Aircraft brake system evaluation and prediction method based on fault simulation and geometric features

Abstract

The invention discloses an aircraft braking system evaluation and prediction method based on fault simulation and geometric features, and belongs to the technical field of electric data processing. The method comprises the following steps: constructing a simulation model of an aircraft brake system, and injecting faults into the simulation model according to different degrees of fault injection control parameters to generate a degradation monitoring data set of the whole life cycle of the system; dividing the degradation monitoring data set into a training data set and a test data set; training the GoogLeNet model by using the training data set to obtain a health assessment model; evaluating the health degree of the aircraft brake system by using the test data set; constructing a topological graph according to the multi-rack operation data and the health degree of the aircraft braking system; nodes in the topological graph are clustered and divided into K clusters; establishing a residual life prediction model for each cluster; and performing similarity matching on the sample and the K clusters, and selecting the best matched residual life prediction model to perform residual life prediction on the system. The method can accurately predict the remaining life of the aircraft brake system.

Description

Technical Field

[0001] The present invention relates to an evaluation and prediction method for an aircraft braking system based on fault simulation and geometric features, and belongs to the technical field of electrical data processing. Background Art

[0002] With the development of modern aviation technology, the aircraft braking system, as an important part of ensuring the safe completion of ground-phase tasks of the aircraft, its safety, reliability and stability are crucial to flight safety. Timely and accurate assessment of the health status of the aircraft braking system and prediction of the remaining life can not only effectively avoid fault accidents and ensure flight safety, but also provide strong technical support for the maintenance support planning of the aircraft. Currently, the health assessment and remaining life prediction technologies for the braking system mainly include the method based on expert knowledge and the method based on data-driven. Among them, the method based on expert knowledge mainly uses the experience and rules of domain experts, combined with some performance monitoring parameters of the braking system to judge its health status. However, the limitation of this method lies in the high dependence on expert knowledge, and the construction of the rule knowledge base is relatively difficult, lacking flexibility and generalization. Another type of method is the data-driven health status assessment and remaining life prediction method. Due to the difficulty in collecting braking system data and obtaining full-life degradation data, the method based on the generative model generates samples similar to the actual life degradation data to alleviate the problem of data scarcity. However, this type of method has high requirements for the quality of the original data, and it is necessary to include sufficient monitoring data of different degradation stages in the original data to generate more comprehensive and accurate available data. Summary of the Invention

[0003] To overcome the shortcomings of the prior art, the invention purpose of the present invention is to provide an evaluation and prediction method for an aircraft braking system based on fault simulation and geometric features, and its invention purpose is to solve the problem of inaccurate prediction of the remaining life of the aircraft braking system caused by the difficulty in collecting braking system data and obtaining full-life degradation data.

[0004] To overcome the shortcomings of the prior art, the invention purpose of the present invention is to provide an evaluation and prediction method for an aircraft braking system based on fault simulation and geometric features, which includes: Step 1: Construct a simulation model of the aircraft braking system, obtain a set of typical fault modes, select fault injection control parameters according to the fault modes in the set of typical fault modes, and inject faults in the simulation model according to the selected fault injection control parameters of different degrees to generate a degradation monitoring data set for the full life cycle of the aircraft braking system; Step 2: Divide the degradation monitoring data set into a training data set and a test data set; Step 3: Obtain the training fusion feature set from the training data set, and use the training fusion feature set to train the GoogLeNet model to obtain a health assessment model; obtain the test geometric feature set from the test data set, and the health assessment model evaluates the health of the aircraft brake system according to the test geometric feature set; Step 4: Construct a topology graph based on the full-life geometric features and the health assessment results of the N-flight data of the aircraft brake system; cluster the nodes in the topology graph into K clusters, where K is greater than or equal to 3; establish a remaining life prediction model for each cluster; obtain the sample graph node features from the samples to be tested, perform similarity matching between the sample graph node features and the K clusters, and select the remaining life prediction model established by the best-matched cluster to predict the remaining life of the aircraft brake system, where N is a positive integer greater than or equal to 2.

[0005] Compared with the prior art, the aircraft brake system evaluation and prediction method based on fault simulation and geometric features provided by the present invention solves the problem of inaccurate monitoring of the health state of the aircraft brake system caused by the difficulty in collecting brake system data and the difficulty in obtaining full-life degradation data through the above technical solutions, so as to accurately predict the remaining life of the aircraft brake system. Description of the Drawings

[0006] Figure 1 is a flowchart of the aircraft brake system evaluation and prediction method based on fault simulation and geometric features provided by the present invention.

[0007] Figure 2 is a graph showing the relationship between the control current of sensor drift fault injection and the braking time provided by the present invention.

[0008] Figure 3 is a graph showing the relationship between the braking pressure of sensor drift fault injection and the braking time provided by the present invention.

[0009] Figure 4 is a graph showing the relationship between the wheel speed of sensor drift fault injection and the braking time provided by the present invention.

[0010] Figure 5 is a graph showing the relationship between the control current of electro-hydraulic servo valve fault injection and the braking time provided by the present invention.

[0011] Figure 6 is a graph showing the relationship between the braking pressure of electro-hydraulic servo valve fault injection and the braking time provided by the present invention.

[0012] Figure 7 is a graph showing the relationship between the wheel speed of electro-hydraulic servo valve fault injection and the braking time provided by the present invention. Detailed Embodiments

[0013] To make the technical means, creative features, achieved objectives and functions of the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0014] In the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0015] Figure 1 is a flowchart of the aircraft brake system evaluation and prediction method based on fault simulation and geometric features provided by the present invention. As Figure 1 shown, the aircraft brake system evaluation and prediction method based on fault simulation and geometric features provided by the present invention includes: Step 1: Build a simulation model of the aircraft brake system, obtain a set of typical fault modes, select fault injection control parameters according to the fault modes in the set of typical fault modes, and inject faults in the simulation model according to the selected fault injection control parameters at different levels to generate a degradation monitoring data set for the entire life cycle of the aircraft brake system; Step 2: Divide the degradation monitoring data set into a training data set and a test data set; Step 3: Obtain a training fusion feature set from the training data set, and use the training fusion feature set to train the GoogLeNet model to obtain a health assessment model; obtain a test geometric feature set from the test data set, and the health assessment model evaluates the health of the aircraft brake system according to the test geometric feature set; Step 4: Build a topology map according to the full-life geometric features and health assessment results of the N flight operation data of the aircraft brake system; cluster the nodes in the topology map into K clusters, where K is greater than or equal to 3; establish a remaining life prediction model for each cluster; obtain sample graph node features from the sample to be tested, perform similarity matching between the sample graph node features and the K clusters, and select the remaining life prediction model established by the best-matched cluster to predict the remaining life of the aircraft brake system, where N is a positive integer greater than or equal to 2.

[0016] In Step 1, the simulation model of the aircraft brake system includes the simulation model of the aircraft brake system, the wheel motion model, the hydraulic system model and the brake device model. Among them, the airframe mechanics model is used to analyze the forces in the aircraft's heading direction and vertical direction; the wheel motion model is mainly used to analyze the bonding force and the corresponding bonding moment at the tire-ground contact point; the hydraulic system model is used to analyze the automatic control characteristics of the current feedback loop based on the speed signal; the brake device model is used to analyze the braking ability.

[0017] In Step 1, the selection of failure modes in the typical failure mode set needs to be combined with the actual usage of the aircraft braking system and some expert knowledge to select failures with a relatively high probability of occurrence and that can be simulated by the simulation model. Failure modes include, for example, sensor drift failure and electro-hydraulic servo valve failure. For sensor drift failure, the core principle is that there is a deviation between the wheel speed data transmitted by the wheel speed sensor and the actual wheel speed data. As the degree of failure increases, the degree of this deviation will gradually expand. Electro-hydraulic servo valve failure mainly manifests as the weakening of the control ability of the electro-hydraulic servo valve on the hydraulic system due to aging, resulting in system failures. For the above two failure modes, control current, brake pressure, wheel speed, and brake time are selected as failure characterization parameters respectively. The four selected parameters can all reflect the overall health status of the aircraft braking system to a certain extent, and thus can serve as a strong support for evaluation and prediction.

[0018] In Step 1, the selection of fault injection control parameters needs to consider the working mechanism of typical failure modes, select appropriate control parameter variables, and finally, based on the selected fault injection control parameters, by changing the values of the fault injection control parameters, different degrees of fault injection are realized, so as to simulate the performance degradation process of the aircraft braking system in the whole life cycle and obtain the degradation monitoring data in the whole life cycle. For example, the greater the deviation between the wheel speed sensor and the actual wheel speed data, the deeper the degree of failure, and the weaker the control ability of the electro-hydraulic servo valve on the hydraulic system, the deeper the degree of failure. Therefore, when the gain coefficient in the control system increases, the degree of sensor drift failure deepens, and when the gain coefficient decreases, the degree of electro-hydraulic servo valve failure deepens. Therefore, the gain coefficient is used as the fault injection control parameter, and different gain coefficients are selected to inject corresponding simulated faults into the wheel speed sensor and electro-hydraulic servo valve for the two failure modes respectively. The injected gain coefficient is used as a quantitative expression of the degree of failure.

[0019] In Step 1, the fault injection control parameter is denoted as C. Each time the value of C is set, a fault simulation can be completed. Suppose there is a set of fault injection control parameters which means that the aircraft fails after operating N flight cycles, that is, a total of N simulations are performed, and finally a performance degradation data set for the whole life cycle is obtained , where the data set obtained from the nth simulation , N is a positive integer greater than or equal to 5, and M is a positive integer greater than or equal to 2.

[0020] For example, such as Figure 2 、 Figure 3 and Figure 4As shown, for the sensor drift fault, the gain coefficient range of the sensor drift fault is selected as 1 - 1.34. When the gain coefficient is 1.0, the healthiest data of the aircraft braking system is obtained, that is, the health of the aircraft braking system is 1.0. Similarly, when the gain coefficient of the sensor drift fault is 1.34, the health degree of the aircraft braking system is 0. The fault interval is between 1.18 and 1.34 for the gain coefficient.

[0021] As Figure 5 , Figure 6 and Figure 7 shown, for the electro-hydraulic servo valve fault, the gain coefficient range of the sensor drift fault is selected as 1 - 0.7. When the gain coefficient is 1.0, the healthiest data of the aircraft braking system is obtained, that is, the health of the aircraft braking system is 1.0. Similarly, when the gain coefficient of the sensor drift fault is 0.7, the health degree of the aircraft braking system is 0. The fault interval is between 0.7 and 0.82 for the gain coefficient.

[0022] Since the fault characteristics of the two fault modes are different, in order to obtain more comprehensive degradation data over the entire life cycle, it is finally determined that for the sensor drift fault, the total number of simulation times is 341, and the simulation data is 341 groups; for the electro-hydraulic servo valve fault, the total number of simulation times is 251, and the simulation data obtained is 251 groups.

[0023] In step 2, the training set ratio r is selected, and the performance degradation data set D over the entire life cycle is divided into a training data set and a test data set . , . For example, for the health assessment task, r = 0.7, that is, the data of 238 flight operations are shuffled and used as the training set, and the remaining 113 groups of flight operation data are used as the test set; for the remaining life prediction task, r = 0.6, that is, the data of 150 flight operations are used as the training set, and the remaining 101 groups of data are used as the test set.

[0024] In step 2, the data in the training data set and the test data set are normalized for subsequent processing. The normalization process is, for example, the maximum-minimum normalization. For example, the normalization process for the data in the training data set includes: Obtain the maximum value and the minimum value of the data in the training data set , then the normalized training data set , where .

[0025] In step 3, the fused features in the training fused feature set are the concatenation of the training geometric features extracted from the degradation monitoring data in the training dataset and the high-dimensional robust health representations extracted from the training geometric features.

[0026] Taking the extraction of geometric features from the monitoring data in the training dataset as an example, the extraction of geometric features includes: S3-1: Aggregate the time-domain information of the monitoring data trajectory formed by the degradation monitoring data in the degradation monitoring dataset of the aircraft brake system through a sliding window to obtain a time-domain index sequence. Specifically, perform principal component extraction on the normalized training dataset to obtain the principal component normalized training dataset , and cut it with a sliding window of width W and step size ST to obtain segments. For each segment, extract time-domain indices to obtain a time-domain index sequence; S3-2: Divide the time-domain index sequence with the inflection point as the segmentation point to obtain P segments, and perform trend modeling on the time-domain index sequence of each segment to obtain the time-domain index curve of each segment, where P is a positive integer greater than or equal to 2; S3-3: Extract the training geometric features of the aircraft brake system according to the time-domain index curve of each segment, and analyze the change law of the training geometric features of the aircraft brake system to fit the degradation characteristics of the health state of the aircraft brake system.

[0027] The geometric features include, for example, the average slope of the curve, the area of the curve, etc. The calculation formula for the average slope of the curve is as shown in Formula 1: Formula 1, where, represents the extracted value of the fault characterization parameter at the end time T of the p-th segment, represents the extracted value of the fault characterization parameter at the start time 0 of the p-th segment, and T represents the total duration of the segment.

[0028] The calculation formula for the curve area is as shown in Formula 2: Formula 2, where, represents the extracted value of the fault characterization parameter at time t.

[0029] Finally, the geometric features of the p-th segment are: .

[0030] The high-dimensional robust health representation is extracted from the geometric features through a high-dimensional robust health representation extraction model. The high-dimensional robust health representation extraction model is trained by a compressed denoising autoencoder model, and its training process includes: S3-4: Obtain the coding features of the aircraft braking system according to Formula 3: , Formula 3, where, is the activation function of the encoder; and are the parameters of the encoder; is the geometric feature of the p-th segment of the aircraft braking system with added noise; S3-5: Obtain the reconstruction features of the decoder according to Formula 4: Formula 4, where, is the activation function of the decoder; and are the parameters of the decoder; S3-6: Calculate the first loss function value according to the reconstruction features and the geometric features ; ; S3-7: Judge whether the first loss function value is the minimum. If so, the features output by the encoder are high-dimensional robust health representations; if not, modify the parameters , , and values, and then return to step S3-4.

[0031] Finally, obtain the fusion feature of the p-th segment: , where, represents concatenation, and the fusion features of P segments are concatenated again to obtain the training fusion feature .

[0032] In step 3, the health assessment model includes a feature extraction module and a degradation state measurement module; the feature extraction module is configured to obtain performance degradation features according to the fusion features; the degradation state measurement module is configured to obtain the health degree of the aircraft braking system according to the performance degradation features.

[0033] The feature extraction module being configured to obtain performance degradation features according to the fusion features includes: S3-8: Perform initial convolution and max pooling on the fusion features through Formula 5 to obtain the first feature : Formula 5, where, represents the training fusion feature; represents the convolution weight, Represents the convolutional bias term; Represents convolution; Represents max pooling; S3-9: For the first feature Perform multi-scale feature extraction on the first feature through Equation 6 to obtain a second feature set , where I is a positive integer greater than or equal to 2; for the first feature Perform multi-scale pooling on the first feature through Equation 7 to obtain a third feature set , where J is a positive integer greater than or equal to 1; preferably, I = 4, J = 1, where, Equation 6, In the formula, Represents a convolution with a convolutional kernel of ; Represents activation; Represents a convolution with a convolutional kernel of ; Equation 7, In the formula, Represents max pooling with a kernel of ; S3-10: Concatenate all the second features in the second feature set to obtain a first feature vector, concatenate all the third features in the third feature set to obtain a second feature vector, and then concatenate the first feature vector and the second feature vector to obtain a performance degradation feature .

[0034] The degradation state measurement module is configured to obtain the health of the aircraft braking system based on the performance degradation feature, including: S3-11: Obtain a fourth feature according to the performance degradation feature through Equation 8: Equation 8, In the formula, Represents the degradation feature; Represents the first fully connected weight, Represents the first fully connected bias term; Is an activation function; S3-12: Obtain the predicted health of the nth flight of the aircraft braking system according to the fourth feature through Equation 9: Equation 9, In the formula, Represents the second fully connected weight, Represents the second fully connected bias term; Represents a mapping function. The value range is between [0, 1]. The closer the health is to 1, the healthier the system state is, and the closer it is to 0, the more severely the system state has degraded; Optionally, the predicted health H of the aircraft braking system is obtained through steps 3-12; S3-13: Label the health of the aircraft braking system for the nth flight according to the fault injection control parameter as: Formula 10, In the above formula is the fault injection control parameter for the nth flight, is the sequence of fault injection control parameters for all fault simulation injection flights, is the true health label for the nth flight: S3-14: Construct the second loss function according to Formula 11: Formula 11, In the above formula, is the number of training samples (the number of training flights); S3-15: Determine whether the second loss function is the minimum. If so, output the optimal parameters , , , ; otherwise, adjust the parameters , , , and return to step S3-11.

[0035] In the present invention, the GoogLeNet model is trained into a health assessment model through steps S3-4 to step S2-15.

[0036] In step 3, the health state of the braking system is measured through the health assessment model, specifically including inputting the geometric features of the test samples into the health assessment model to obtain the health state assessment results of each test sample. The present invention uses the mean absolute error to measure the accuracy of the model, and the calculation method is as shown in Formula 12: Formula 12, In the above formula, , are respectively the predicted value and the true value of the health of the braking system obtained according to the test sample i.

[0037] Step 4 aims to construct a topological graph structure by using the geometric feature data and the health assessment results of the aircraft braking system operation flight data, so as to be used for the subsequent remaining life prediction model to mine the life change law of the aircraft braking system from the graph data, so as to realize a more accurate remaining life prediction.

[0038] Step 4 includes: S4-1: Construct a topological graph structure based on the full-life geometric feature data and the health assessment results of the N flight operation data of the aircraft braking system, so as to be used for the subsequent remaining life prediction model to mine the life change law of the aircraft braking system from the graph data, thereby achieving a more accurate remaining life prediction; S4-2: Based on the constructed health state topological graph, cluster and partition the N flight operation samples through the graph structure clustering method, and build an independent remaining life prediction model within each cluster, thereby realizing the construction of a more refined remaining life prediction model; S4-3: Extract the sample graph node features from the sample data, perform similarity matching between the sample graph node features and K clusters, select the remaining life model of the most matching cluster, and use it to predict the remaining life of the aircraft braking system. This method can realize large-sample transfer modeling based on health state similarity, and improve the prediction accuracy and robustness.

[0039] Step S4-1 specifically includes: S4-1-1: Regard each flight operation as a node in the graph. The features of each node include geometric features and the health of the braking system. Concatenate the above two types of features to form graph node features: , In the formula: is the graph node feature; represents the geometric feature; H represents the health assessment result; represents the vector concatenation operation.

[0040] Similarly, extract the geometric features for the test samples, and conduct a health assessment on the braking system according to the test samples. Concatenate the geometric features and the health assessment results, then there is the sample graph node feature . The graph node features comprehensively reflect the original degradation trend and deep potential feature structure of the braking system, which is conducive to improving the accuracy of subsequent remaining life prediction.

[0041] S4-1-2: Cluster the nodes in the topological graph into K clusters, connect the nodes in the K clusters with undirected edges to obtain the health state topological graph G=(V,E), where V represents the set of all flight operation nodes, and E represents the edge set constructed based on feature similarity. The K is greater than or equal to 3; establish a remaining life prediction model for each cluster.

[0042] Step S4-2 specifically includes: S4-2-1: Divide the nodes into K clustering clusters Each clustering cluster represents a set of flight operation data with similar structures and health state trends in the health state space, providing a basis for the construction and training of the subsequent remaining life prediction model; S4-2-2: Build a remaining useful life prediction model, which is trained by a convolutional attention network (Attention-CNN); S4-2-3: Use the remaining useful life prediction model to predict the remaining useful life of the aircraft braking system.

[0043] In the present invention, the training process of training the convolutional attention network into a remaining useful life prediction model includes: S4-2-2-1: Input the graph node features into the input layer of the convolutional neural network, and extract the hidden features from the output layer , and calculate the query vector Q, key vector K, and value vector V according to the hidden features : ; ; ; In the formula, , and are the weights of the query vector Q, key vector K, and value vector V respectively; , and are the biases of the query vector Q, key vector K, and value vector V respectively; S4-2-2-2: Obtain the attention features according to the query vector Q, key vector K, and value vector V: ; In the formula, T represents transpose, is the training gradient, and p is the number of training times; S4-2-2-3: Input the attention features into the fully connected layer to construct the mapping relationship with the remaining useful life. The calculation process is as follows: , In the formula, is the predicted remaining useful life value of the nth flight of the aircraft braking system, and are the weight vector and bias vector of the fully connected layer respectively; S4-2-2-4: Construct the remaining useful life label according to the fault injection control parameter: , In the above formula is the fault injection control parameter of the nth flight, is the sequence of fault injection control parameters for all fault simulation injection flights, is the true remaining useful life label of the nth flight of the aircraft braking system; S4-2-2-5: Construct the third loss function according to the following formula: , In the above formula, is the number of training samples (the number of training sorties); S-2-2-6: Determine the third loss function whether it is the minimum. If so, output the optimal parameters , , , , , , , , otherwise adjust the parameters by the backpropagation gradient method and return to step S4-2-2-1.

[0044] The present invention trains the convolutional attention network into a remaining life prediction model through steps S4-2-2-1 to S4-2-2-6.

[0045] In the present invention, using the remaining life prediction model to predict the remaining life of the brake system specifically includes: S4-2-3-1: For the test sample to be predicted, obtain its graph node features , calculate the similarity between the graph node features and the centers of each clustering cluster. Let the set of feature vectors of each cluster center of the clustering cluster be: , then the calculation method of the clustering cluster closest to the test sample is as follows: , where dist(·) represents the Euclidean distance metric function; S4-2-3-2: After determining the clustering-matched cluster , input the graph node features into the input layer of the remaining life prediction model of the cluster , and the output layer outputs the corresponding remaining life prediction value of the aircraft brake system. Calculate the fourth loss function according to formula 12: Formula 12, In the above formula, , are respectively the true value and the predicted value of the remaining service life of the aircraft brake system obtained according to the test sample i.

[0046] Finally, through the above steps, the health assessment and remaining life prediction of all test samples can be completed.

[0047] The present invention further provides an aircraft brake system evaluation and prediction system based on fault simulation and geometric features, which includes a storage medium and one or more processors. The storage medium stores a computer program, and the computer program is called by one or more processors to implement the above-mentioned aircraft brake system evaluation and prediction method based on fault simulation and geometric features.

[0048] The present invention also provides a computer program product, which compiles the above-mentioned aircraft brake system evaluation and prediction method based on fault simulation and geometric features into a computer program called and executed by one or more processors using a computer language.

[0049] In the present invention, the combination of one or more of the above steps still belongs to the scope disclosed by the present invention.

[0050] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification only illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. An aircraft brake system evaluation and prediction method based on fault simulation and geometric features, characterized in that Including: Step 1: Construct a simulation model of the aircraft braking system, obtain a set of typical fault modes, select fault injection control parameters according to the fault modes in the set of typical fault modes, and inject faults in the simulation model according to the selected fault injection control parameters at different levels to generate a degradation monitoring data set for the entire life cycle of the aircraft braking system; Step 2: Divide the degradation monitoring data set into a training data set and a test data set; Step 3: Obtain a training fusion feature set from the training data set, and use the training fusion feature set to train the GoogLeNet model to obtain a health assessment model; obtain a test geometric feature set from the test data set, and the health assessment model evaluates the health of the aircraft braking system according to the test geometric feature set; Step 4: Construct a topological graph according to the full-life geometric features and the health assessment results of the N flight data of the aircraft braking system; cluster the nodes in the topological graph into K clusters, where K is greater than or equal to 3; establish a remaining life prediction model for each cluster; obtain sample graph node features from the samples to be tested, perform similarity matching between the sample graph node features and the K clusters, and select the remaining life prediction model established by the best-matched cluster to predict the remaining life of the aircraft braking system, where N is a positive integer greater than or equal to 2.

2. The aircraft brake system evaluation and prediction method based on fault simulation and geometric features according to claim 1, characterized in that, The fusion feature in the training fusion feature set is the splicing of the training geometric features extracted from the degradation monitoring data in the training data set and the high-dimensional robust health representation extracted from the training geometric features.

3. The aircraft brake system evaluation and prediction method based on fault simulation and geometric features according to claim 2, wherein The extraction of the training geometric features includes: S3-1: Aggregate the time-domain information of the monitoring data trajectory formed by the degradation monitoring data in the degradation monitoring data set of the aircraft braking system through a sliding window to obtain a time-domain index sequence; S3-2: Divide the time-domain index sequence at the inflection point as the segmentation point to obtain P segments, and perform trend modeling on the time-domain index sequence of each segment to obtain the time-domain index curve of each segment, where P is a positive integer greater than or equal to 2; S3-3: Extract the training geometric features of the aircraft braking system according to the time-domain index curve of each segment, and analyze the change law of the training geometric features of the aircraft braking system to fit the degradation characteristics of the health state of the aircraft braking system.

4. The aircraft brake system evaluation and prediction method based on fault simulation and geometric features according to claim 3, characterized in that The high-dimensional robust health representation is extracted from the training geometric features of the aircraft braking system through a high-dimensional robust health representation extraction model, and the high-dimensional robust health representation extraction model is trained by a compressed denoising autoencoder model. Its training process includes: S3-4: Obtain the encoding features of the aircraft braking system according to the following formula: , , where is the activation function of the encoder; and are the parameters of the encoder; is the geometric feature of the p-th segment of the aircraft braking system feature with added noise; S3-5: Obtain the reconstructed features of the decoder according to the following formula: , In the formula, is the activation function of the decoder; and are the parameters of the decoder; S3-6: Calculate the first loss function value according to the reconstructed features and geometric features ; ; S3-7: Determine the first loss function value Whether it is the smallest. If so, the features output by the encoder are high-dimensional robust health representations; if not, modify the parameters , , and values, and then return to step S3-4.

5. The method for evaluating and predicting an aircraft braking system based on fault simulation and geometric features according to claim 1, wherein The health assessment model includes a feature extraction module and a degradation state measurement module; the feature extraction module is configured to obtain performance degradation features according to the training fusion features; The degradation state measurement module is configured to obtain the health of the aircraft braking system according to the performance degradation features.

6. The aircraft brake system evaluation and prediction method based on fault simulation and geometric features according to claim 5, characterized in that, The feature extraction module is configured to obtain performance degradation features according to the training fusion features, including: S3-8: Perform initial convolution and max pooling on the training fusion features to obtain the first feature : , In the formula, represents the training fusion feature; represents the convolution weight, represents the convolution bias term; represents the convolution; represents the max pooling; S3-9: Perform multi-scale feature extraction on the first feature to obtain a second feature set , where I is a positive integer greater than or equal to 2; perform multi-scale pooling on the first feature to obtain a third feature set , where J is a positive integer greater than or equal to 1; S3-10: Splice the second features in the second feature set to obtain a first spliced vector, splice the third features in the third feature set to obtain a second spliced vector, and then splice the first spliced vector and the second spliced vector to obtain performance degradation features.

7. The aircraft brake system evaluation and prediction method based on fault simulation and geometric features according to claim 6, characterized in that ， In the formula, represents the convolution with the convolution kernel ; represents activation; represents the convolution with the convolution kernel ; , where represents a max pooling with a kernel of .

8. The aircraft brake system evaluation and prediction method based on fault simulation and geometric features according to claim 7, wherein The degradation state measurement module is configured to obtain the health of the aircraft brake system according to the performance degradation characteristics, including: S3-11: Obtain the fourth feature based on the performance degradation characteristics: , In the formula, represents the performance degradation feature; represents the first fully connected weight, represents the first fully connected bias term, is the activation function; S3-12: Obtain the predicted health of the aircraft braking system for the nth flight based on the fourth feature: In the formula, represents the second fully-connected weight, represents the second fully-connected bias term; represents the mapping function.

9. The aircraft brake system evaluation and prediction method based on fault simulation and geometric features according to claim 1, wherein The remaining life prediction model is trained by a convolutional attention network.

10. The aircraft brake system evaluation and prediction method based on fault simulation and geometric features according to any one of claims 1-9, characterized in that, The simulation model of the aircraft brake system includes the simulation model of the aircraft brake system, the wheel motion model, the hydraulic system model and the brake device model. Among them, the airframe mechanics model is used to analyze the forces in the aircraft heading direction and the vertical direction; the wheel motion model is used to analyze the bonding force and the corresponding bonding moment at the tire-ground contact point; the hydraulic system model is used to analyze the automatic control characteristics of the current feedback loop based on the speed signal; the brake device model is used to analyze the braking ability.

Images