Intelligent diagnosis method and device for engine faults based on abnormal data screening

Abstract

The present application relates to an intelligent engine fault diagnosis method and device based on abnormal data screening, wherein the method includes: dividing the original data into non-fault flight data and fault flight data; constructing a feature vector space based on parameter combinations; delineating a safety boundary; judging whether the mapped sample points exceed the safety boundary based on the data distribution; if the mapped sample points exceed the safety boundary, determining the mapped sample points as fault samples and labeling them as fault samples; otherwise, determining the mapped sample points as non-fault samples; constructing a data sample set; training a support vector machine (SVM) classifier using the training set and corresponding labels, and generating an aircraft engine fault diagnosis result based on the final fault diagnosis model. This solves the problems in related technologies such as the small number of engine fault samples and the high acquisition cost, the difficulty in accurately classifying engine faults, and the complex and changeable operating conditions and unclear status of aircraft engine data.

Description

Technical Field

[0001] The present application relates to the technical field of fault diagnosis, and in particular to an intelligent engine fault diagnosis method and device based on abnormal data screening. Background Art

[0002] The engine is known as the "heart" of an aircraft, and its healthy and stable operation directly impacts flight safety and quality. On the one hand, aircraft engines face constantly changing operating conditions and operate in complex and harsh environments, making them prone to failure. On the other hand, engine maintenance is expensive, and even occasional failures can result in significant losses. Scientifically and reliably diagnosing engine failure risks has become a key concern for all parties in the aviation industry. Currently, the challenges facing engine maintenance, safety, and economic efficiency are becoming increasingly prominent, making research on engine fault diagnosis methods of great significance.

[0003] Related technologies for aircraft engine fault diagnosis can be broadly categorized into model-based and data-based approaches. Model-based approaches primarily rely on physical analysis models or mathematical models of aircraft engines. The drawback of these approaches is their high reliance on prior knowledge, resulting in poor generalizability. Furthermore, as engine system complexity increases, the modeling process becomes increasingly difficult. Data-driven anomaly monitoring technology, on the other hand, effectively avoids the drawbacks of modeling difficulties, high costs, and limited applicability by collecting data from various online monitoring sensors to train models, thus offering broad engineering applications. In recent years, various classification models, such as artificial neural networks, support vector machines (SVMs), and extreme learning machines (ELMs), have demonstrated promising results. With the increasing interest in deep learning technology, a growing number of algorithms and innovative approaches have been introduced to the field of engine anomaly monitoring and fault diagnosis. These include convolutional neural networks, graph neural networks, recurrent neural networks, long short-term memory (LSTM) neural networks, autoencoders, generative adversarial networks, transformers, and various model fusion methods.

[0004] However, the performance and generalization capabilities of related technologies depend on sample quality. Fault samples are rare in actual aircraft engine applications, and the primary data source is monitoring results under normal conditions. These samples are also expensive to obtain. Under these class imbalances, deep neural network models that require extensive annotated data are often unsuitable. Furthermore, the complex and variable operating conditions of actual aircraft engine monitoring data, coupled with unclear data labels, also present challenges for data-driven engine fault diagnosis. Specifically, data in actual engineering applications is categorized and stored by flight, and data within the same flight may vary in status. Even in flights with clear faults, the actual fault samples may only account for a small fraction of the flight time, while the majority of the flight data remains normal. For example, for typical sudden faults such as bird strikes, the engine was operating normally before the incident. Therefore, when analyzing and processing data, it is not appropriate to simply assume that all data from flights with bird strikes is faulty data. The fault samples required by machine learning classifiers should be data that clearly demonstrates the fault. Mixing in normal data will distort the classification results. Summary of the Invention

[0005] The present application provides an intelligent engine fault diagnosis method and device based on abnormal data screening to solve the problems in related technologies, such as the small number of engine fault samples and high acquisition costs, the difficulty in accurately classifying engine faults, and the complex and changeable operating conditions of aviation engine data and unclear status.

[0006] The first embodiment of the present application provides an intelligent engine fault diagnosis method based on abnormal data screening, which is applied to the model building stage and includes the following steps: obtaining the original data of the aircraft engine and dividing the original data into non-fault flight data and fault flight data; performing multi-parameter feature mining on the target sample to generate parameters that meet the preset fault sensitive conditions, and generating parameter combinations based on the parameters that meet the preset fault sensitive conditions, so as to construct a feature vector space based on the parameter combinations; analyzing the non-fault flight data to generate cluster centers, and delineating safety boundaries in the feature vector space based on the cluster centers and a preset safety margin; mapping the sample points in the fault flight data under preset working conditions into the feature vector space to determine the mapped sample points, and based on the data According to the distribution, whether the mapped sample point exceeds the safety boundary is judged; if the mapped sample point exceeds the safety boundary, the mapped sample point is judged to be a fault sample, and the fault sample is marked to determine the label of the fault sample, otherwise the mapped sample point is judged to be a non-fault sample; a data sample set is constructed according to the non-fault sample, the label of the non-fault sample, the fault sample and the label of the fault sample, and the data sample set is processed to generate a processed data sample set; the processed data sample set is divided into a training set and a test set, and a support vector machine (SVM) classifier is trained using the training set and the corresponding labels to obtain a fault diagnosis model, and the test set is input into the trained fault diagnosis model to generate a final fault diagnosis model.

[0007] Optionally, in one embodiment of the present application, multi-parameter feature mining is performed on the target sample to obtain parameters that meet preset fault-sensitive conditions, and a parameter combination is generated based on the parameters that meet the preset fault-sensitive conditions to construct a feature vector space based on the parameter combination, including: obtaining the parameter distribution law of the aircraft engine, and screening out the parameters that meet the preset fault-sensitive conditions from the parameter distribution law; combining the parameters that meet the preset fault-sensitive conditions to generate the parameter combination, and establishing feature vector coordinates based on the parameter combination to construct the feature vector space based on the feature vector coordinates.

[0008] Optionally, in one embodiment of the present application, the defining of the safety boundary in the feature vector space based on the cluster center and the preset safety margin includes: based on the cluster center and the range of the preset safety margin on each feature dimension, defining the distribution boundary of the non-fault samples in the feature vector space, and determining the safety boundary based on the distribution boundary.

[0009] Optionally, in one embodiment of the present application, before constructing a data sample set based on the non-fault samples, the labels of the non-fault samples, the fault samples and the labels of the fault samples, it also includes: filtering out the fault data from the fault flight data, and determining the labels of the fault samples based on the fault type of the fault data.

[0010] The second aspect of the present application provides an intelligent engine fault diagnosis method based on abnormal data screening, which is applied to the model application stage and includes the following steps: obtaining a test set and inputting the test set into the trained fault diagnosis model to generate a final fault diagnosis model, wherein the fault diagnosis model is constructed by training a support vector machine (SVM) classifier based on the training set and the corresponding label; and generating a fault diagnosis result of the aircraft engine according to the final fault diagnosis model.

[0011] The third embodiment of the present application provides an engine fault intelligent diagnosis device based on abnormal data screening, which is applied to the model construction stage, including: an acquisition module for acquiring the original data of the aircraft engine and dividing the original data into non-fault flight data and fault flight data; a construction module for performing multi-parameter feature mining on the target sample to generate parameters that meet the preset fault sensitive conditions, and generating parameter combinations based on the parameters that meet the preset fault sensitive conditions, so as to construct a feature vector space based on the parameter combinations; an analysis module for analyzing the non-fault flight data to generate cluster centers, and delineating safety boundaries in the feature vector space based on the cluster centers and a preset safety margin; a judgment module for mapping sample points in the fault flight data under preset working conditions into the feature vector space to determine the mapped sample points, and based on the Based on the data distribution, it is judged whether the mapped sample point exceeds the safety boundary; a determination module is used to determine that the mapped sample point is a fault sample when the mapped sample point exceeds the safety boundary, and mark the fault sample to determine the label of the fault sample, otherwise the mapped sample point is determined to be a non-fault sample; a processing module is used to construct a data sample set according to the non-fault sample, the label of the non-fault sample, the fault sample and the label of the fault sample, and process the data sample set to generate a processed data sample set; a generation module is used to divide the processed data sample set into a training set and a test set, use the training set and the corresponding label to train a support vector machine SVM classifier to obtain a fault diagnosis model, and input the test set into the trained fault diagnosis model to generate a final fault diagnosis model.

[0012] Optionally, in one embodiment of the present application, the construction module includes: an acquisition unit, used to obtain the parameter distribution law of the aircraft engine, and filter out the parameters that meet the preset fault-sensitive conditions from the parameter distribution law; a construction unit, used to combine the parameters that meet the preset fault-sensitive conditions to generate the parameter combination, and establish eigenvector coordinates based on the parameter combination to construct the eigenvector space based on the eigenvector coordinates.

[0013] Optionally, in one embodiment of the present application, the analysis module includes: a determination unit, configured to define the distribution boundary of the non-fault samples in the feature vector space based on the cluster center and the range of the preset safety margin on each feature dimension, and determine the safety boundary based on the distribution boundary.

[0014] Optionally, in one embodiment of the present application, it further includes: a screening module for screening out the fault data from the fault flight data before constructing a data sample set based on the non-fault samples, the labels of the non-fault samples, the fault samples and the labels of the fault samples, and determining the labels of the fault samples based on the fault types of the fault data.

[0015] A fourth embodiment of the present application provides an intelligent engine fault diagnosis device based on abnormal data screening, which is applied in the model application stage and includes: an input module for obtaining a test set and inputting the test set into a trained fault diagnosis model to generate a final fault diagnosis model, wherein the fault diagnosis model is constructed by training a support vector machine (SVM) classifier using the training set and corresponding labels; and a diagnosis module for generating a fault diagnosis result for the aircraft engine based on the final fault diagnosis model.

[0016] The fifth aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the intelligent diagnosis method for engine faults based on abnormal data screening as described in the above embodiment.

[0017] The sixth aspect of the present application provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, it implements the above-mentioned intelligent diagnosis method for engine faults based on abnormal data screening.

[0018] The seventh aspect of the present application provides a computer program product, which stores a computer program that, when executed by a processor, implements the above-mentioned intelligent diagnosis method for engine faults based on abnormal data screening.

[0019] The present invention, based on the concepts of flight parameter distribution and normal domain, can pre-classify operating conditions and filter and label abnormal data before fault classification. This provides stable and accurate data and its labels for subsequent classifiers, which are then input into a support vector machine (SVM) classifier to train the diagnostic model and achieve the goal of accurately classifying engine faults. This solves the problems in related technologies such as the limited number of aircraft engine fault samples and the high acquisition cost, making accurate engine fault classification difficult, and the complex and changeable operating conditions of aircraft engine data and unclear status.

[0020] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0022] Figure 1 A flowchart of an engine fault intelligent diagnosis method based on abnormal data screening provided in an embodiment of the present application applied to a model building stage;

[0023] Figure 2 Flowchart of an intelligent engine fault diagnosis method based on abnormal data screening according to one embodiment of the present application;

[0024] Figure 3 A flowchart of an engine fault intelligent diagnosis method based on abnormal data screening provided in an embodiment of the present application applied to a model application stage;

[0025] Figure 4 Schematic diagram of engine fault diagnosis test results according to one embodiment of the present application;

[0026] Figure 5 This is a structural diagram of an engine fault intelligent diagnosis device based on abnormal data screening provided by an embodiment of the present application applied to a model building stage;

[0027] Figure 6 This is a structural diagram of an engine fault intelligent diagnosis device based on abnormal data screening provided by an embodiment of the present application applied to a model application stage;

[0028] Figure 7 A schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0030] The following describes the engine fault intelligent diagnosis method and device based on abnormal data screening according to the embodiment of the present application with reference to the accompanying drawings. In response to the problems mentioned in the above background technology that the engine fault samples are few and the acquisition cost is high, it is difficult to achieve accurate classification of engine faults, and the operating conditions of aircraft engine data are complex and changeable and the status is unclear, the present application provides an engine fault intelligent diagnosis method based on abnormal data screening. In this method, before fault classification, the operating conditions can be first divided using the flight parameter distribution, and then abnormal data can be screened based on the idea of ​​the feature space safety domain to improve data quality. The screened data is then input into the SVM classifier to train the diagnostic model, thereby improving the accuracy of aircraft engine fault diagnosis. In this way, the problems such as the few engine fault samples and the high acquisition cost, the difficulty in achieving accurate classification of engine faults, the complex and changeable operating conditions of aircraft engine data and the unclear status are solved.

[0031] Specifically, Figure 1 A flow chart of an intelligent engine fault diagnosis method based on abnormal data screening provided in an embodiment of the present application.

[0032] Before introducing the intelligent engine fault diagnosis method based on abnormal data screening proposed in the embodiment of the present application, the relevant background is first introduced:

[0033] Traditional scheduled maintenance methods are highly inefficient, require enormous manpower, material, and financial resources, and cannot prevent flight accidents caused by aircraft engine failures. Data-driven, condition-based maintenance, which enables proactive preventive maintenance management, has received increasing attention. In recent years, the rapid development of artificial intelligence algorithms has brought new research hotspots and new avenues to this field. Common artificial intelligence fault diagnosis methods include neural networks and support vector machines (SVMs). SVMs perform well without requiring large amounts of data, making them ideal for engineering applications where accumulated historical engine data is scarce. The basic principle of SVMs is to find an optimal classification hyperplane between two data sets, ensuring that the hyperplane maintains a certain level of classification accuracy while maximizing the distance between the hyperplane and both data sets. This method can theoretically achieve optimal classification for linearly separable data.

[0034] like Figure 1 As shown, the engine fault intelligent diagnosis method based on abnormal data screening is applied to the model building stage and includes the following steps:

[0035] In step S101 , original data of the aircraft engine is obtained, and the original data is divided into non-fault flight data and fault flight data.

[0036] It can be understood that the non-fault flight data in the embodiment of the present application is normal flight data.

[0037] In actual implementation, the present embodiment can organize aircraft engine data and preliminarily divide the raw data into normal flight data and faulty flight data based on the actual engine operating status. Furthermore, considering the differences in flight conditions and mission profiles, the flight data is limited to operating conditions based on typical flight parameters, and data under similar operating conditions is selected for subsequent analysis and processing.

[0038] Typical flight parameters refer to parameters that can reflect the flight status, such as turbine after-temperature T1, high-pressure rotor speed n2, normal acceleration n y , airspeed v, etc.; similar operating conditions refer to dividing the engine data into multiple operating conditions based on the value range of typical flight parameters, and selecting data under the same operating condition for subsequent analysis and processing.

[0039] In the embodiment of the present application, before fault classification, the operating conditions are first divided using the flight parameter distribution, and then abnormal data is screened based on the idea of ​​feature space safety domain to improve data quality and provide stable and accurate data and its labels for subsequent classifiers.

[0040] In step S102, multi-parameter feature mining is performed on the target sample to generate parameters that meet preset fault sensitivity conditions, and parameter combinations are generated based on the parameters that meet the preset fault sensitivity conditions to construct a feature vector space based on the parameter combinations.

[0041] During the actual implementation process, the embodiment of the present application can perform multi-parameter feature mining on the target sample, select parameter combinations that are sensitive to faults, and construct a feature vector space based on the parameter combinations, thereby providing support for the subsequent demarcation of safety boundaries in the feature vector space.

[0042] It should be noted that the preset fault-sensitive conditions can be set by those skilled in the art according to actual conditions and are not specifically limited here.

[0043] Optionally, in one embodiment of the present application, multi-parameter feature mining is performed on the target sample to obtain parameters that meet preset fault-sensitive conditions, and parameter combinations are generated based on the parameters that meet the preset fault-sensitive conditions to construct a feature vector space based on the parameter combinations, including: obtaining the parameter distribution law of the aircraft engine, and screening out parameters that meet the preset fault-sensitive conditions from the parameter distribution law; combining the parameters that meet the preset fault-sensitive conditions to generate a parameter combination, and establishing feature vector coordinates based on the parameter combination to construct a feature vector space based on the feature vector coordinates.

[0044] During the actual execution process, the embodiment of the present application can obtain the distribution rules of various parameters of the engine, find out the three groups of characteristic parameters that are sensitive to typical faults, combine the three groups of characteristic parameters that are sensitive to typical faults, generate parameter combinations, and use the parameter combinations as the characteristic vector coordinates of each group of data to construct a characteristic vector space, where the characteristic vector space refers to the use of three groups of features p1, p2, and p3 to establish a spatial coordinate system O-p1p2p3.

[0045] In step S103 , the non-fault flight data is analyzed to generate cluster centers, and a safety boundary is defined in the feature vector space based on the cluster centers and a preset safety margin.

[0046] It is understandable that the embodiment of the present application may use a spectral clustering algorithm to perform cluster analysis on the non-fault flight data.

[0047] In the actual implementation process, the embodiment of the present application can use the spectral clustering method to perform cluster analysis on the non-fault flight data to obtain the cluster center, and calculate the standard deviation of each feature dimension of the sample data as a safety margin to define the safety boundary in the feature space.

[0048] Optionally, in one embodiment of the present application, a safety boundary is defined in the feature vector space based on the cluster center and a preset safety margin, including: based on the cluster center and the range of the preset safety margin on each feature dimension, such as the range of the standard deviation of the data in each feature dimension, the distribution boundary of non-fault samples is defined in the feature vector space, and the safety boundary is determined based on the distribution boundary.

[0049] It is understandable that in the embodiment of the present application, K=1 and the feature dimension is 3 to obtain the cluster center (h, k, l) of the normal sample distribution in the space O-p1p2p3. In the embodiment of the present application, the safety margin is set to the range (a, b, c) of 3 times the standard deviation from the cluster center on each feature dimension, a=3σ p1 ,b=3σ p2 ,c=3σ p3, the distribution boundary Θ obtained by training with normal samples in the feature vector space O-p1p2p3 is used, and the safety boundary is determined based on the distribution boundary. It should be noted that data distributed within the distribution boundary range is considered to be normal values.

[0050] In step S104, the sample points in the fault flight data under the preset working conditions are mapped into the feature vector space to determine the mapped sample points, and it is determined whether the mapped sample points exceed the safety boundary based on the data distribution.

[0051] It can be understood that the preset working conditions in the embodiments of the present application can be the same working condition.

[0052] During the actual execution process, the embodiment of the present application can map sample points under the same working conditions in the fault flight data to the feature vector space to determine the mapped sample points, and judge whether the mapped sample points exceed the safety boundary based on the data distribution, so as to conduct targeted engine fault diagnosis analysis according to the judgment results.

[0053] In step S105 , if the mapped sample point exceeds the safety boundary, the mapped sample point is determined to be a fault sample and marked as the fault sample to determine the label of the fault sample; otherwise, the mapped sample point is determined to be a non-fault sample.

[0054] It can be understood that, in the embodiment of the present application, exceeding the safety boundary may mean that the sample point falls outside the safety boundary Θ when mapped to the feature vector space O-p1p2p3.

[0055] During actual execution, the embodiment of the present application may determine that the mapped sample point is a fault sample when the mapped sample point exceeds the safety boundary, and mark the fault sample to determine the label of the fault sample.

[0056] In step S106 , a data sample set is constructed according to the non-fault samples, the labels of the non-fault samples, the fault samples, and the labels of the fault samples, and the data sample set is processed to generate a processed data sample set.

[0057] It can be understood that the data sample set in the embodiment of the present application can be a sample set for fault diagnosis constructed using normal data and its label "0", fault data and its labels "1, 2, L, N".

[0058] Specifically, the embodiments of the present application can construct a data sample set for fault diagnosis based on non-fault samples, labels of non-fault samples, fault samples and labels of fault samples, and normalize the data sample set to generate a processed data sample set, thereby providing support for the subsequent division of training sets and test sets.

[0059] Optionally, in one embodiment of the present application, before constructing a data sample set based on non-fault samples, labels of non-fault samples, fault samples and labels of fault samples, it also includes: filtering out fault data from fault flight data, and determining the labels of fault samples based on the fault type of the fault data.

[0060] In the actual implementation process, the embodiment of the present application can mark the fault samples, and mark the fault samples according to the sample P in the fault flight data. i The fault data is filtered out based on whether it exceeds the limit, and the sample P is set according to the fault type. i The labels are 1, 2, L, N, where N is the number of fault types.

[0061] In step S107, the processed data sample set is divided into a training set and a test set, and the support vector machine (SVM) classifier is trained using the training set and the corresponding labels to obtain a fault diagnosis model, and the test set is input into the trained fault diagnosis model to generate the final fault diagnosis model.

[0062] During the actual execution process, the embodiment of the present application can divide the normalized data sample set into a training set and a test set according to a certain ratio, set the SVM model hyperparameters, use the training set and the corresponding labels to train the support vector machine SVM classifier, iterate until the optimal, obtain the fault diagnosis model, and input the test set into the trained fault diagnosis model to generate the final fault diagnosis model, and use the test set data to verify the accuracy of the model.

[0063] The embodiment of the present application can input the filtered data into the SVM classifier to train the diagnosis model, thereby improving the accuracy of aircraft engine fault diagnosis.

[0064] Specifically, it can be combined Figure 2 As shown, the working principle of the engine fault intelligent diagnosis method based on abnormal data screening in the embodiment of the present application is described in detail with a specific embodiment.

[0065] Step S201: aircraft engine sortie data.

[0066] Step S202: Normal flight.

[0067] Step S203: fault number.

[0068] Step S204: Select similar working conditions.

[0069] Step S205: Study on multi-parameter distribution rules.

[0070] Step S206: Select characteristic parameters.

[0071] Step S207: Normal flight.

[0072] Step S208: Fault number.

[0073] Step S209: spectral clustering, cluster center, safety region and safety margin.

[0074] Step S210: spectral clustering, cluster centers, safety regions, and safety margins.

[0075] Step S211: Determine whether the limit is exceeded. If not, proceed to step S213.

[0076] Step S212: Determine whether the limit is exceeded. If so, proceed to step S214.

[0077] Step S213: Normal sample.

[0078] Step S214: Fault sample.

[0079] Step S215: Data sample set.

[0080] Step S216: training set.

[0081] Step S217: Test set.

[0082] Step S218: SVM classifier.

[0083] Step S219: Diagnosis result.

[0084] like Figure 3 As shown, the engine fault intelligent diagnosis method based on abnormal data screening is applied to the model application stage and includes the following steps:

[0085] In step S301, a test set is obtained and input into the trained fault diagnosis model to generate a final fault diagnosis model, wherein the fault diagnosis model is constructed by training a support vector machine (SVM) classifier using the training set and corresponding labels.

[0086] In step S302, a fault diagnosis result of the aircraft engine is generated according to the final fault diagnosis model.

[0087] Before performing engine fault diagnosis, the distribution of normal values ​​in the feature space is found based on the idea of ​​normal sample clustering. Fault samples are screened and labeled based on whether the data distribution exceeds the safety domain, improving data quality. The processed data and labels are then input into the classifier to achieve higher classification accuracy. Through the application of this application, the accuracy of aircraft engine fault diagnosis can be improved under sample-scarce conditions.

[0088] In order to verify the effectiveness of the proposed method, flight data was selected as a case study. In this case, after sorting out the historical flight data of aircraft engines, it was found that there were three operating states: normal operation, explosion failure, and wear failure, which were recorded as 0#, 1#, and 2# respectively. Among them, there were 4 groups of flights with number 0#, 1 group of flights with number 1#, and 3 groups of flights with number 2#. Then, considering the differences in flight conditions and mission profiles, according to the high-pressure rotor speed n2 and the normal acceleration n y The two indicators divide the data of each flight into multiple operating conditions. In this case, the high-speed and high-overload operating conditions are selected for subsequent analysis.

[0089] Next, multi-parameter feature mining was performed on the samples. Through parameter sensitivity analysis, flight parameter features such as "left engine exhaust temperature, left engine throttle lever displacement, and left engine high-pressure speed" were selected as the eigenvector coordinates for flight data numbered 1. Flight parameter features such as "true airspeed, lateral overload, and right engine low-pressure speed" were selected as the eigenvector coordinates for flight data numbered 2. This led to the construction of a feature vector space. Next, spectral clustering was used to cluster the normal flight data, obtaining cluster centers. By calculating the standard deviation of each feature dimension in the sample data, the 3σ rule was used to determine the safety margin and delineate the safety boundary in the feature space. In this case, the feature dimension was 3, which can be approximated using an ellipsoid model. The faulty flight data was mapped into the corresponding feature vector space. Anomaly monitoring was performed based on whether the data distribution exceeded the safety boundary, and faulty samples were screened and labeled. In the data for flight 1#, the sample points that exceed the feature space are labeled 1, representing a "deflagration fault"; in the data for flight 2#, the sample points that exceed the feature space are labeled 2, representing a "wear fault"; and in the data for flight 0#, the sample points that do not exceed the feature space are labeled 0, representing "normal data." These data and their corresponding labels together constitute the data sample set for the classifier. The constructed data sample set is normalized, and 80% is randomly selected as the training set, with the remaining 20% ​​as the test set. The training set and corresponding labels are used to train an SVM classifier to obtain a fault diagnosis model. The test set data is then fed into the fault diagnosis model to obtain diagnostic test results. Figure 4 The final test results are shown. Figure 4 It can be seen that after advance data labeling, the classification accuracy of explosion faults, wear faults and normal data has been greatly improved.

[0090] Next, an intelligent engine fault diagnosis device based on abnormal data screening proposed in accordance with an embodiment of the present application will be described with reference to the accompanying drawings.

[0091] Figure 5 It is a structural diagram of an engine fault intelligent diagnosis device based on abnormal data screening according to an embodiment of the present application applied to the model building stage.

[0092] like Figure 5 As shown, the engine fault intelligent diagnosis device 10 based on abnormal data screening includes: an acquisition module 100, a construction module 200, an analysis module 300, a judgment module 400, a determination module 500, a processing module 600 and a generation module 700.

[0093] Specifically, the acquisition module 100 is used to acquire original data of the aircraft engine and divide the original data into non-fault flight data and fault flight data.

[0094] The construction module 200 is used to perform multi-parameter feature mining on the target sample to generate parameters that meet the preset fault sensitivity conditions, and generate parameter combinations based on the parameters that meet the preset fault sensitivity conditions to construct a feature vector space based on the parameter combinations.

[0095] The analysis module 300 is used to analyze the non-fault flight data to generate cluster centers, and to define safety boundaries in the feature vector space based on the cluster centers and a preset safety margin.

[0096] The judgment module 400 is used to map the sample points in the fault flight data under the preset working conditions into the feature vector space to determine the mapped sample points, and judge whether the mapped sample points exceed the safety boundary based on the data distribution.

[0097] The determination module 500 is used to determine that the mapped sample point is a fault sample and mark the fault sample to determine the label of the fault sample when the mapped sample point exceeds the safety boundary; otherwise, the mapped sample point is determined to be a non-fault sample.

[0098] The processing module 600 is configured to construct a data sample set according to non-fault samples, labels of non-fault samples, fault samples and labels of fault samples, and process the data sample set to generate a processed data sample set.

[0099] The generation module 700 is used to divide the processed data sample set into a training set and a test set, use the training set and the corresponding labels to train the support vector machine (SVM) classifier to obtain a fault diagnosis model, and input the test set into the trained fault diagnosis model to generate the final fault diagnosis model.

[0100] Optionally, in one embodiment of the present application, the construction module 200 includes: an acquisition unit and a construction unit.

[0101] The acquisition unit is used to acquire the parameter distribution law of the engine and select parameters that meet the preset fault sensitivity conditions from the parameter distribution law.

[0102] The construction unit is used to combine parameters that meet preset fault-sensitive conditions to generate a parameter combination, and establish characteristic vector coordinates according to the parameter combination to construct a characteristic vector space according to the characteristic vector coordinates.

[0103] Optionally, in one embodiment of the present application, the analysis module 300 includes: a determination unit.

[0104] The determination unit is used to delineate the distribution boundary of non-fault samples in the feature vector space based on the cluster center and the range of the preset safety margin on each feature dimension, and determine the safety boundary according to the distribution boundary.

[0105] Optionally, in one embodiment of the present application, the engine fault intelligent diagnosis device 10 based on abnormal data screening further includes: a screening module.

[0106] Among them, the screening module is used to screen out fault data from the fault flight data before constructing a data sample set based on non-fault samples, labels of non-fault samples, fault samples and labels of fault samples, and determine the labels of fault samples according to the fault type of the fault data.

[0107] Figure 6 It is a structural diagram of an engine fault intelligent diagnosis device based on abnormal data screening according to an embodiment of the present application applied to the model application stage.

[0108] like Figure 6 As shown, the engine fault intelligent diagnosis device 20 based on abnormal data screening includes: an input module 800 and a diagnosis module 900.

[0109] The input module 800 is used to obtain a test set and input the test set into the trained fault diagnosis model to generate a final fault diagnosis model, wherein the fault diagnosis model is constructed by training a support vector machine (SVM) classifier based on the training set and the corresponding labels.

[0110] The diagnosis module 900 is used to generate a fault diagnosis result of the aircraft engine according to the final fault diagnosis model.

[0111] It should be noted that the aforementioned explanation of the embodiment of the engine fault intelligent diagnosis method based on abnormal data screening is also applicable to the engine fault intelligent diagnosis device based on abnormal data screening in this embodiment, and will not be repeated here.

[0112] The intelligent engine fault diagnosis device based on abnormal data screening, proposed in the embodiments of this application, can first use flight parameter distribution to divide operating conditions before fault classification. It then uses the concept of feature space safety domains to screen abnormal data to improve data quality. The screened data is then fed into an SVM classifier to train the diagnostic model, thereby improving the accuracy of aircraft engine fault diagnosis. This solves the problems of limited and expensive acquisition of engine fault samples, making accurate engine fault classification difficult, and the complex, changeable, and unclear operating conditions of aircraft engine data.

[0113] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0114] Memory 701 , processor 702 , and computer programs stored in the memory 701 and executable on the processor 702 .

[0115] When the processor 702 executes the program, the engine fault intelligent diagnosis method based on abnormal data screening provided in the above embodiment is implemented.

[0116] Furthermore, the electronic device further includes:

[0117] The communication interface 703 is used for communication between the memory 701 and the processor 702 .

[0118] The memory 701 is used to store computer programs that can be run on the processor 702 .

[0119] The memory 701 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0120] If the memory 701, processor 702, and communication interface 703 are implemented independently, the communication interface 703, memory 701, and processor 702 can be interconnected via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 7 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0121] Optionally, in a specific implementation, if the memory 701, the processor 702 and the communication interface 703 are integrated on a chip, the memory 701, the processor 702 and the communication interface 703 can communicate with each other through an internal interface.

[0122] The processor 702 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0123] This embodiment also provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the above-mentioned intelligent diagnosis method for engine faults based on abnormal data screening is implemented.

[0124] An embodiment of the present application also provides a computer program product, which stores a computer program. When the program is executed by a processor, it implements the above-mentioned intelligent diagnosis method for engine faults based on abnormal data screening.

[0125] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0126] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of this application, "N" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0127] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing a custom logical function or process step, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may be performed in a different order than shown or discussed, including performing functions in a substantially simultaneous manner or in a reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present application pertain.

[0128] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or N wires (electronic devices), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program can be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing it in other suitable ways as necessary, and then storing it in a computer memory.

[0129] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiment, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0130] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0131] In addition, the functional units in the various embodiments of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into a module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0132] The storage medium mentioned above may be a read-only memory, a magnetic disk, or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present application. Persons skilled in the art may make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. An intelligent engine fault diagnosis method based on abnormal data screening, characterized in that: Applied to the model building stage, wherein the method comprises the following steps: Acquiring raw data of the aircraft engine, and dividing the raw data into non-fault flight data and fault flight data; Performing multi-parameter feature mining on the target sample to generate parameters that meet preset fault-sensitive conditions, and generating parameter combinations based on the parameters that meet the preset fault-sensitive conditions, so as to construct a feature vector space based on the parameter combinations; Analyzing the non-fault flight data to generate cluster centers, and delineating a safety boundary in the feature vector space based on the cluster centers and a preset safety margin; Mapping sample points in the fault flight data under a preset working condition into the feature vector space to determine mapped sample points, and determining whether the mapped sample points exceed the safety boundary based on data distribution; If the mapped sample point exceeds the safety boundary, the mapped sample point is determined to be a fault sample, and the fault sample is marked to determine the label of the fault sample; otherwise, the mapped sample point is determined to be a non-fault sample; Constructing a data sample set according to the non-fault samples, the labels of the non-fault samples, the fault samples and the labels of the fault samples, and processing the data sample set to generate a processed data sample set; The processed data sample set is divided into a training set and a test set, the training set and the corresponding labels are used to train a support vector machine (SVM) classifier to obtain a fault diagnosis model, and the test set is input into the trained fault diagnosis model to generate a final fault diagnosis model.

2. The method according to claim 1, characterized in that The multi-parameter feature mining is performed on the target sample to obtain parameters that meet the preset fault-sensitive conditions, and a parameter combination is generated according to the parameters that meet the preset fault-sensitive conditions, so as to construct a feature vector space according to the parameter combination, including: Obtaining a parameter distribution law of the aircraft engine, and screening out the parameters that meet the preset fault-sensitive conditions from the parameter distribution law; The parameters satisfying the preset fault-sensitive conditions are combined to generate the parameter combination, and characteristic vector coordinates are established according to the parameter combination, so as to construct the characteristic vector space according to the characteristic vector coordinates.

3. The method according to claim 1, characterized in that Delineating a safety boundary in the feature vector space based on the cluster center and a preset safety margin includes: Based on the cluster center and the range of the preset safety margin in each feature dimension, the distribution boundary of the non-fault samples is delineated in the feature vector space, and the safety boundary is determined according to the distribution boundary.

4. The method according to claim 1, wherein Before constructing a data sample set according to the non-fault samples, the labels of the non-fault samples, the fault samples and the labels of the fault samples, the method further includes: Fault data is filtered out from the fault flight data, and labels of the fault samples are determined according to the fault types of the fault data.

5. An intelligent engine fault diagnosis method based on abnormal data screening, characterized in that: An intelligent engine fault diagnosis method based on abnormal data screening according to any one of claims 1 to 4 is applied in the model application stage, wherein the method comprises the following steps: Obtaining a test set and inputting the test set into the trained fault diagnosis model to generate the final fault diagnosis model, wherein the fault diagnosis model is constructed by training a support vector machine (SVM) classifier based on the training set and corresponding labels; A fault diagnosis result of the aircraft engine is generated according to the final fault diagnosis model.

6. An intelligent engine fault diagnosis device based on abnormal data screening, characterized in that: Applied to the model building stage, wherein the device includes: An acquisition module, configured to acquire raw data of the aircraft engine and divide the raw data into non-fault flight data and fault flight data; A construction module, configured to perform multi-parameter feature mining on a target sample to generate parameters that satisfy preset fault-sensitive conditions, and to generate parameter combinations based on the parameters that satisfy the preset fault-sensitive conditions, so as to construct a feature vector space based on the parameter combinations; an analysis module, configured to analyze the non-fault flight data to generate cluster centers, and to define a safety boundary in the feature vector space based on the cluster centers and a preset safety margin; a judgment module, configured to map sample points in the fault flight data under a preset working condition into the feature vector space to determine the mapped sample points, and to judge whether the mapped sample points exceed the safety boundary based on data distribution; a determination module, configured to determine, when the mapped sample point exceeds the safety boundary, that the mapped sample point is a fault sample and mark the fault sample to determine a label of the fault sample; otherwise, determine that the mapped sample point is a non-fault sample; a processing module, configured to construct a data sample set according to the non-fault samples, the labels of the non-fault samples, the fault samples and the labels of the fault samples, and process the data sample set to generate a processed data sample set; A generation module is used to divide the processed data sample set into a training set and a test set, use the training set and the corresponding labels to train a support vector machine (SVM) classifier to obtain a fault diagnosis model, and input the test set into the trained fault diagnosis model to generate a final fault diagnosis model.

7. An intelligent engine fault diagnosis device based on abnormal data screening, characterized in that: An intelligent engine fault diagnosis method based on abnormal data screening according to any one of claims 1 to 4 is applied in the model application stage, wherein the device comprises: An input module is used to obtain a test set and input the test set into the trained fault diagnosis model to generate the final fault diagnosis model, wherein the fault diagnosis model is constructed by training a support vector machine (SVM) classifier based on the training set and corresponding labels; The diagnosis module is used to generate a fault diagnosis result of the aircraft engine according to the final fault diagnosis model.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the intelligent engine fault diagnosis method based on abnormal data screening as described in any one of claims 1 to 4 or 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the engine fault intelligent diagnosis method based on abnormal data screening as described in any one of claims 1-4 or 5.

10. A computer program product comprising a computer program, characterized in that The computer program is executed to implement the engine fault intelligent diagnosis method based on abnormal data screening as described in any one of claims 1 to 4 or 5.

Images