 1 + Process fluctuation characteristic quantity / Reference fluctuation baseline value. For example, if the process fluctuation characteristic quantity is 2.16, and the reference fluctuation baseline value for this epoxy system is calibrated to 1.20; the rate fluctuation correction factor = 1 + 2.16 / 1.20 = 2.80.

[0083] Furthermore, a volume effect correction factor is determined based on the fiber volume fraction, wherein the magnitude of the volume effect correction factor decreases as the fiber volume fraction increases. The volume correction coefficient is a resin system-specific experimental constant, ranging from 0 to 1, obtained through regression analysis of baseline calibration experiments. The volume effect correction factor characterizes the inhibitory effect of carbon fiber occupancy on the resin reaction, and its value decreases as the fiber volume fraction increases. The calculation formula is: Volume effect correction factor = 1 - Volume correction coefficient × Fiber volume fraction. For example, the volume correction coefficient for this general-purpose epoxy resin system is calibrated to 0.35, and the target prepreg fiber volume fraction is 65%; the volume effect correction factor = 1 - 0.35 × 0.65 = 0.7725.

[0084] Furthermore, the standard curing reaction rate constant is corrected using the rate fluctuation correction factor and the volume effect correction factor to obtain the preliminary curing reaction rate constant. Based on the standard curing reaction rate constant, a dual correction factor calculation is performed: Preliminary curing reaction rate constant = Standard curing reaction rate constant × Rate fluctuation correction factor × Volume effect correction factor. For example, the preliminary curing reaction rate constant... .

[0085] Finally, the initial curing reaction rate constant is obtained by combining the rate value corresponding to the gel point curing degree obtained from the curing degree-process response baseline with the initial curing rate value, thus obtaining the curing reaction rate constant. The first weighting coefficient and the second weighting coefficient are weighted fusion allocation parameters, and their sum is always equal to 1. The rate value corresponding to the gel point curing degree refers to the characteristic reaction rate under the critical state of resin gelation, characterizing the reaction transition properties. The weighting coefficients are set, and the weighted fusion formula is used: Curing reaction rate constant = Initial curing reaction rate constant × First weighting coefficient + Rate value corresponding to the gel point curing degree × Second weighting coefficient. For example, setting the first weighting coefficient to 0.8 and the second weighting coefficient to 0.2, the rate value corresponding to the gel point is... After fusion calculation, the final curing reaction rate constant is: .

[0086] The distribution of the degree of cure along the process time axis is predicted, resulting in characteristic parameters for the predicted degree of cure distribution, including: Starting from the zero point of curing degree, the curing process is divided into multiple continuous time intervals along the process time axis, and the length of each time interval is equal to the sampling time interval. Based on the degree of curing at the beginning of the current time interval, combined with the curing reaction rate constant and the standard curing reaction rate corresponding to the degree of curing-process response baseline, the actual curing reaction rate within the current time interval is determined. Based on the actual curing reaction rate and the sampling time interval, the degree of curing increment within the current time interval is obtained, and combined with the degree of curing at the beginning of the current time interval, the degree of curing at the beginning of the next time interval is obtained. Repeat the process of determining the actual curing reaction rate, calculating the degree of curing increment, and recursively calculating the degree of curing until the entire curing process timeline is traversed to generate a predicted degree of curing timeline curve. Calculate the arithmetic mean of the predicted cure degree corresponding to all time intervals on the predicted cure degree time series curve, and use it as the cure degree mean; The standard deviation of curing degree is calculated based on the degree of difference between the predicted curing degree and the mean curing degree in each time interval.

[0087] First, starting from the zero point of cure, the curing process is divided into multiple continuous time intervals along the process time axis, with the length of each time interval equal to the sampling time interval. For example, using the zero point of cure as the starting point for calculation, and based on a uniform sampling time interval of 5 seconds, the complete curing period is divided into several continuous, equally long independent time intervals, and the cure degree is calculated recursively segment by segment.

[0088] Secondly, based on the degree of cure at the start of the current time interval, combined with the curing reaction rate constant and the standard curing reaction rate corresponding to the degree of cure-process response baseline, the actual curing reaction rate within the current time interval is determined. The standard curing reaction rate refers to the increment of degree of cure per unit time corresponding to the standard baseline at the same degree of cure. The adjustment factor is the ratio of the actual rate constant to the standard rate constant, used to correct the theoretical rate.

[0089] Specifically, based on the initial degree of cure in the current interval, the cure degree-process response baseline is queried, and the corresponding standard curing reaction rate is matched; the ratio of the curing reaction rate constant to the standard curing reaction rate constant is calculated as an adjustment factor; the standard curing reaction rate is multiplied by the adjustment factor to obtain the actual curing reaction rate in the current interval. For example, if the initial degree of cure in a certain interval is 41.2%, the baseline standard curing reaction rate is 0.32% / s, and after adjustment, the actual curing reaction rate in this interval is 0.45% / s.

[0090] Next, based on the actual curing reaction rate and the sampling time interval, the degree of cure increment within the current time interval is obtained, and combined with the degree of cure at the beginning of the current time interval, the degree of cure at the beginning of the next time interval is obtained. According to the time-series recursive formula: Degree of cure at the beginning of the next time interval = Degree of cure at the beginning of the current time interval + Actual curing reaction rate × Sampling time interval. For example, if the initial degree of cure in the current interval is 41.2%, the actual curing reaction rate is 0.45% / s, and the sampling interval is 5s; the initial degree of cure in the next interval = 41.2% + 0.45% / s × 5s = 43.45%.

[0091] Then, the process of determining the actual curing reaction rate, calculating the degree of cure increment, and recursively updating the degree of cure is repeated until the entire curing process timeline is traversed, generating a predicted degree of cure time series curve. The operations of solving for the actual reaction rate, calculating the degree of cure increment, and recursively updating the degree of cure are repeated iteratively to complete the calculations for all time intervals in sequence. The predicted degree of cure values ​​at all times are then summarized, sorted by the timeline, and a continuous and complete predicted degree of cure time series curve is generated.

[0092] Subsequently, the arithmetic mean of the predicted cure rates for all time intervals on the predicted cure rate time-series curve is calculated as the cure rate mean. The cure rate mean refers to the average level of the predicted cure rate at all time points throughout the entire curing cycle, characterizing the overall curing degree of the prepreg. The predicted cure rates for all time intervals on the time-series curve are extracted, and the arithmetic mean is calculated as the cure rate mean. For example, averaging multiple sets of cure rate data over the entire time period yields a cure rate mean of 82.65% for this prepreg.

[0093] Finally, the standard deviation of the curing degree is calculated based on the degree of difference between the predicted curing degree and the mean curing degree within each time interval. The standard deviation of the curing degree characterizes the degree of dispersion of the curing degree relative to the overall mean at each time point; the smaller the value, the better the curing uniformity. The difference between the predicted curing degree and the mean curing degree at all times is statistically analyzed, and the square of each difference is calculated. The square root of the sum of the squares and the average is then taken to obtain the standard deviation of the curing degree. For example, based on discrete statistical calculations, the standard deviation of the curing degree in this curing process is 3.28%, indicating that the fluctuation range of the curing degree throughout the entire cycle is small, and the overall curing uniformity is good.

[0094] In this embodiment of the invention, based on standard curing kinetic parameters and combined with a dual correction mechanism of process fluctuations and fiber volume fraction, the curing reaction rate constant is accurately corrected to match the reaction characteristics under actual working conditions. Through segmented recursive calculations over equal time intervals, continuous prediction of the degree of curing throughout the entire process timeline is achieved, fully reconstructing the curing evolution law across the entire stages of heating, isothermal control, and cooling. The average degree of curing is used to evaluate the overall curing completion, and the standard deviation of the degree of curing is used to quantify curing uniformity, achieving multi-dimensional quantitative characterization of the curing state. This effectively compensates for the deficiency that exothermic parameters can only reflect thermal effects and cannot directly reflect the degree of curing, providing reliable kinetic and curing characteristic parameter support for subsequent prediction of composite material mechanical properties and optimization of curing processes.

[0095] S500: Input the predicted curing degree distribution characteristic parameters into the pre-constructed fatigue resistance performance mapping model to obtain fatigue resistance performance characterization values.

[0096] In this embodiment of the invention, the predicted curing degree distribution characteristic parameters are input into a pre-constructed fatigue resistance performance mapping model to obtain fatigue resistance performance characterization values. The mean curing degree determines the overall resin crosslinking perfection of the carbon fiber composite material, and the standard deviation of curing degree reflects the uniformity of curing across the entire domain. These two parameters directly affect the residual stress inside the material, the interlayer interface bonding strength, and the matrix density, and have a strong nonlinear correlation with the fatigue life of the component. Traditional mechanical testing has a long testing cycle and high cost, which cannot meet the needs of rapid evaluation in batch processes. This step uses the curing degree distribution characteristic parameters as input and fatigue performance indicators as output to build a multi-layer fully connected regression network. Through supervised training with experimental sample data, a quantitative nonlinear mapping relationship between curing parameters and fatigue resistance performance is established. Based on the trained mapping model, rapid and accurate prediction of fatigue resistance performance is achieved.

[0097] Step S500 in the method provided in this embodiment of the invention includes: The pre-construction step of the fatigue resistance performance mapping model includes: Multiple groups of prepreg samples with different mean curing degree and different standard deviation of curing degree were collected. Laminate samples were prepared for each group of samples according to the standard curing process, and fatigue tests were performed to obtain the measured fatigue life values ​​of each group of samples. The mean and standard deviation of curing degree of each sample group are used as input feature vectors, and the corresponding measured fatigue life values ​​are used as output labels to form a sample dataset. An initial regression mapping network is constructed with the nonlinear mapping relationship between the input feature vector and the output label as the objective, wherein the initial regression mapping network adopts a multi-layer fully connected structure; The mean squared error loss function is used to measure the deviation between the predicted fatigue life output by the network and the measured fatigue life. The internal weight parameters of the network are iteratively updated by backpropagation combined with an adaptive moment estimation optimizer. When the mean squared error loss of the network on the independent validation set no longer decreases after several consecutive iterations, training is stopped, and the trained regression mapping network is used as the fatigue resistance performance mapping model.

[0098] First, multiple groups of prepreg samples with different mean and standard deviations of degree of cure were collected. Laminate specimens were prepared for each group of samples according to a standard curing process, and fatigue tests were performed to obtain the measured fatigue life values ​​for each group. Differentiated prepreg samples refer to prepreg specimens with varying mean and standard deviations of degree of cure prepared by adjusting curing process parameters. Laminate specimens are standard composite material test pieces prepared using a hot-pressing process, used for standardized fatigue performance testing. The measured fatigue life value refers to the number of cycles under a set cyclic load condition before fatigue failure, serving as a quantitative label for fatigue resistance.

[0099] Specifically, a general-purpose epoxy prepreg consistent with the target part was selected, with a uniform fiber volume fraction of 65%. Multiple sets of blanks with different curing states were prepared by finely adjusting the heating rate and the isothermal duration. Each set of blanks was pressed into standard-sized laminate samples according to the same standard curing process. A high-frequency reciprocating fatigue testing machine was used to carry out constant amplitude cyclic load fatigue tests, and the number of cycles corresponding to the fracture failure of each set of samples was recorded as the measured fatigue life value.

[0100] For example, a total of 35 differentiated samples were prepared, covering different curing ranges of low, medium and high: the average degree of curing ranged from 65% to 92%, and the standard deviation of degree of curing ranged from 1.25% to 6.85%. The fatigue test showed that the fatigue life of the sample with excellent curing uniformity could reach 125,000 cycles, while the fatigue life of the sample with large curing fluctuations was only 63,000 cycles, thus forming gradient performance data.

[0101] Secondly, the mean and standard deviation of curing degree for each sample group are used as input feature vectors, and the corresponding measured fatigue life values ​​are used as output labels, forming a sample dataset. The input feature vector is a two-dimensional feature parameter composed of the mean and standard deviation of curing degree. The output label is the measured fatigue life value corresponding to a single sample. The sample dataset is a standardized data set where the input feature vectors and output labels are matched one-to-one, providing data support for training the fatigue resistance performance mapping model.

[0102] Specifically, taking a single sample as a unit, the mean and standard deviation of the degree of curing obtained from the test of each group of samples are combined into a two-dimensional input feature vector, and the corresponding measured fatigue life value is used as the unique output label. After all sample data are sorted, they are randomly divided into a training set and an independent validation set in a 7:3 ratio. Abnormal discrete data are removed, and the dataset is standardized and preprocessed.

[0103] For example, the format of a single sample data set is: input feature vector [82.65%, 3.28%], corresponding to output labels [108,200]. After dividing all 35 data sets, there are 24 training sets and 11 independent validation sets, ensuring that model training and performance validation are independent of each other.

[0104] Next, an initial regression mapping network is constructed with the nonlinear mapping relationship between the input feature vector and the output label as the objective. This initial regression mapping network employs a multi-layer fully connected structure. A multi-layer fully connected structure refers to a basic neural network structure where neurons within a layer are not connected, but neurons in adjacent layers are all interconnected.

[0105] Specifically, a four-layer fully connected regression network is constructed, and the number of neurons and activation functions of each layer are defined. The specific model structure is as follows: The input layer contains 2 neurons, corresponding to the mean of solidification degree and the standard deviation of solidification degree, respectively; the first hidden layer contains 16 neurons, and the activation function is ReLU, which is responsible for shallow nonlinear feature extraction; the second hidden layer contains 32 neurons, and the activation function is ReLU, which completes deep correlation feature fusion; the third hidden layer contains 16 neurons, and the activation function is ReLU, which realizes feature dimensionality reduction and optimization; the output layer contains 1 neuron, and the activation function is Linear activation, which directly outputs the continuous fatigue resistance performance characterization value.

[0106] Furthermore, the mean squared error loss function is used to measure the deviation between the predicted fatigue life value output by the network and the measured fatigue life value. Backpropagation combined with an adaptive moment estimation optimizer iteratively updates the network's internal weight parameters. The mean squared error loss function quantifies the deviation between the model's predicted values ​​and the experimental measured values; a smaller deviation indicates higher model fitting accuracy. The adaptive moment estimation optimizer (Adam) is a mainstream optimization algorithm that combines momentum and adaptive learning rate to accelerate network convergence and improve training stability. Backpropagation refers to the training mechanism that adjusts the network weights and bias parameters layer by layer based on the deviation of the loss function.

[0107] Specifically, the mean squared error loss function (MSE) is used as the network loss evaluation metric, and the Adam adaptive moment estimation optimizer is selected with an initial learning rate of 0.001. The predicted values ​​are output through forward propagation, the loss function value is calculated, and then the internal weights and bias parameters of the network are iteratively updated layer by layer through the backpropagation algorithm, continuously reducing the prediction bias. For example, the loss value in the first iteration is 12.65. After multiple rounds of parameter updates, the training set loss continues to decrease, the deviation between the predicted values ​​and the actual fatigue life gradually narrows, and the fitting accuracy steadily improves.

[0108] Finally, when the mean squared error loss of the network on the independent validation set no longer decreases after multiple consecutive iterations, training is stopped, and the trained regression mapping network is used as the fatigue-resistant performance mapping model. The independent validation set refers to a dedicated validation dataset that does not participate in weight update training, used to objectively evaluate the model's generalization ability. Iterative convergence refers to the training termination state where the model performance tends to stabilize and the loss no longer decreases.

[0109] Specifically, after each iteration during training, the mean squared error loss of the independent validation set is calculated synchronously. A convergence condition is set: if the mean squared error loss of the independent validation set no longer decreases after 15 consecutive iterations, the model training is considered complete, and iteration is immediately stopped. The optimal weights and structural parameters of the network are saved, and the trained and solidified multilayer fully connected network is defined as the fatigue resistance performance mapping model. For example, if the validation set loss stabilizes at 0.32 and shows no decreasing trend after 15 consecutive iterations when training reaches the 47th iteration, the convergence condition is met, training is terminated, and the model file is saved, resulting in the trained fatigue resistance performance mapping model.

[0110] Based on this, the predicted curing degree distribution characteristic parameters are input into a pre-constructed fatigue resistance performance mapping model to obtain fatigue resistance performance characterization values. The predicted curing degree distribution characteristic parameters calculated by S400, namely the curing degree mean and curing degree standard deviation, are organized into a standard input feature vector and imported into the pre-constructed fatigue resistance performance mapping model. Through internal forward computation, the model directly outputs continuous numerical results as the fatigue resistance performance characterization values ​​of the target prepreg. For example, inputting the characteristic parameters [82.65%, 3.28%] of the test piece into the fatigue resistance performance mapping model, the model outputs 107,600 fatigue resistance performance characterization values ​​after computation, achieving rapid quantitative prediction of fatigue performance from curing parameters.

[0111] In this embodiment of the invention, a supervised learning dataset is constructed based on actual experimental samples. A three-layer fully connected network structure with hidden layers is adopted, and the ReLU activation function is used to enhance the nonlinear fitting ability, accurately representing the complex relationship between the mean of curing degree, the standard deviation of curing degree, and the fatigue resistance of composite materials. Through the Adam optimizer and a strict validation set convergence mechanism, training efficiency and generalization ability are balanced, and the model prediction accuracy is stable and reliable. Relying on the pre-trained fatigue resistance mapping model, only the curing degree distribution feature parameters need to be input to quickly obtain the fatigue resistance characterization value, eliminating the need for complicated mechanical fatigue tests, improving the efficiency of curing process evaluation and material performance testing, and realizing an integrated correlation evaluation of curing quality and service performance.

[0112] S600: Calculate the curing degree deviation correction factor based on the average curing degree, and use the curing degree deviation correction factor to correct the fatigue resistance performance characterization value to obtain the final fatigue resistance performance evaluation result.

[0113] In the embodiments of the present invention, a curing degree deviation correction factor is calculated based on the average curing degree, and the anti-fatigue performance characterization value is corrected using the curing degree deviation correction factor to obtain the final anti-fatigue performance evaluation result. The anti-fatigue performance characterization value output by the anti-fatigue performance mapping model is mainly predicted based on the correlation between the average curing degree and the standard deviation, without fully considering the benchmark difference of the average curing degree itself. The lower the average curing degree, the more incomplete the resin cross-linking reaction, the more internal defects in the material, and the greater the deviation of the actual anti-fatigue performance corresponding to the same characterization value; on the contrary, the higher the average curing degree, the smaller the deviation between the characterization value and the actual performance. If the uncorrected characterization value is directly used as the evaluation result, it will lead to evaluation errors due to different curing degree benchmarks and cannot accurately reflect the true anti-fatigue level of the component. Therefore, it is necessary to calculate the deviation correction factor based on the average curing degree, calibrate the characterization value, eliminate the influence brought by the curing degree benchmark difference, and ensure the accuracy and reliability of the final evaluation result.

[0114] First, extract the average curing degree from the predicted curing degree distribution characteristic parameters calculated in S400. The curing degree deviation correction factor = 1 / average curing degree. The value of this factor decreases as the average curing degree increases, and can accurately compensate for the performance evaluation deviation under different curing degree benchmarks.

[0115] Secondly, retrieve the anti-fatigue performance characterization value output by the anti-fatigue performance mapping model in S500. The final anti-fatigue performance evaluation result = anti-fatigue performance characterization value × curing degree deviation correction factor. Complete the numerical correction calculation to obtain the final anti-fatigue performance evaluation result of the target prepreg component. This evaluation result can be used to quickly determine the quality grade of the prepreg component, guide the optimization and adjustment of the curing process parameters, and provide a quantitative basis for predicting the service life of the component, controlling the mass production quality, and selecting models for engineering applications without performing destructive fatigue tests in the later stage.

[0116] Exemplarily, the predicted average curing degree obtained in S400 is 82.65%, and the curing degree deviation correction factor . Retrieve the anti-fatigue performance characterization value output by S500 as 107,600 times. The final anti-fatigue performance evaluation result . This result is the final anti-fatigue performance evaluation value of the target epoxy prepreg component after curing degree deviation correction. Combining with the relevant industry standards, if the anti-fatigue life of this type of prepreg component is ≥ 100,000 times, it is judged as qualified, and ≥ 120,000 times is judged as excellent. Then this evaluation result indicates that the anti-fatigue performance of this component reaches the excellent level, and at the same time, it can guide the subsequent curing process to appropriately extend the constant temperature duration, further increase the average curing degree, and optimize the anti-fatigue performance of the component.

[0117] In this embodiment of the invention, the curing degree deviation correction factor effectively eliminates the fatigue performance evaluation deviation caused by different curing degree average benchmarks, solves the problem of the disconnect between model prediction values ​​and actual component performance, and makes the final fatigue performance evaluation results more consistent with the actual curing state and service performance of prepreg components, thus improving the evaluation accuracy. The correction process only needs to rely on existing curing degree averages and fatigue performance characterization values, with simple calculation logic and convenient operation, without the need for additional complex experiments, effectively reducing evaluation costs and time consumption. At the same time, it clarifies the multiple practical applications of the final evaluation results, realizing a closed-loop process from curing parameter detection, performance prediction, deviation correction to quality judgment and process optimization, providing accurate and efficient quantitative support for the batch production control, engineering application selection, and service safety prediction of prepreg components, further expanding the engineering application value and practicality of the entire evaluation method.

[0118] Through the specific implementation methods described above, the embodiments of the present invention achieve the following technical effects: This invention provides an intelligent evaluation method and system for the performance of carbon fiber prepregs. It employs multi-regional independent temperature acquisition and data optimization processing to accurately obtain exothermic parameters throughout the curing process, effectively reducing single-point detection errors and signal noise. Through standard sample calibration, a benchmark correlation between exothermic power and degree of curing is established, and the overall fluctuation of the curing process is quantitatively evaluated by comparing time-series parameters. Based on material ratios and operating condition disturbances, the curing reaction kinetic parameters are corrected, and the distribution law and uniformity index of the degree of curing are accurately predicted segmentally and recursively. Using a trained neural network model, a nonlinear mapping relationship between curing characteristics and fatigue resistance is constructed, eliminating the need for destructive mechanical testing and rapidly outputting performance prediction results. Simultaneously, a curing degree deviation correction mechanism is introduced to calibrate the performance prediction values, further improving the accuracy of the evaluation. This invention achieves integrated non-destructive testing and quantitative evaluation of the prepreg's curing state, process stability, and service performance, providing a reliable technical basis for curing process optimization, mass production quality control, and component performance prediction.

[0119] Example 2, as Figure 2 As shown, the present invention provides an intelligent evaluation system for the performance of carbon fiber prepregs, the system comprising: The response parameter acquisition module 11 is used to acquire the dynamic process response parameter sequence of the target prepreg during the curing process, wherein the dynamic process response parameter sequence includes the measured exothermic power of at least one characteristic temperature point distributed along the process time axis. The response baseline calling module 12 is used to call the corresponding curing degree-process response baseline according to the resin system type of the target prepreg, wherein the curing degree-process response baseline defines the mapping relationship between the exothermic power and the curing degree under the standard curing process. The fluctuation characteristic calculation module 13 is used to compare the dynamic process response parameter sequence with the curing degree-process response baseline on the corresponding time axis, and calculate the integral value of the time-series exothermic deviation as a process fluctuation characteristic quantity. The curing degree distribution prediction module 14 is used to obtain the curing reaction rate constant based on the process fluctuation characteristic quantity and the fiber volume fraction of the target prepreg, combined with the curing degree-process response baseline, and to predict the distribution of curing degree along the process time axis, thereby obtaining the predicted curing degree distribution characteristic parameters, wherein the predicted curing degree distribution characteristic parameters include at least the curing degree mean and the curing degree standard deviation. The fatigue performance mapping module 15 is used to input the predicted curing degree distribution characteristic parameters into the pre-constructed fatigue resistance performance mapping model to obtain fatigue resistance performance characterization values. The performance deviation correction module 16 is used to calculate the curing degree deviation correction factor based on the curing degree mean value, and use the curing degree deviation correction factor to correct the fatigue resistance performance characterization value to obtain the final fatigue resistance performance evaluation result.

[0120] In one embodiment, the response parameter acquisition module 11 is further configured to: Multiple temperature sensing nodes are arranged along the resin flow direction on the inner surface of the curing mold of the target prepreg, and the position of each temperature sensing node corresponds to a characteristic temperature point. The measured temperature values ​​of each temperature sensing node in the curing heating section, isothermal section and cooling section are continuously collected at a fixed sampling frequency to generate a temperature-time series of each characteristic temperature point. Based on the mass and specific heat capacity of the target prepreg, and combined with the temperature-time series of each characteristic temperature point, the instantaneous heat release power at each sampling time of each characteristic temperature point is calculated to obtain the instantaneous heat release power series of each characteristic temperature point. For the instantaneous heat release power sequence at the same characteristic temperature point, based on the power change rate between adjacent sampling times, power jump points caused by sensor noise are removed to obtain the corrected instantaneous heat release power. The average instantaneous heat release power of each characteristic temperature point at the same sampling time is taken and arranged in chronological order to form a sequence of measured heat release power distributed along the process time axis.

[0121] Specifically, based on the mass and specific heat capacity of the target prepreg, and combined with the temperature-time series of each characteristic temperature point, the instantaneous heat release power at each sampling time of each characteristic temperature point is calculated to obtain the instantaneous heat release power sequence of each characteristic temperature point, including: For each characteristic temperature point, the measured temperature value at the current sampling time is extracted from the temperature-time series of the corresponding characteristic temperature point; Extract the measured temperature value of the previous adjacent sampling time from the current sampling time, calculate the difference between the measured temperature value of the current sampling time and the measured temperature value of the previous adjacent sampling time, and combine it with the sampling time interval to obtain the forward temperature change rate at the sampling time. Extract the measured temperature value of the next adjacent sampling time from the sampling time, calculate the difference between the measured temperature value of the next adjacent sampling time and the measured temperature value, and combine it with the sampling time interval to obtain the backward temperature change rate at the sampling time. Calculate the difference between the forward temperature change rate and the backward temperature change rate, and combine it with the sampling time interval to obtain the temperature change acceleration at the sampling moment; The mass and specific heat capacity of the target prepreg are obtained, the heat capacity product is calculated, and the basic exothermic power is obtained by combining the forward temperature change rate. The basic heat release power is inertially corrected based on the heat capacity product and the temperature change acceleration to obtain the instantaneous heat release power at the sampling time; Iterate through all sampling times to obtain the instantaneous heat release power of the current characteristic temperature point at each sampling time; By iterating through all characteristic temperature points, the instantaneous heat release power sequence of each characteristic temperature point is obtained.

[0122] The process of obtaining the mass and specific heat capacity of the target prepreg and calculating the heat capacity product includes: Before curing, the target prepreg is placed on a weighing device, and three consecutive weighing readings are taken. The arithmetic mean of the three readings is calculated as the mass of the target prepreg. Based on the resin system type of the target prepreg, the corresponding fiber specific heat capacity and resin specific heat capacity are extracted from the pre-stored material property database. Obtain the fiber volume fraction of the target prepreg and calculate the product of the fiber volume fraction and the fiber specific heat capacity as the fiber heat capacity contribution value; Based on the fiber volume fraction and the resin specific heat capacity, the resin heat capacity contribution value is calculated, and combined with the fiber heat capacity contribution value, the specific heat capacity of the target prepreg is obtained. The heat capacity product is determined based on the mass of the target prepreg and the specific heat capacity.

[0123] In one embodiment, the response baseline call module 12 is further configured to: The steps for constructing the curing degree-process response baseline include: Prepare a standard sample with the same resin system type and the same fiber volume fraction as the target prepreg, place the standard sample in a differential scanning calorimeter, and heat it according to the standard curing process. During the curing process, the real-time exothermic power of the standard sample is continuously collected, and the measured value of the degree of curing is extracted from the differential scanning calorimeter at fixed time intervals to obtain the discrete point pair of exothermic power-degree of curing. The discrete points of the exothermic power-degree of cure were sorted, and a scatter plot was drawn with exothermic power as the abscissa and degree of cure as the ordinate. From the scatter plot, the curing degree value before the heat release power starts to rise continuously is selected as the curing degree zero point, and the curing degree value corresponding to the point where the heat release power reaches the peak value and then drops to half of the peak value is selected as the gel point curing degree. Adjacent discrete point pairs are connected using a linear interpolation method to form a continuous curve, which serves as the baseline for the degree of curing-process response.

[0124] In one embodiment, the fluctuation characteristic calculation module 13 is further configured to: The measured heat release power at each sampling moment is extracted from the dynamic process response parameter sequence to form a set of measured heat release power time series values; Based on the curing degree-process response baseline, the standard exothermic power corresponding to each sampling time under the standard curing process is obtained, forming a set of standard exothermic power time series values. Based on the measured heat release power and the standard heat release power at the same sampling time, the instantaneous heat release deviation value at each sampling time is determined; The instantaneous heat release deviation value is integrated and accumulated along the process time axis to obtain the time-series heat release deviation integral value, which is used as the process fluctuation characteristic quantity.

[0125] In one embodiment, the curing degree distribution prediction module 14 is further configured to: Specifically, based on the process fluctuation characteristics and the fiber volume fraction of the target prepreg, and combined with the degree of curing-process response baseline, the curing reaction rate constant is obtained, including: Extracting the standard curing reaction rate constant corresponding to the standard curing process from the curing degree-process response baseline includes: Obtain the exothermic power sequence and degree of cure sequence at each sampling time under the standard curing process; Based on the change in curing degree at adjacent sampling times and the sampling time interval in the curing degree sequence, the rate of change in curing degree at each sampling time is determined; By combining the exothermic power and the rate of change of degree of cure at the same sampling time, the apparent reaction rate constant corresponding to each sampling time is calculated; Extract the apparent reaction rate constants corresponding to all sampling times within the curing degree interval from zero point to gel point, and calculate the arithmetic mean as the standard curing reaction rate constant; A rate fluctuation correction factor is determined based on the process fluctuation characteristic quantity, wherein the magnitude of the rate fluctuation correction factor increases as the process fluctuation characteristic quantity increases; A volume effect correction factor is determined based on the fiber volume fraction, wherein the magnitude of the volume effect correction factor decreases as the fiber volume fraction increases; The standard curing reaction rate constant is corrected using the rate fluctuation correction factor and the volume effect correction factor to obtain the preliminary curing reaction rate constant; By combining the rate value corresponding to the gel point curing degree obtained from the curing degree-process response baseline, the initial curing reaction rate constant is fused to obtain the curing reaction rate constant.

[0126] The distribution of the degree of cure along the process time axis is predicted, resulting in characteristic parameters for the predicted degree of cure distribution, including: Starting from the zero point of curing degree, the curing process is divided into multiple continuous time intervals along the process time axis, and the length of each time interval is equal to the sampling time interval. Based on the degree of curing at the beginning of the current time interval, combined with the curing reaction rate constant and the standard curing reaction rate corresponding to the degree of curing-process response baseline, the actual curing reaction rate within the current time interval is determined. Based on the actual curing reaction rate and the sampling time interval, the degree of curing increment within the current time interval is obtained, and combined with the degree of curing at the beginning of the current time interval, the degree of curing at the beginning of the next time interval is obtained. Repeat the process of determining the actual curing reaction rate, calculating the degree of curing increment, and recursively calculating the degree of curing until the entire curing process timeline is traversed to generate a predicted degree of curing timeline curve. Calculate the arithmetic mean of the predicted cure degree corresponding to all time intervals on the predicted cure degree time series curve, and use it as the cure degree mean; The standard deviation of curing degree is calculated based on the degree of difference between the predicted curing degree and the mean curing degree in each time interval.

[0127] In one embodiment, the fatigue performance mapping module 15 is further configured to: The pre-construction step of the fatigue resistance performance mapping model includes: Multiple groups of prepreg samples with different mean curing degree and different standard deviation of curing degree were collected. Laminate samples were prepared for each group of samples according to the standard curing process, and fatigue tests were performed to obtain the measured fatigue life values ​​of each group of samples. The mean and standard deviation of curing degree of each sample group are used as input feature vectors, and the corresponding measured fatigue life values ​​are used as output labels to form a sample dataset. An initial regression mapping network is constructed with the nonlinear mapping relationship between the input feature vector and the output label as the objective, wherein the initial regression mapping network adopts a multi-layer fully connected structure; The mean squared error loss function is used to measure the deviation between the predicted fatigue life output by the network and the measured fatigue life. The internal weight parameters of the network are iteratively updated by backpropagation combined with an adaptive moment estimation optimizer. When the mean squared error loss of the network on the independent validation set no longer decreases after several consecutive iterations, training is stopped, and the trained regression mapping network is used as the fatigue resistance performance mapping model.

[0128] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent evaluation of the performance of carbon fiber prepreg, characterized in that, The method includes: Obtain the dynamic process response parameter sequence of the target prepreg during the curing process, wherein the dynamic process response parameter sequence includes the measured exothermic power of at least one characteristic temperature point distributed along the process time axis. Based on the resin system type of the target prepreg, the corresponding curing degree-process response baseline is invoked, wherein the curing degree-process response baseline defines the mapping relationship between exothermic power and curing degree under standard curing process; The dynamic process response parameter sequence is compared with the curing degree-process response baseline on the corresponding time axis, and the integral value of the time-series exothermic deviation is calculated as a process fluctuation characteristic quantity. Based on the process fluctuation characteristics and the fiber volume fraction of the target prepreg, combined with the curing degree-process response baseline, the curing reaction rate constant is obtained, and the distribution of curing degree along the process time axis is predicted to obtain the predicted curing degree distribution characteristic parameters, wherein the predicted curing degree distribution characteristic parameters include at least the curing degree mean and the curing degree standard deviation. The predicted curing degree distribution characteristic parameters are input into the pre-constructed fatigue resistance performance mapping model to obtain fatigue resistance performance characterization values; The curing degree deviation correction factor is calculated based on the average curing degree value. The fatigue resistance performance characterization value is then corrected using the curing degree deviation correction factor to obtain the final fatigue resistance performance evaluation result.

2. The intelligent evaluation method for the performance of carbon fiber prepreg according to claim 1, characterized in that, Obtain the dynamic process response parameter sequence of the target prepreg during the curing process, including: Multiple temperature sensing nodes are arranged along the resin flow direction on the inner surface of the curing mold of the target prepreg, and the position of each temperature sensing node corresponds to a characteristic temperature point. The measured temperature values ​​of each temperature sensing node in the curing heating section, isothermal section and cooling section are continuously collected at a fixed sampling frequency to generate a temperature-time series of each characteristic temperature point. Based on the mass and specific heat capacity of the target prepreg, and combined with the temperature-time series of each characteristic temperature point, the instantaneous heat release power at each sampling time of each characteristic temperature point is calculated to obtain the instantaneous heat release power series of each characteristic temperature point. For the instantaneous heat release power sequence at the same characteristic temperature point, based on the power change rate between adjacent sampling times, power jump points caused by sensor noise are removed to obtain the corrected instantaneous heat release power. The average instantaneous heat release power of each characteristic temperature point at the same sampling time is taken and arranged in chronological order to form a sequence of measured heat release power distributed along the process time axis.

3. The intelligent evaluation method for the performance of carbon fiber prepreg according to claim 2, characterized in that, Based on the mass and specific heat capacity of the target prepreg, and combined with the temperature-time series of each characteristic temperature point, the instantaneous heat release power at each sampling time of each characteristic temperature point is calculated to obtain the instantaneous heat release power sequence of each characteristic temperature point, including: For each characteristic temperature point, the measured temperature value at the current sampling time is extracted from the temperature-time series of the corresponding characteristic temperature point; Extract the measured temperature value of the previous adjacent sampling time from the current sampling time, calculate the difference between the measured temperature value of the current sampling time and the measured temperature value of the previous adjacent sampling time, and combine it with the sampling time interval to obtain the forward temperature change rate at the sampling time. Extract the measured temperature value of the next adjacent sampling time from the sampling time, calculate the difference between the measured temperature value of the next adjacent sampling time and the measured temperature value, and combine it with the sampling time interval to obtain the backward temperature change rate at the sampling time. Calculate the difference between the forward temperature change rate and the backward temperature change rate, and combine it with the sampling time interval to obtain the temperature change acceleration at the sampling moment; The mass and specific heat capacity of the target prepreg are obtained, the heat capacity product is calculated, and the basic exothermic power is obtained by combining the forward temperature change rate. The basic heat release power is inertially corrected based on the heat capacity product and the temperature change acceleration to obtain the instantaneous heat release power at the sampling time; Iterate through all sampling times to obtain the instantaneous heat release power of the current characteristic temperature point at each sampling time; By iterating through all characteristic temperature points, the instantaneous heat release power sequence of each characteristic temperature point is obtained.

4. The intelligent evaluation method for the performance of carbon fiber prepreg according to claim 3, characterized in that, Obtain the mass and specific heat capacity of the target prepreg, and calculate the heat capacity product, including: Before curing, the target prepreg is placed on a weighing device, and three consecutive weighing readings are taken. The arithmetic mean of the three readings is calculated as the mass of the target prepreg. Based on the resin system type of the target prepreg, the corresponding fiber specific heat capacity and resin specific heat capacity are extracted from the pre-stored material property database. Obtain the fiber volume fraction of the target prepreg and calculate the product of the fiber volume fraction and the fiber specific heat capacity as the fiber heat capacity contribution value; Based on the fiber volume fraction and the resin specific heat capacity, the resin heat capacity contribution value is calculated, and combined with the fiber heat capacity contribution value, the specific heat capacity of the target prepreg is obtained. The heat capacity product is determined based on the mass of the target prepreg and the specific heat capacity.

5. The intelligent evaluation method for the performance of carbon fiber prepreg according to claim 1, characterized in that, The steps for constructing the cure degree-process response baseline include: Prepare a standard sample with the same resin system type and the same fiber volume fraction as the target prepreg, place the standard sample in a differential scanning calorimeter, and heat it according to the standard curing process. During the curing process, the real-time exothermic power of the standard sample is continuously collected, and the measured value of the degree of curing is extracted from the differential scanning calorimeter at fixed time intervals to obtain the discrete point pair of exothermic power-degree of curing. The discrete points of the exothermic power-degree of cure were sorted, and a scatter plot was drawn with exothermic power as the abscissa and degree of cure as the ordinate. From the scatter plot, the curing degree value before the heat release power starts to rise continuously is selected as the curing degree zero point, and the curing degree value corresponding to the point where the heat release power reaches the peak value and then drops to half of the peak value is selected as the gel point curing degree. Adjacent discrete point pairs are connected using a linear interpolation method to form a continuous curve, which serves as the baseline for the degree of curing-process response.

6. The intelligent evaluation method for the performance of carbon fiber prepreg according to claim 1, characterized in that, The dynamic process response parameter sequence is compared with the cure degree-process response baseline on the corresponding time axis, and the integral value of the time-series exothermic deviation is calculated as a process fluctuation characteristic quantity, including: The measured heat release power at each sampling moment is extracted from the dynamic process response parameter sequence to form a set of measured heat release power time series values; Based on the curing degree-process response baseline, the standard exothermic power corresponding to each sampling time under the standard curing process is obtained, forming a set of standard exothermic power time series values. Based on the measured heat release power and the standard heat release power at the same sampling time, the instantaneous heat release deviation value at each sampling time is determined; The instantaneous heat release deviation value is integrated and accumulated along the process time axis to obtain the time-series heat release deviation integral value, which is used as the process fluctuation characteristic quantity.

7. The intelligent evaluation method for the performance of carbon fiber prepreg according to claim 1, characterized in that, Based on the process fluctuation characteristics and the fiber volume fraction of the target prepreg, and combined with the degree of curing-process response baseline, the curing reaction rate constant is obtained, including: Extracting the standard curing reaction rate constant corresponding to the standard curing process from the curing degree-process response baseline includes: Obtain the exothermic power sequence and degree of cure sequence at each sampling time under the standard curing process; Based on the change in curing degree at adjacent sampling times and the sampling time interval in the curing degree sequence, the rate of change in curing degree at each sampling time is determined; By combining the exothermic power and the rate of change of degree of cure at the same sampling time, the apparent reaction rate constant corresponding to each sampling time is calculated; Extract the apparent reaction rate constants corresponding to all sampling times within the curing degree interval from zero point to gel point, and calculate the arithmetic mean as the standard curing reaction rate constant; A rate fluctuation correction factor is determined based on the process fluctuation characteristic quantity, wherein the magnitude of the rate fluctuation correction factor increases as the process fluctuation characteristic quantity increases; A volume effect correction factor is determined based on the fiber volume fraction, wherein the magnitude of the volume effect correction factor decreases as the fiber volume fraction increases; The standard curing reaction rate constant is corrected using the rate fluctuation correction factor and the volume effect correction factor to obtain the preliminary curing reaction rate constant; By combining the rate value corresponding to the gel point curing degree obtained from the curing degree-process response baseline, the initial curing reaction rate constant is fused to obtain the curing reaction rate constant.

8. The intelligent evaluation method for the performance of carbon fiber prepreg according to claim 1, characterized in that, The distribution of curing degree along the process time axis is predicted, and the characteristic parameters of the predicted curing degree distribution are obtained, including: Starting from the zero point of curing degree, the curing process is divided into multiple continuous time intervals along the process time axis, and the length of each time interval is equal to the sampling time interval. Based on the degree of curing at the beginning of the current time interval, combined with the curing reaction rate constant and the standard curing reaction rate corresponding to the degree of curing-process response baseline, the actual curing reaction rate within the current time interval is determined. Based on the actual curing reaction rate and the sampling time interval, the degree of curing increment within the current time interval is obtained, and combined with the degree of curing at the beginning of the current time interval, the degree of curing at the beginning of the next time interval is obtained. Repeat the process of determining the actual curing reaction rate, calculating the degree of curing increment, and recursively calculating the degree of curing until the entire curing process timeline is traversed to generate a predicted degree of curing timeline curve. Calculate the arithmetic mean of the predicted cure degree corresponding to all time intervals on the predicted cure degree time series curve, and use it as the cure degree mean; The standard deviation of curing degree is calculated based on the degree of difference between the predicted curing degree and the mean curing degree in each time interval.

9. The intelligent evaluation method for the performance of carbon fiber prepreg according to claim 1, characterized in that, The pre-construction steps of the fatigue resistance performance mapping model include: Multiple groups of prepreg samples with different mean curing degree and different standard deviation of curing degree were collected. Laminate samples were prepared for each group of samples according to the standard curing process, and fatigue tests were performed to obtain the measured fatigue life values ​​of each group of samples. The mean and standard deviation of curing degree of each sample group are used as input feature vectors, and the corresponding measured fatigue life values ​​are used as output labels to form a sample dataset. An initial regression mapping network is constructed with the nonlinear mapping relationship between the input feature vector and the output label as the objective, wherein the initial regression mapping network adopts a multi-layer fully connected structure; The mean squared error loss function is used to measure the deviation between the predicted fatigue life output by the network and the measured fatigue life. The internal weight parameters of the network are iteratively updated by backpropagation combined with an adaptive moment estimation optimizer. When the mean squared error loss of the network on the independent validation set no longer decreases after several consecutive iterations, training is stopped, and the trained regression mapping network is used as the fatigue resistance performance mapping model.

10. A smart evaluation system for the performance of carbon fiber prepreg, characterized in that, The system is used to implement the intelligent performance evaluation method for carbon fiber prepreg according to any one of claims 1-9, the system comprising: The response parameter acquisition module is used to acquire the dynamic process response parameter sequence of the target prepreg during the curing process, wherein the dynamic process response parameter sequence includes the measured exothermic power of at least one characteristic temperature point distributed along the process time axis. The response baseline calling module is used to call the corresponding curing degree-process response baseline according to the resin system type of the target prepreg, wherein the curing degree-process response baseline defines the mapping relationship between exothermic power and curing degree under standard curing process; The fluctuation characteristic calculation module is used to compare the dynamic process response parameter sequence with the curing degree-process response baseline on the corresponding time axis, and calculate the integral value of the time-series exothermic deviation as a process fluctuation characteristic quantity. The curing degree distribution prediction module is used to obtain the curing reaction rate constant based on the process fluctuation characteristic quantity and the fiber volume fraction of the target prepreg, combined with the curing degree-process response baseline, and to predict the distribution of curing degree along the process time axis, thereby obtaining the predicted curing degree distribution characteristic parameters, wherein the predicted curing degree distribution characteristic parameters include at least the curing degree mean and the curing degree standard deviation. The fatigue performance mapping module is used to input the predicted curing degree distribution characteristic parameters into the pre-constructed fatigue resistance performance mapping model to obtain fatigue resistance performance characterization values. The performance deviation correction module is used to calculate the curing degree deviation correction factor based on the average curing degree, and use the curing degree deviation correction factor to correct the fatigue resistance performance characterization value to obtain the final fatigue resistance performance evaluation result.

Images