A resin liquid oxygen compatibility prediction method based on element content and toughness coordination
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SUZHOU LABORATORY
- Filing Date
- 2026-05-14
- Publication Date
- 2026-06-12
Smart Images
Figure CN122193279A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and specifically to a method for predicting the compatibility of resin with liquid oxygen based on the synergy of element content and toughness. Background Technology
[0002] Liquid oxygen compatibility evaluation of resin materials is a crucial technical step in determining their applicability in aerospace propulsion systems, cryogenic storage and transportation equipment, and structural components for high-energy oxidizing environments. The liquid oxygen environment is characterized by strong oxidizing properties coupled with cryogenic temperatures; under impact or friction conditions, materials may experience dangerous reactions such as combustion or explosion due to localized energy concentration. Therefore, liquid oxygen compatibility is typically evaluated using liquid oxygen impact testing. This involves applying a specified energy mechanical impact to resin samples under liquid oxygen immersion conditions using specialized testing equipment, simulating actual service conditions, and observing whether ignition, combustion, or explosion occurs. The result is then used to determine whether the test is "passed" or "failed." The evaluation results directly relate to the engineering suitability of the resin material and are an important basis for formulation selection and engineering application decisions.
[0003] Currently, commonly used resin materials in liquid oxygen environments include thermosetting resins, thermoplastic resins, and their modified systems. During resin formulation development and high-throughput screening, various functional elements, such as phosphorus, nitrogen, silicon, fluorine, aluminum, and sulfur, are often introduced into the system to achieve purposes such as flame retardancy, charring, inhibition of energy concentration, or improvement of mechanical properties. However, the compatibility of resins with liquid oxygen is not a linear function of the content of a single element, but rather a coupling result between elemental composition, distribution, and mechanical properties such as material toughness. Different elements may exhibit synergistic or antagonistic effects, and the material's fracture toughness, impact energy absorption capacity, and other mechanical parameters significantly influence the transfer and dissipation of transient impact energy, thereby affecting reaction behavior in liquid oxygen environments.
[0004] On the other hand, liquid oxygen impact testing is inherently dangerous, costly, and time-consuming. A single test can only provide a discrete "pass or fail" result, making it difficult to provide continuous quantitative trend information. In the process of screening multiple formulations, relying solely on single-factor indicators (such as the content of a single element or a single toughness parameter) for evaluation can easily overlook the coupling effect of multiple factors, leading to unclear directions for formulation optimization and even misjudgments. Furthermore, traditional static evaluation methods (such as element content detection, isolated mechanical property testing, or direct liquid oxygen impact assessment) cannot reveal the nonlinear synergistic relationship between elemental composition and material toughness, nor can they establish compatibility prediction models under the coupling effect of multiple factors, thus limiting the efficiency and accuracy of formulation design.
[0005] Therefore, existing methods for evaluating the compatibility of resins with liquid oxygen still have significant shortcomings in terms of multi-factor collaborative modeling capabilities, trend prediction capabilities, and high-throughput screening applicability. There is an urgent need for a liquid oxygen compatibility prediction method that can be based on the collaborative analysis of elemental content and toughness parameters, so as to reduce the number of actual liquid oxygen impact tests while achieving quantitative prediction and risk assessment of compatibility results. Summary of the Invention
[0006] To address the technical problem of low compatibility prediction accuracy caused by the inability of static evaluation methods to capture the synergistic effect of multiple factors, the present invention aims to provide a resin-liquid oxygen compatibility prediction method based on the synergy of element content and toughness. The specific technical solution adopted is as follows: Obtain elemental content data and impact toughness data of the resin-cured sample in multiple dimensions; Based on the multi-dimensional feature interaction relationship between element content data and impact toughness data in each dimension, a collaborative feature vector of resin-cured samples is constructed; based on the test results of resin-cured samples, label data of resin-cured samples is obtained. Based on the collaborative feature vector and label data of the resin-cured sample, the pass probability of the resin-cured sample is obtained using a data model, and the detection result of the resin-cured sample is judged.
[0007] Preferably, the construction of a collaborative feature vector for the resin-cured sample based on the multi-dimensional feature interaction relationship between elemental content data and impact toughness data in each dimension specifically includes: Based on the coupling relationship between the element content data and impact toughness data of each dimension, the first synergistic term of the element content data of each dimension is determined; Based on the ratio relationship between element content data from different dimensions, the second synergistic term is determined; The third synergistic term is determined based on the superposition relationship between element content data from different dimensions; The co-functional feature vector of the resin-cured sample includes elemental content data and impact toughness data in all dimensions, a first co-functional term, a second co-functional term, and a third co-functional term.
[0008] Preferably, determining the first synergistic term of the elemental content data for each dimension based on the coupling relationship between the elemental content data and the impact toughness data for each dimension specifically includes: The product of the element content data and the impact toughness data for each dimension is taken as the first co-term of the element content data for each dimension.
[0009] Preferably, determining the second synergistic term based on the ratio relationship between element content data of different dimensions specifically includes: The second synergistic term is determined based on the ratio between the element content data of each two different dimensions and the ratio between the element content data of each dimension and the sum of the element content data of the two dimensions.
[0010] Preferably, determining the third synergistic term based on the superposition relationship between element content data of different dimensions specifically includes: The sum of the element content data between any two different dimensions is used as the third co-term.
[0011] Preferably, obtaining the label data of the resin-cured sample based on the test results of the resin-cured sample specifically includes: A liquid oxygen shock compatibility test is performed on the resin-cured sample. If the test passes, the label data of the resin-cured sample is set to the first value; if the test fails, the label data of the resin-cured sample is set to the second value.
[0012] Preferably, the determination of the test results of the resin-cured sample specifically includes: If the pass probability of the resin-cured sample is greater than or equal to the preset test threshold, the resin-cured sample passes the test. If the pass probability of the resin-cured sample is less than the preset test threshold, the resin-cured sample fails the test.
[0013] Preferably, the elemental content data includes: The content of phosphorus, nitrogen, silicon, fluorine, aluminum, and sulfur.
[0014] The embodiments of the present invention have at least the following beneficial effects: This invention first establishes a standardized composition-mechanical dual-dimensional data base, providing accurate raw data support for subsequent feature construction. It establishes the original correlation between elemental systems and toughness properties, laying a solid data foundation for collaborative feature mining.
[0015] Then, based on the coupling, ratio, and superposition relationship between element content and impact toughness, a high-dimensional vector containing basic features and three types of synergistic features is constructed. This decouples the aliasing problem of single element content and isolated toughness indices, generating feature and label data that can be directly called by the machine learning model. Finally, the model obtains the probability of liquid oxygen compatibility of samples, enabling quantitative prediction and risk management of resin liquid oxygen compatibility. This replaces the traditional binary judgment of pass or fail, significantly reducing the risk of misjudgment and providing an efficient quantitative basis for high-throughput screening of resin formulations. Attached Figure Description
[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a flowchart of a method for predicting the compatibility of resin with liquid oxygen based on the synergy of element content and toughness. Detailed Implementation
[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a resin liquid oxygen compatibility prediction method based on the synergy of elemental content and toughness proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0020] The following description, in conjunction with the accompanying drawings, details a specific scheme for predicting the compatibility of resin with liquid oxygen based on the synergy of element content and toughness provided by this invention.
[0021] Please see Figure 1 The diagram illustrates a flowchart of a method for predicting the compatibility of resin with liquid oxygen based on the synergy of elemental content and toughness, according to an embodiment of the present invention. The method includes the following steps: Step S100: Obtain elemental content data and impact toughness data of the resin-cured sample in multiple dimensions.
[0022] First, take the resin system to be evaluated (thermosetting resin and its modified system) and prepare standard cured samples; the curing conditions are as follows: initial curing at 80–150℃ for 0.5–2h; then curing at 150–200℃ for 1–4h.
[0023] Then, the elemental content of the resin-cured sample was determined to obtain multi-dimensional elemental content data. Specifically, each dimension corresponds to one type of element, and the elemental content data includes phosphorus content, nitrogen content, silicon content, fluorine content, aluminum content, and sulfur content. The units are all mass fractions, which refer to the percentage ratio between the mass of each type of element and the total mass of the resin-cured sample.
[0024] More specifically, phosphorus, nitrogen, silicon, and aluminum were detected using X-ray fluorescence spectroscopy (XRF) or inductively coupled plasma optical emission spectroscopy (ICP-OES); sulfur and fluorine were detected using ion chromatography (IC) or combustion absorption-ion selective electrode method.
[0025] It should be noted that, depending on the instrument used to detect different elements, an appropriate method should be adaptively selected, such as acid digestion, alkali fusion, or combustion absorption, to pretreat the resin-cured sample so that it can be accurately identified by the detection instrument. Acid digestion involves using strong acids (such as nitric acid, hydrochloric acid, and hydrofluoric acid) to decompose the sample under heating conditions, which can destroy the organic structure of the resin and convert elements such as phosphorus (P), silicon (Si), and aluminum (Al) into soluble ionic forms. Alkali fusion involves melting the sample at high temperatures using an alkaline flux (such as sodium carbonate or sodium peroxide), suitable for processing insoluble resin components that are difficult to decompose by acid digestion, such as some high-temperature resistant resins containing ceramic fillers. Combustion absorption involves fully burning the resin sample, converting sulfur (S) and fluorine (F) into gaseous compounds, and then absorbing these gaseous substances with a specific absorbent, converting them into ionic forms in solution.
[0026] Different detection instruments have different requirements for sample morphology. For example, X-ray fluorescence spectroscopy (XRF) or inductively coupled plasma optical emission spectroscopy (ICP-OES) for measuring phosphorus (P), nitrogen (N), silicon (Si), and aluminum (Al) requires samples that are soluble salt solutions or fused sections; while ion chromatography (IC) for measuring sulfur (S) and fluorine (F) requires samples that are ionic solutions. Pretreatment methods are used to match the instrument requirements and ensure the accuracy of subsequent elemental content detection results.
[0027] Furthermore, impact test specimens are prepared from the resin-cured sample and impact toughness tests are performed. The impact toughness testing method is a well-known technique and will not be elaborated here. For example, a simple beam impact test (Charpy) or a cantilever beam impact test (Izod) can be used. Specifically, the average impact toughness of a certain number of test specimens can be used as the impact toughness data of the resin-cured sample. The number of test specimens can be at least 5, and the implementer can set this number according to the specific implementation scenario.
[0028] Step S200: Based on the multi-dimensional feature interaction relationship between the element content data and impact toughness data of each dimension, construct the collaborative feature vector of the resin-cured sample; based on the test results of the resin-cured sample, obtain the label data of the resin-cured sample.
[0029] The main purpose of this step is to decouple the aliasing problem of single element content and isolated toughness index in the evaluation of resin liquid oxygen compatibility. By quantifying the degree of interaction and correlation of multi-dimensional features, a feature vector that can characterize the synergistic effect of elements and toughness is constructed and a true value label is assigned, providing reliable input for the accurate prediction of subsequent machine learning models.
[0030] The oxygen compatibility of resin liquid is determined by both the elemental system and the toughness index. There is a significant nonlinear coupling effect between the two. Relying solely on the content of a single element or independent toughness data cannot accurately capture its intrinsic mechanism. This can easily lead to misjudgments such as the element content meeting the standard but the toughness being insufficient, or the element ratio being unbalanced, resulting in biased results.
[0031] Therefore, this step requires a hierarchical association construction and truth value calibration to complete the output of features and labels. First, based on the coupling enhancement relationship between element content data and impact toughness data in each dimension, the first synergistic term reflecting the synergistic gain effect of element-toughness is determined; then, based on the proportion and balance relationship between element content data in different dimensions, the second synergistic term reflecting the relative proportion characteristics of the element system is determined; then, based on the superposition and complementarity relationship between element content data in different dimensions, the third synergistic term characterizing the comprehensive effect of multi-element interaction is determined; finally, the element content data, impact toughness data, and the three types of synergistic terms in all dimensions are integrated to form a complete synergistic feature vector of the resin-cured sample, and based on the measured results of the liquid oxygen impact test of the resin-cured sample, the corresponding truth value label data for compatibility pass or failure is calibrated.
[0032] The first step is to determine the first synergistic term of the element content data for each dimension based on the coupling relationship between the element content data and the impact toughness data for each dimension.
[0033] More specifically, the product of the element content data and the impact toughness data for each dimension is taken as the first co-term of the element content data for each dimension.
[0034] As a concrete example, the first co-term of the element content data for each dimension can be represented as follows: , , , , , ,in, , , , , These represent the elemental content data for phosphorus (P), nitrogen (N), silicon (Si), fluorine (F), aluminum (Al), and sulfur (S) under the corresponding dimensions. This represents impact toughness data.
[0035] The first synergistic term reflects the coupled enhancement effect between the content of each element and toughness. The higher the content of a certain element and the better the material toughness, the stronger the improvement effect on compatibility under the combined effect of the two, highlighting the superimposed gain when the two factors work together.
[0036] The second step is to determine the second synergistic term based on the ratio relationship between element content data in different dimensions.
[0037] More specifically, the second synergistic term is determined based on the ratio between the element content data of each two different dimensions and the ratio between the element content data of each dimension and the sum of the element content data of the two dimensions.
[0038] As a concrete example, the second collaborative term can be represented as follows: , , .
[0039] The second synergistic term reflects the balance effect between different types of elements. It focuses on the relative proportions within the element system, rather than the absolute content, and is used to detect whether the proportions between elements are reasonable and whether there is a balance relationship where the proportion of one type of element is too high, inhibiting the effect of another element.
[0040] The third step is to determine the third synergistic term based on the superposition relationship between element content data in different dimensions.
[0041] Specifically, the sum of element content data from each pair of different dimensions is used as the third co-term.
[0042] As a concrete example, the third collaborative term can be represented as follows: , , .
[0043] The third synergistic term reflects the superimposed complementary effect of multiple elements. It is used to characterize the functional complementarity of different elements in improving compatibility, and reflects the comprehensive effect formed by the superposition of the individual effects when multiple elements coexist.
[0044] Finally, the synergistic feature vector of the resin-cured sample includes elemental content data and impact toughness data in all dimensions, as well as the first synergistic term, the second synergistic term, and the third synergistic term.
[0045] It should be understood that the elemental content data and impact toughness data of all dimensions are used as basic features, and the first, second and third synergistic terms are used as synergistic features. They are then sequentially spliced together in a certain order to form the synergistic feature vector of the resin-cured sample, which is a fusion of multi-dimensional features.
[0046] Step S300: Based on the collaborative feature vector and label data of the resin-cured sample, the pass probability of the resin-cured sample is obtained using a data model, and the detection result of the resin-cured sample is determined.
[0047] The main purpose of this step is to train and infer the machine learning model based on the collaborative feature vector and ground truth label data, output the probability of liquid oxygen compatibility of the resin-cured sample, and complete the engineering judgment based on the preset threshold, so as to provide a quantitative basis for the rapid screening of resin formulations.
[0048] Traditional liquid oxygen compatibility assessment relies solely on a binary result of pass or failure, failing to quantify risk levels and hindering high-throughput formulation screening. Therefore, this step employs a progressive logic of model training, probability output, and threshold comparison. First, the collaborative feature vectors and corresponding ground truth labels are divided into training and validation sets. Then, the training set is used to train and optimize the machine learning classification model, enabling it to output compatibility pass probabilities based on feature vectors. Next, the validation set is used to calibrate and validate the model's probability, ensuring the reliability of the output probabilities. Finally, the collaborative feature vectors of the sample to be evaluated are input into the trained model to obtain the corresponding pass probability. This probability is compared to a preset test threshold. If the pass probability is greater than or equal to the threshold, the sample is deemed to have acceptable liquid oxygen compatibility; if it is less than the threshold, the sample is deemed unacceptable or requires further testing.
[0049] The first step is to conduct a liquid oxygen shock compatibility test on the resin-cured sample. If the test is passed, the label data of the resin-cured sample is set to the first value; if the test fails, the label data of the resin-cured sample is set to the second value.
[0050] As a specific example, the first value and the second value are used to represent different states of the test result of passing or failing. In this embodiment, the first value can be set to 1 and the second value can be set to 0.
[0051] The liquid oxygen impact test can be conducted according to standard methods, preferably ASTM G86-17 or equivalent test methods. The label data of the resin-cured sample reflects the corresponding test results, and the label data of the resin-cured sample can be used as the true value label for model training and calibration.
[0052] The second step involves using a data model to obtain the pass probability of the resin-cured sample based on the collaborative feature vector and label data of the resin-cured sample.
[0053] As a concrete example, a fully connected neural network is used to obtain the pass probability of resin-cured samples by analyzing the collaborative feature vectors and label data of the resin-cured samples. In the fully connected neural network, the acquired collaborative feature vectors and label data of the resin-cured samples are used as the dataset. The network is trained using a stochastic gradient descent algorithm based on the cross-entropy loss function until convergence, yielding the output pass probability.
[0054] It should be noted that, in order to avoid the impact of different dimensions of data on the feature analysis process, the values of each dimension of the collaborative feature vector X of each sample data can be normalized before inputting into the neural network model. For example, the minimax normalization method can be used, and the implementer can choose according to the specific implementation scenario.
[0055] More specifically, the collaborative feature vector X (already normalized) of N groups of resin-cured samples is selected as the model input, and the corresponding liquid oxygen impact test ground truth label Y (0 or 1) is used as the target output. The dataset is randomly divided into a training set (e.g., 70%), a test set (e.g., 15%), and a validation set (e.g., 15%). The model is trained using a stochastic gradient descent algorithm based on the cross-entropy loss function until convergence. After training, the network weight parameters are fixed. The feature vector of the resin sample to be tested is input into the network, and the output layer directly outputs the value, which is the pass probability of the sample.
[0056] Furthermore, the test results of the resin-cured sample are judged as follows: if the pass probability of the resin-cured sample is greater than or equal to the preset test threshold, the resin-cured sample passes the test; if the pass probability of the resin-cured sample is less than the preset test threshold, the resin-cured sample fails the test, indicating that the test has failed and needs to be retested.
[0057] In this embodiment, because liquid oxygen compatibility involves system safety, it is crucial to avoid false negatives (i.e., materials that are actually incompatible but are incorrectly classified as passing by the model). Therefore, a high threshold strategy is adopted. Specifically, in this embodiment, the test threshold ranges from 0.8 to 0.95. This means that the material is only considered "passed" when the lower confidence bound of the model's predicted "pass probability" exceeds 80%. This implies that even if the model predicts a 79% probability of passing, the system will still classify it as requiring retesting / failure, thereby forcing physical verification to ensure safety.
[0058] In summary, the embodiments of this invention fully consider the close relationship between the resin's reaction sensitivity under liquid oxygen impact conditions and local hot spots, crack propagation, and energy dissipation. Various elemental systems influence compatibility in terms of inerting, isolating char / ceramization residues, increasing bond energy and chemical inertness, promoting thermal diffusion, and reducing hot spot temperature rise. By introducing synergistic features (element content and toughness coupling terms, ratio terms), differences caused by "same element content but different toughness" or "same toughness but different element ratios" can be captured, thereby reducing misjudgments. Through a confidence lower bound determination mechanism, the probability of missing high-risk samples in engineering assessments can be significantly reduced. Model screening can greatly reduce the number of candidate formulations requiring liquid oxygen hazard testing, thereby shortening the development cycle and reducing testing risks.
[0059] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.
Claims
1. A method for predicting the compatibility of resin with liquid oxygen based on the synergy of elemental content and toughness, characterized in that, The method includes: Obtain elemental content data and impact toughness data of the resin-cured sample in multiple dimensions; Based on the multi-dimensional feature interaction relationship between element content data and impact toughness data in each dimension, a collaborative feature vector of resin-cured samples is constructed; based on the test results of resin-cured samples, label data of resin-cured samples is obtained. Based on the collaborative feature vector and label data of resin-cured samples, the pass probability of resin-cured samples is obtained using a data model, and the detection results of resin-cured samples are judged. Specifically, the construction of a collaborative feature vector for the resin-cured sample based on the multi-dimensional feature interaction relationship between elemental content data and impact toughness data in each dimension includes: Based on the coupling relationship between the element content data and impact toughness data of each dimension, the first synergistic term of the element content data of each dimension is determined; Based on the ratio relationship between element content data from different dimensions, the second synergistic term is determined; The third synergistic term is determined based on the superposition relationship between element content data from different dimensions; The co-functional feature vector of the resin-cured sample includes elemental content data and impact toughness data in all dimensions, a first co-functional term, a second co-functional term, and a third co-functional term.
2. The method for predicting resin-liquid oxygen compatibility based on the synergy of elemental content and toughness according to claim 1, characterized in that, The coupling relationship between elemental content data and impact toughness data in each dimension is used to determine the first synergistic term of elemental content data in each dimension, specifically including: The product of the element content data and the impact toughness data for each dimension is taken as the first co-term of the element content data for each dimension.
3. The method for predicting resin-liquid oxygen compatibility based on the synergy of elemental content and toughness according to claim 1, characterized in that, The determination of the second synergistic term based on the ratio relationship between element content data of different dimensions specifically includes: The second synergistic term is determined based on the ratio between the element content data of each two different dimensions and the ratio between the element content data of each dimension and the sum of the element content data of the two dimensions.
4. The method for predicting resin-liquid oxygen compatibility based on the synergy of elemental content and toughness according to claim 1, characterized in that, The third synergistic term is determined based on the superposition relationship between element content data from different dimensions, specifically including: The sum of the element content data between any two different dimensions is used as the third co-term.
5. The method for predicting resin-liquid oxygen compatibility based on the synergy of elemental content and toughness according to claim 1, characterized in that, The label data for the resin-cured samples, obtained from the test results, specifically includes: A liquid oxygen shock compatibility test is performed on the resin-cured sample. If the test passes, the label data of the resin-cured sample is set to the first value; if the test fails, the label data of the resin-cured sample is set to the second value.
6. The method for predicting resin-liquid oxygen compatibility based on the synergy of elemental content and toughness according to claim 1, characterized in that, The determination of the test results for the resin-cured sample specifically includes: If the pass probability of the resin-cured sample is greater than or equal to the preset test threshold, the resin-cured sample passes the test. If the pass probability of the resin-cured sample is less than the preset test threshold, the resin-cured sample fails the test.
7. The method for predicting resin-liquid oxygen compatibility based on the synergy of elemental content and toughness according to claim 1, characterized in that, The element content data includes: The content of phosphorus, nitrogen, silicon, fluorine, aluminum, and sulfur.