Composite material structure damage probability evaluation method and device and electronic equipment

Abstract

The invention relates to a composite material structure damage probability assessment method and device and electronic equipment, and relates to the technical field of aircraft maintenance, and the method comprises the steps: obtaining the structure influence factor data of a to-be-assessed composite material structure; inputting the structure influence factor data into the trained random forest model, and performing damage probability evaluation processing on the to-be-evaluated composite material structure through all decision trees in the random forest model to generate an evaluation result; the evaluation results are subjected to voting processing, the evaluation result with the maximum voting number serves as a predicted structure damage evaluation result of the composite material structure to be evaluated, and the predicted structure damage evaluation result at least comprises slight damage, moderate damage and serious damage. According to the scheme, the limitation of a traditional method in the absence of sensor data is effectively overcome, and the damage prediction requirement under the condition of no sensor data or incomplete data is met.

Description

Technical Field

[0001] This disclosure relates to the field of aircraft maintenance technology, specifically to a method, apparatus, and electronic equipment for assessing the probability of damage to composite material structures. Background Technology

[0002] As a critical component of an aircraft, the nacelle operates in a complex and ever-changing environment, and its structural condition directly affects flight safety and performance. Therefore, accurately predicting the condition of the nacelle structure, especially assessing the probability of damage occurrence and development, is crucial for ensuring the continued airworthiness of the aircraft and for developing scientifically sound maintenance strategies.

[0003] Traditional methods for predicting damage to aircraft structures mainly include rule-based methods, logistic regression, and support vector machines (SVM), among other data analysis techniques. Taking rule-based methods as an example, they typically rely on the experience and knowledge of domain experts and signal processing techniques to formulate diagnostic rules. While this approach can leverage rich engineering experience, the completeness and adaptability of the rules are often difficult to guarantee when dealing with complex objects like nacelle structures with diverse damage patterns. Especially in the absence of comprehensive sensor monitoring data, its predictive ability is significantly limited, making it difficult to meet the needs of damage prediction in situations with no sensor data or incomplete data. Summary of the Invention

[0004] In view of this, this disclosure provides a method, apparatus and electronic device for assessing the probability of damage to composite material structures. The main purpose is to solve the technical problem that the predictive ability of traditional aircraft structural damage prediction methods is significantly limited in the absence of comprehensive sensor monitoring data, making it difficult to meet the needs of damage prediction in the absence of sensor data or incomplete data.

[0005] According to a first aspect of this disclosure, a method for assessing the damage probability of a composite material structure is provided, the method comprising: Obtain data on the structural influencing factors of the composite material structure to be evaluated; The structural influencing factor data are input into the trained random forest model, and the damage probability assessment of the composite material structure to be evaluated is performed through all decision trees in the random forest model to generate the evaluation result. The evaluation results are voted on, and the evaluation result with the most votes is taken as the predicted structural damage evaluation result of the composite material structure to be evaluated. The predicted structural damage evaluation result includes at least minor damage, moderate damage, and severe damage.

[0006] According to a second aspect of this disclosure, a device for assessing the probability of damage to composite material structures is provided, the device comprising: The acquisition module is used to acquire data on structural influencing factors of the composite material structure to be evaluated. The evaluation module is used to input the structural influencing factor data into the trained random forest model, and to perform damage probability evaluation on the composite material structure to be evaluated through all decision trees in the random forest model to generate evaluation results. The voting module is used to process the evaluation results through voting, and the evaluation result with the largest number of votes is used as the predicted structural damage evaluation result of the composite material structure to be evaluated. The predicted structural damage evaluation result includes at least minor damage, moderate damage, and severe damage.

[0007] According to a third aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of the first aspect described above.

[0008] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause a computer to perform the method of the first aspect described above.

[0009] The composite material structure damage probability assessment method, apparatus, and electronic equipment disclosed herein, compared with existing technologies, acquire structural influencing factor data of the composite material structure to be assessed; input the structural influencing factor data into a trained random forest model; perform damage probability assessment on the composite material structure to be assessed through all decision trees in the random forest model to generate assessment results; perform voting on the assessment results, and take the assessment result with the highest number of votes as the predicted structural damage assessment result of the composite material structure to be assessed. The predicted structural damage assessment result includes at least minor damage, moderate damage, and severe damage. By applying the scheme of this disclosure, the acquired structural influencing factor data can be used to input the structural influencing factor data into a trained random forest model to leverage the robustness and ensemble advantages of the random forest model, and provide probabilistic and hierarchical predicted structural damage assessment results. This effectively overcomes the limitations of traditional methods in the absence of sensor data and meets the needs of damage prediction in the absence of sensor data or incomplete data. Attached Figure Description

[0010] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0011] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart illustrating a method for assessing the damage probability of a composite material structure provided in an embodiment of this disclosure. Figure 2 A component damage state assessment process prediction diagram based on the random forest algorithm provided in this embodiment of the disclosure; Figure 3 An obfuscation matrix diagram provided in an embodiment of this disclosure; Figure 4 An obfuscation matrix diagram provided in an embodiment of this disclosure; Figure 5 This is a schematic diagram of a composite material structure damage probability assessment device provided in an embodiment of the present disclosure. Detailed Implementation

[0013] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description. It should be noted that, unless otherwise specified, the embodiments of this disclosure and the features described therein can be combined with each other.

[0014] The following description, with reference to the accompanying drawings, outlines a method, apparatus, and electronic device for assessing the damage probability of composite material structures according to embodiments of this disclosure.

[0015] This disclosure provides a method, apparatus, and electronic device for assessing the probability of damage to composite material structures. The main purpose is to address the technical problem that the predictive ability of traditional aircraft structural damage prediction methods is significantly limited in the absence of comprehensive sensor monitoring data, making it difficult to meet the needs of damage prediction in the absence of sensor data or incomplete data.

[0016] like Figure 1 As shown, embodiments of this disclosure provide a method for assessing the damage probability of composite material structures, including: Step 101: Obtain data on structural influencing factors of the composite material structure to be evaluated.

[0017] For the embodiments of this disclosure, various structural influencing factor data related to the composite material structure to be evaluated can be collected from multiple sources such as design documents, manufacturing records, maintenance logs, and environmental monitoring records. The structural influencing factor data may include working parameter data, service environment data, and other structural influencing factors.

[0018] Among them, the working parameter data may include the age of the component and the number of usage cycles of the component. The age of the component can be the service life of the composite material structure, such as the service time of the aircraft wing or fuselage structure; the number of usage cycles of the component can be the number of times the structural component has been used, such as the number of aircraft flight cycles or the number of engine start-stop cycles.

[0019] Service environment data may include average annual precipitation, average annual temperature, number of months with average monthly temperature greater than 20 degrees Celsius, average annual humidity, average number of frost days per year, and whether it is a coastal area.

[0020] Other structural influencing factors may include load history, manufacturing process parameters, material properties, maintenance records, etc.

[0021] Data cleaning can be performed on the acquired structural influencing factors to remove outliers, and numerical features can be mapped to the [0,1] interval through normalization to eliminate the impact of dimensional differences on model training.

[0022] Step 102: Input the structural influencing factor data into the trained random forest model, and use all decision trees in the random forest model to perform damage probability assessment on the composite material structure to be evaluated, and generate the assessment results.

[0023] In the embodiments of this disclosure, before inputting the structural influencing factor data into the trained random forest model, it is first necessary to construct and train the random forest model. Specifically: Step 201: Based on the preset composite material structure damage state evaluation index system, the key influencing factors of composite material structure damage can be determined. In this embodiment, factors significantly related to structural damage of typical engine components (such as nacelle sliding doors) can be screened according to the preset composite material structure damage status evaluation index system shown in Table 1, including operating parameters and service environment parameters.

[0024]

[0025] Table 1 Pre-set evaluation index system for damage state of composite material structures By using engineering experience analysis methods and combining them with the service scenarios of engine components, factors that significantly affect thermal aging (X4, X5), corrosion (X3, X6, X7, X8), and fatigue (X1) can be identified.

[0026] Analysis of variance was performed on historical damage data (e.g., 240 samples within 10 years) to screen variables that were significantly related to the degree of damage (Y1). Finally, variables X1, X3 to X8 in Table 1 were retained as inputs for the random forest model.

[0027] For example, an analysis of 240 samples of a certain type of engine nacelle sliding door showed that for every year the age of the component (X1) increases, the probability of serious damage increases by 12%; the risk of moderate damage in a coastal environment (X8=1) is 2.3 times that in an inland environment (X8=2).

[0028] Step 202: Collect sample data containing key influencing factors and corresponding structural damage assessment results, and divide the sample data into training sample set and test sample set according to a preset ratio using stratified sampling. The structural damage assessment results are used to characterize different damage levels, including at least minor damage, moderate damage and severe damage. In this embodiment, firstly, key influencing factors of composite material structures (such as a sliding door of an engine nacelle) can be collected, namely variables X1, X3 to X8 in step 201, as well as sample data of the corresponding structural damage assessment results.

[0029] The structural damage assessment results can be obtained by periodically inspecting, non-destructive testing or other evaluation methods on the composite material structure to quantify the current damage state of the structure. In this embodiment, the structural damage assessment results can be divided into three levels. Different levels can reflect different degrees of damage. For example, the first level (1) represents minor damage (such as surface scratch depth < 0.2 mm), the second level (2) represents moderate damage (such as interlaminar debonding area < 5%), and the third level (3) represents severe damage (such as structural penetration or debonding area ≥ 5%), which can be used as the prediction target of the model (i.e., output variable Y1).

[0030] To build and evaluate the prediction model, the collected sample data can be divided into a training sample set and a test sample set. In this embodiment, a preset division ratio can be set, such as 7:3, that is, 70% of the total sample data can be allocated to the training sample set for training the random forest model; the remaining 30% of the total sample data can be allocated to the test sample set for evaluating its prediction performance and generalization ability after the model training is completed.

[0031] In particular, stratified sampling can be used when dividing the sample set. Stratified sampling involves grouping the population (i.e., all samples) according to one or more important categorical features, called "strata". Then, random sampling is performed independently within each "stratum", and samples are drawn according to a preset ratio (7:3) and placed into the training sample set and the test sample set respectively.

[0032] In this embodiment, considering that the service environment category is a key factor affecting damage, and that the number of samples under different environment categories may vary, stratified sampling is used to ensure that the proportion of samples from each service environment category in the training and test sets is roughly consistent with the proportion in the original dataset. This avoids the model failing to learn the damage patterns of a particular environment category accurately or inaccurately due to a small proportion of such samples in the training or test sets. In other words, it ensures that the proportion of each environment type in the training or test sets is roughly the same, thereby enhancing the robustness and accuracy of the model's damage prediction under different service environments. Through the above steps, a training sample set containing 168 samples and a test sample set containing 72 samples are finally obtained.

[0033] Step 203: Randomly select training samples with replacement from the training sample set using the bootstrap sampling method to generate multiple training subsets; train the corresponding decision tree using each training subset, and randomly select some features for optimal splitting at each node of the decision tree to generate multiple decision trees; construct a random forest model using multiple decision trees.

[0034] In this embodiment, as Figure 2 As shown, bootstrap sampling can be used to randomly select samples with replacement from the training sample set and repeat this process, for example, selecting a sample size that is the same as or slightly smaller than the original training set, thereby generating multiple independent training subsets. Because sampling is done with replacement, some samples may appear multiple times in the same subset, while other samples may not be selected, thus increasing the randomness of the training process.

[0035] For example, subset 1 may contain repeated samples (such as sample A appearing 3 times), while subset 2 may not contain sample B at all, ensuring the difference between subsets. By introducing sample randomness through sampling with replacement, the difference rate of training data for a single decision tree reaches 63.2% (theoretical value), avoiding model overfitting.

[0036] Next, a decision tree can be independently trained using each generated training subset. During the training of each decision tree, a random attribute selection mechanism is introduced. That is, when a decision tree grows to a node requiring splitting, the algorithm does not consider all available features (i.e., X1 to X8 in Table 1), but randomly selects a subset from these features (e.g., by default, the square root of the total number of features is selected). Then, based solely on the randomly selected feature subset, the optimal split point that best distinguishes the damage state of the node samples is found. This practice of randomly selecting some features when splitting nodes further increases the diversity between each decision tree, effectively reducing the risk of model overfitting and improving the model's generalization ability.

[0037] Through the above process, multiple decision trees with different structures and predictive abilities are ultimately generated. These decision trees together constitute a random forest model. The number of decision trees (Ntree) in the random forest can be set to 500, and the number of features Mtry considered in node splitting can be 3, resulting in a confusion matrix, as shown below. Figure 3 and Figure 4 As shown in Table 2, the category prediction results are obtained based on the confusion matrix.

[0038]

[0039] Table 2 Category Prediction Results After the random forest model is trained, the structural influencing factor data of the composite material structure to be evaluated can be input into the trained random forest model. Inside the random forest model, the structural influencing factor data can be simultaneously passed to each decision tree in the forest. Each decision tree can independently judge the input data according to its own rules and output a preliminary assessment result on the damage state of the structure (e.g., minor damage, moderate damage, or severe damage). The output results of all decision trees are then aggregated to form a preliminary assessment result set.

[0040] In the embodiments of this disclosure, a trained random forest model can be used to perform damage probability assessment and prediction processing on the training sample set to obtain the predicted structural damage assessment results. Construct a confusion matrix, which is used to statistically predict the correspondence between the structural damage assessment results and the actual damage state of the training sample set; Based on the confusion matrix, the precision and recall of the random forest model on the training dataset are calculated to predict the structural damage assessment results. The performance of the random forest model is evaluated based on precision and recall, and the model parameters are dynamically adjusted to optimize the model performance.

[0041] In this embodiment, after the initial construction and training of the random forest model are completed, the model is validated and the model parameters are adjusted using the training sample set in order to evaluate its performance and perform necessary optimizations.

[0042] First, the trained random forest model can be used to perform damage probability assessment and prediction on the training sample set itself. The structural influencing factor data of the training sample set can be input into the constructed random forest model. Each decision tree in the random forest model can classify the samples according to the rules it has learned, and finally generate the predicted structural damage assessment result (i.e., minor damage, moderate damage, or severe damage) for each sample through a voting mechanism.

[0043] Next, a confusion matrix can be constructed. The confusion matrix can be used to evaluate the performance of the classification model. Table 3 shows the interpretation of the confusion matrix results for each category. In classification prediction problems, it is desirable to classify as many samples correctly as possible, that is, to maximize the TP and TN values.

[0044]

[0045] Table 3. Explanation of the Category Confusion Matrix Results for Classification Problems A confusion matrix is ​​a tabular tool used to visually represent the correspondence between model predictions and actual labels, such as... Figure 3 and Figure 4 As shown, the precision and recall of the random forest model on the training dataset can be calculated based on the constructed confusion matrix.

[0046] Precision rate measures the proportion of samples that are actually classified as positive (e.g., predicted as "minor injury") by the model, reflecting the accuracy of the model's predictions. The calculation formula is shown below:

[0047] In the formula, It can represent the accuracy rate. It can represent a real class. It can represent a pseudo-positive class.

[0048] Recall measures the proportion of samples that are actually positive that are correctly predicted as positive by the model. It reflects the model's ability to identify that category. The calculation formula is shown below:

[0049] In the formula, It can represent recall rate. It can represent a real class. It can represent a pseudo-negative class.

[0050] Finally, the performance of the random forest model can be comprehensively evaluated based on the calculated precision and recall metrics. If the evaluation results show that the model performance does not meet expectations, or if overfitting / underfitting is observed, the model parameters can be dynamically adjusted to optimize its performance. For example, the number of decision trees (Ntree) can be adjusted, such as increasing or decreasing it from 500; the number of features considered when splitting nodes (Mtry) can be adjusted, such as changing it from 3 to other values; and other parameters such as the maximum depth of the decision trees and the minimum number of split samples can also be adjusted. After adjustments, retraining and evaluation are necessary until the model reaches the target performance level on both the training and independent test sets, ensuring that the final model has better generalization ability and prediction accuracy.

[0051] Based on the precision and recall calculation formulas, the prediction accuracy of the training and test sets based on the random forest algorithm for damage prediction can be calculated, as shown in Table 4 below.

[0052]

[0053] Table 4 Prediction Accuracy Table In this embodiment of the disclosure, a trained random forest model can be used to perform damage probability assessment and prediction processing on the test dataset to obtain the predicted structural damage assessment results. Construct a confusion matrix, which is used to statistically predict the correspondence between the structural damage assessment results and the actual damage state of the test dataset; Based on the confusion matrix, calculate at least one classification evaluation index result of the random forest model on the test dataset under the prediction of structural damage assessment results. The classification evaluation index result includes at least precision, recall, F1 score, and AUC value. Analyze the classification evaluation index results to evaluate the classification performance of the random forest model.

[0054] In this embodiment, the test dataset can be input into the already trained random forest model. The random forest model can use all its internal decision trees to evaluate the damage probability of each sample in the test dataset based on the previously learned patterns. Each decision tree can independently give a prediction result, and then the final predicted structural damage assessment result (e.g., minor damage, moderate damage, or severe damage) of the sample can be determined through a voting mechanism.

[0055] Next, a confusion matrix can be constructed, which can be used to systematically statistically analyze the correspondence between the predicted structural damage assessment results and the actual damage state of the samples on the test dataset.

[0056] Based on the values ​​of TP, FP, FN, TN, etc., obtained from the confusion matrix, at least one key classification evaluation metric of the random forest model on the test dataset can be calculated. The classification evaluation metric may include: precision, recall, F1 score, and AUC (Area Under Curve).

[0057] Among them, the F1 score is the harmonic mean of precision and recall, which can be used to provide a single indicator to comprehensively evaluate the performance of the model in class imbalance problems; the AUC value can be used to reflect the area under the ROC curve of the model's ability to distinguish positive and negative samples, reflecting the overall ability of the model to distinguish different damage categories (or positive and negative samples). The closer the value is to 1, the stronger the model's ability to distinguish.

[0058] By comprehensively analyzing these metrics, the classification performance of the random forest model on the test dataset can be fully evaluated, and it can be determined whether it meets the requirements of practical applications. For example, the differences in metrics between the training and test sets can be compared to determine whether the model is overfitting; the performance of metrics on different damage categories can be observed to determine the model's ability to handle class imbalance problems.

[0059] In this embodiment of the disclosure, by comparing the classification accuracy of the model prediction results under different service environment types, the adaptability and generalization performance of the model under multiple environmental conditions can be verified, ensuring that it has good stability in practical applications.

[0060] Specifically, the number of first samples that were correctly predicted and the number of second samples that were incorrectly predicted can be counted based on the predicted structural damage assessment results; The accuracy of the training sample set and the test sample set is calculated based on the first sample size and the second sample size, respectively. The overall accuracy of the random forest model is obtained based on the accuracy, the predicted structural damage assessment results of the combined training and test sample sets, and the overall accuracy is used to comprehensively evaluate the prediction performance of the random forest model.

[0061] In this embodiment of the disclosure, after using the trained random forest model to perform damage probability assessment and prediction processing on the training sample set and the test sample set respectively, and obtaining the respective predicted structural damage assessment results, the number of first samples correctly predicted by the model and the number of second samples incorrectly predicted by the model under the predicted structural damage assessment results can be counted.

[0062] The first sample size can be used to characterize the total number of samples whose predicted results are consistent with the actual results, including all correctly classified samples (i.e., the sum of true negative and true negative samples); the second sample size can be used to characterize the total number of samples whose predicted results are inconsistent with the actual results, including all misclassified samples (i.e., the sum of false negative and false positive samples).

[0063] Based on the statistically obtained first and second sample numbers, the accuracy of the model on the training and test sample sets can be calculated respectively. The accuracy calculation formula is shown below:

[0064] In the formula, It can represent accuracy. It can represent a real class. It can represent a pseudo-negative class. It can represent a false positive class. It can represent a true negative class.

[0065] That is, the accuracy rate is determined by the ratio of the first sample size to the sum of the first and second sample sizes.

[0066] Finally, based on the accuracy of the training and test sets (e.g., the accuracy of the training set is 92.9% and the accuracy of the test set is 81.9%), the overall accuracy of the random forest model can be calculated as 89.6% by combining the prediction results of the training and test sets. The overall accuracy can be the proportion of all correctly predicted samples to the total number of samples, which is a key indicator for evaluating the prediction performance of the random forest model.

[0067] Step 103: The evaluation results are voted on, and the evaluation result with the largest number of votes is taken as the predicted structural damage evaluation result of the composite material structure to be evaluated. The predicted structural damage evaluation result includes at least minor damage, moderate damage and severe damage.

[0068] Following step 203 of the embodiment, as follows Figure 2 As shown, after all decision trees output the preliminary evaluation result set, the final decision fusion can be performed. Specifically, a voting mechanism can be used to count the number of votes received by each evaluation result (i.e., damage state category is minor damage, moderate damage, and severe damage) in the evaluation result set. The number of votes can be the number of times each decision tree gives that evaluation result category.

[0069] For example, if a random forest consists of 100 decision trees, and the evaluation results show that 45 decision trees predict "minor damage", 35 decision trees predict "moderate damage", and 20 decision trees predict "severe damage", then "minor damage" receives 45 votes, "moderate damage" receives 35 votes, and "severe damage" receives 20 votes.

[0070] Finally, the damage state category that receives the most votes can be used as the final predicted structural damage assessment result for the composite material structure to be evaluated. For example, since "minor damage" received the most votes (45), the final prediction result is "minor damage". Through this majority voting method, random forests can integrate the opinions of all decision trees to make more robust and accurate judgments, effectively reducing the bias or error that may exist in a single decision tree, thereby providing reliable information on structural health status.

[0071] In the embodiments of this disclosure, the predicted structural damage assessment results under different component age conditions can be obtained in each service environment category; Based on the predicted structural damage assessment results, an age-damage probability curve is plotted to show the evolution trend of structural damage risk under different service environments.

[0072] Specifically, based on the trained random forest model, the damage probability of samples under different service environment types can be further predicted. Random forests can provide a probability estimate of each sample belonging to a damage category by integrating the outputs of multiple decision trees, i.e.:

[0073] By iterating through samples under different component ages, the predicted damage probability is calculated for each environmental category and plotted as a component age-damage probability curve, thus visually demonstrating the evolution trend of structural damage risk under different service environments. This prediction result not only reveals the coupling relationship between service environment and component age in damage evolution but also provides important data support for operating units to formulate environmentally adaptable maintenance strategies.

[0074] This disclosure establishes a damage evaluation index system for composite material structures based on maintenance data of composite material structures from over a hundred in-service civil aircraft of a certain airline (covering a wide geographical area and long service life), through a combination of cluster analysis and analysis of variance.

[0075] Based on the damage assessment index system, this paper considers nacelle structure condition prediction technology under complex service environments. It integrates damage mechanism and data modeling as dual drivers to develop a structural damage probability prediction algorithm with time-varying modeling capabilities, thereby improving the accuracy and stability of prediction results in complex service environments. Simultaneously, based on damage level data from actual aircraft structural maintenance, a classification algorithm is established using machine learning and trained multiple times to achieve structural damage classification and assessment, providing accurate damage grading judgment criteria for determining safe maintenance tolerance.

[0076] In summary, the composite material structure damage probability assessment method provided in this disclosure, compared with existing technologies, can obtain structural influencing factor data of the composite material structure to be assessed; input the structural influencing factor data into a trained random forest model; perform damage probability assessment on the composite material structure to be assessed through all decision trees in the random forest model to generate assessment results; perform voting on the assessment results, and take the assessment result with the highest number of votes as the predicted structural damage assessment result of the composite material structure to be assessed. The predicted structural damage assessment result includes at least minor damage, moderate damage, and severe damage. By applying the scheme of this disclosure, the obtained structural influencing factor data can be used to input the structural influencing factor data into a trained random forest model to leverage the robustness and ensemble advantages of the random forest model, and provide probabilistic and hierarchical predicted structural damage assessment results. This effectively overcomes the limitations of traditional methods in the absence of sensor data and meets the needs of damage prediction in the absence of sensor data or incomplete data.

[0077] Based on the above Figure 1 The specific implementation of the method shown in this embodiment provides a composite material structure damage probability assessment device, such as... Figure 5 As shown, the device includes: an acquisition module 31, an evaluation module 32, and a voting module 33; Module 31 is used to acquire data on structural influencing factors of the composite material structure to be evaluated. Evaluation module 32 is used to input the structural influencing factor data into the trained random forest model, and to perform damage probability evaluation on the composite material structure to be evaluated through all decision trees in the random forest model to generate evaluation results. The voting module 33 is used to process the evaluation results by voting, and the evaluation result with the largest number of votes is used as the predicted structural damage evaluation result of the composite material structure to be evaluated. The predicted structural damage evaluation result includes at least minor damage, moderate damage and severe damage.

[0078] In specific application scenarios, such as Figure 5 As shown, the device also includes: a construction module 34; The construction module 34 is used to randomly sample training samples with replacement from the training sample set using a bootstrap sampling method to generate multiple training subsets; to train a corresponding decision tree using each of the training subsets; to randomly select some features for optimal splitting at each node of the decision tree to generate multiple decision trees; and to construct the random forest model using the multiple decision trees.

[0079] In specific application scenarios, such as Figure 5 As shown, the device also includes: a division module 35; The partitioning module 35 is used to determine the key influencing factors of composite material structure damage based on the preset composite material structure damage state evaluation index system; collect sample data containing the key influencing factors and the corresponding structural damage assessment results; and divide the sample data into a training sample set and a test sample set according to a preset ratio using stratified sampling. The structural damage assessment results are used to characterize different damage levels, including at least minor damage, moderate damage and severe damage.

[0080] In specific application scenarios, such as Figure 5 As shown, the device also includes: an optimization module 36; The optimization module 36 is used to perform damage probability assessment and prediction processing on the training sample set using the trained random forest model to obtain the predicted structural damage assessment result; construct a confusion matrix, which is used to statistically determine the correspondence between the predicted structural damage assessment result and the actual damage state of the training sample set; calculate the precision and recall of the random forest model on the training dataset under the predicted structural damage assessment result based on the confusion matrix; evaluate the model performance of the random forest model based on the precision and recall, and dynamically adjust the model parameters of the random forest model to optimize the model performance of the random forest model.

[0081] In specific application scenarios, such as Figure 5 As shown, the device also includes: an evaluation module 37; Evaluation module 37 is used to perform damage probability assessment and prediction processing on the test dataset using the trained random forest model to obtain predicted structural damage assessment results; construct a confusion matrix, which is used to statistically analyze the correspondence between the predicted structural damage assessment results and the actual damage state of the test dataset; based on the confusion matrix, calculate at least one classification evaluation index result of the random forest model for the test dataset under the predicted structural damage assessment results, wherein the classification evaluation index result includes at least precision, recall, F1 score, and AUC value; and analyze the classification evaluation index result to evaluate the classification performance of the random forest model.

[0082] In specific application scenarios, such as Figure 5 As shown, the device also includes: a computing module 38; The calculation module 38 is used to count the number of first samples correctly predicted and the number of second samples incorrectly predicted under the predicted structural damage assessment results; calculate the accuracy of the training sample set and the test sample set based on the first sample number and the second sample number respectively; calculate the overall accuracy of the random forest model based on the accuracy and the combined predicted structural damage assessment results of the training sample set and the test sample set, and the overall accuracy is used to comprehensively evaluate the prediction effect of the random forest model.

[0083] In specific application scenarios, such as Figure 5 As shown, the device also includes: a drawing module 39; The plotting module 39 is used to obtain the predicted structural damage assessment results under different component age conditions in each service environment category; and plot the component age-damage probability curve based on the predicted structural damage assessment results to show the evolution trend of structural damage risk under different service environments.

[0084] Based on the above, Figure 1 Accordingly, this embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. Figure 1 The method shown.

[0085] Based on this understanding, the technical solution disclosed herein can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, or portable hard drive) and includes several instructions to cause a computer device (such as a personal computer, server, or network device) to execute the methods of various implementation scenarios of this disclosure.

[0086] Based on the above, Figure 1 The method shown, and Figure 5 To achieve the above objectives, this disclosure also provides an electronic device, configurable on a vehicle (e.g., an electric vehicle), in accordance with the illustrated virtual device embodiment. This device includes a storage medium and a processor; the storage medium stores a computer program; the processor executes the computer program to implement the above-described virtual device. Figure 1 The method shown.

[0087] Optionally, the aforementioned physical devices may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB interfaces, card reader interfaces, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Wi-Fi interfaces), etc.

[0088] Those skilled in the art will understand that the physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or combine certain components, or have different component arrangements.

[0089] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the aforementioned physical device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software in the information processing physical device.

[0090] Through the above description of the embodiments, those skilled in the art can clearly understand that this disclosure can be implemented using software plus necessary general-purpose hardware platforms, or it can be implemented through hardware. Compared with the prior art, the technical solution of this disclosure obtains structural influencing factor data of the composite material structure to be evaluated; inputs the structural influencing factor data into a trained random forest model; performs damage probability assessment on the composite material structure to be evaluated through all decision trees in the random forest model, generating an assessment result; performs voting on the assessment results, and uses the assessment result with the highest number of votes as the predicted structural damage assessment result of the composite material structure to be evaluated. The predicted structural damage assessment result includes at least minor damage, moderate damage, and severe damage. By applying the solution of this disclosure, the obtained structural influencing factor data can be used to input the structural influencing factor data into a trained random forest model to leverage the robustness and ensemble advantages of the random forest model, and provide probabilistic and hierarchical predicted structural damage assessment results. This effectively overcomes the limitations of traditional methods in the absence of sensor data and meets the needs for damage prediction in the absence of sensor data or incomplete data.

[0091] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprising" or any other variations thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0092] The above are merely specific embodiments of this disclosure, enabling those skilled in the art to understand or implement this disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to these embodiments, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for assessing the probability of damage to composite material structures, characterized in that, The method includes: Obtain data on the structural influencing factors of the composite material structure to be evaluated; The structural influencing factor data are input into the trained random forest model, and the damage probability assessment of the composite material structure to be evaluated is performed through all decision trees in the random forest model to generate the evaluation result. The evaluation results are voted on, and the evaluation result with the most votes is taken as the predicted structural damage evaluation result of the composite material structure to be evaluated. The predicted structural damage evaluation result includes at least minor damage, moderate damage, and severe damage.

2. The method according to claim 1, characterized in that, The random forest model construction process includes: Multiple training subsets are generated by randomly selecting training samples with replacement from the training sample set using a bootstrap sampling method. A corresponding decision tree is trained using each of the training subsets. When splitting at each node of the decision tree, a subset of features is randomly selected for optimal splitting to generate multiple decision trees. The random forest model is constructed using the multiple decision trees.

3. The method according to claim 2, characterized in that, Before generating multiple training subsets by randomly sampling training samples with replacement from the training sample set using a bootstrap sampling method, the method further includes: Based on a pre-defined index system for evaluating the damage state of composite material structures, the key influencing factors of composite material structure damage are determined. Collect sample data containing the key influencing factors and corresponding structural damage assessment results, and divide the sample data into a training sample set and a test sample set according to a preset ratio using stratified sampling. The structural damage assessment results are used to characterize different damage levels, including at least minor damage, moderate damage, and severe damage.

4. The method according to claim 3, characterized in that, The optimization process of the random forest model includes: The trained random forest model is used to perform damage probability assessment and prediction processing on the training sample set to obtain the predicted structural damage assessment results. Construct a confusion matrix, which is used to statistically determine the correspondence between the predicted structural damage assessment results and the actual damage state of the training sample set; Based on the confusion matrix, the precision and recall of the random forest model on the training dataset under the predicted structural damage assessment results are calculated. Based on the precision and recall, the model performance of the random forest model is evaluated, and the model parameters of the random forest model are dynamically adjusted to optimize the model performance.

5. The method according to claim 3, characterized in that, The evaluation process of the random forest model includes: The trained random forest model is used to perform damage probability assessment and prediction processing on the test dataset to obtain the predicted structural damage assessment results. Construct a confusion matrix, which is used to statistically correlate the predicted structural damage assessment results with the actual damage state of the test dataset; Based on the confusion matrix, the random forest model calculates at least one classification evaluation index result for the test dataset under the predicted structural damage assessment result, wherein the classification evaluation index result includes at least precision, recall, F1 score, and AUC value; The classification evaluation index results are analyzed to assess the classification performance of the random forest model.

6. The method according to claim 4 or 5, characterized in that, The method further includes: The number of first samples that were correctly predicted and the number of second samples that were incorrectly predicted were counted based on the predicted structural damage assessment results. The accuracy of the training sample set and the test sample set are calculated based on the first number of samples and the second number of samples, respectively. The overall accuracy of the random forest model is calculated based on the accuracy and the predicted structural damage assessment results of the training sample set and the test sample set. The overall accuracy is used to comprehensively evaluate the prediction performance of the random forest model.

7. The method according to claim 1, characterized in that, The method further includes: In each service environment category, the predicted structural damage assessment results under different component age conditions are obtained; Based on the predicted structural damage assessment results, an age-damage probability curve is plotted to demonstrate the evolution trend of structural damage risk under different service environments.

8. A device for assessing the probability of damage to composite material structures, characterized in that, include: The acquisition module is used to acquire data on structural influencing factors of the composite material structure to be evaluated. The evaluation module is used to input the structural influencing factor data into the trained random forest model, and to perform damage probability evaluation on the composite material structure to be evaluated through all decision trees in the random forest model to generate evaluation results. The voting module is used to process the evaluation results through voting, and the evaluation result with the largest number of votes is used as the predicted structural damage evaluation result of the composite material structure to be evaluated. The predicted structural damage evaluation result includes at least minor damage, moderate damage, and severe damage.

9. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-7.

Images