An aero-engine fault diagnosis method, system, terminal and medium

Abstract

The application relates to the field of fault diagnosis, and is specifically a kind of aero-engine fault diagnosis method, system, terminal and medium based on deep auto-encoder transfer learning technology, wherein a source domain basic model is trained by using a data set of a certain type of object, and on this basis, when the fault of the same type of object of different types needs to be diagnosed, only a small amount of target domain data set is needed, and the source domain basic model can be fine-tuned by combining the transfer learning technology, so that a target domain deep auto-encoder transfer learning model is obtained, the deep features of the target domain data are extracted, and finally, the fault diagnosis is realized by using an SVM classifier for classification. Compared with the traditional engine fault diagnosis method, the method has the advantages of easy migration of the diagnosis object, low requirement for the fault sample amount, high diagnosis precision and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of fault diagnosis, based on deep autoencoder transfer learning technology, and particularly to an aero-engine fault diagnosis method, system, terminal and medium. BACKGROUND

[0002] An aero-engine system is a complex system, including a mechanical transmission system, a sensor monitoring system, an energy system and an electronic monitoring system, etc. The frequent use of aero-engines under drastic and harsh conditions increases the frequency of aero-engine failures. The current fault diagnosis objects of aero-engines mainly include three types: faults of a gas path system, faults of a mechanical rotor system, and sensor faults. Due to the different models of aero-engines used by different aircraft, even the different models of sensors or rotor bearings used in a certain aero-engine on the same aircraft. The traditional fault diagnosis algorithm can only diagnose the faults of a specific object based on the object data used during training, and cannot diagnose the faults of other object data, which is too limited in diagnosis objects. SUMMARY

[0003] In view of the problem that the traditional fault diagnosis algorithm in the prior art is too limited in diagnosis objects and cannot diagnose the faults of other object data during training, the present application provides an aero-engine fault diagnosis method, system, terminal and medium, which can be extended to similar target domain objects of different models and diagnose other object data, effectively improving the diagnosis accuracy.

[0004] The present application is achieved by the following technical solutions:

[0005] An aero-engine fault diagnosis method, comprising the following steps:

[0006] Establishing a source domain data set and a target domain data set;

[0007] Establishing a source domain deep autoencoder basic fault classifier model through the source domain data set, and performing fault classification on the source domain deep autoencoder basic fault classifier model to obtain a classification result;

[0008] Training the source domain deep autoencoder basic fault classifier model with the target domain data set to obtain a target domain deep autoencoder transfer learning model, and performing fault classification on the target domain deep autoencoder transfer learning model to obtain a classification result;

[0009] Comparing the classification result obtained by the source domain deep autoencoder basic fault classifier model with the classification result obtained by the target domain deep autoencoder transfer learning model, and inputting the comparison error as error data into the target domain deep autoencoder transfer learning model for training and verification to obtain a target domain deep autoencoder transfer learning model with the best performance;

[0010] The target domain corresponding to different data types of the same fault diagnosis object is input into the target domain deep autoencoder transfer learning model with the best performance to train and obtain the classification result corresponding to the target domain to which the fault classification is performed, abnormal data in the classification result is extracted, and the fault diagnosis work is completed.

[0011] Preferably, the source domain data set and the target domain data set are established for the aero-engine object in the health management data set.

[0012] Preferably, the process of establishing the source domain deep autoencoder basic fault classifier model is as follows:

[0013] The source domain data set is divided into a source domain training set and a source domain test set, the source domain training set is trained to obtain a deep autoencoder basic model, and the deep autoencoder basic model is verified through the source domain test set.

[0014] Preferably, the source domain deep autoencoder basic fault classifier model performs fault classification through an SVM classifier, and the SVM classifier classifies the hidden layer of the last layer of the source domain deep autoencoder basic fault classifier model to obtain a classification result.

[0015] Preferably, the process of establishing the target domain deep autoencoder transfer learning model is as follows:

[0016] The target domain data set is divided into a target domain training set and a target domain test set, the target domain training set is input into the source domain deep autoencoder basic fault classifier model for training, and the target domain deep autoencoder transfer learning model is obtained by testing the source domain deep autoencoder basic fault classifier model after training through the target domain test set.

[0017] Preferably, the target domain deep autoencoder transfer learning model performs fault classification through an SVM classifier, and the SVM classifier classifies the hidden layer of the last layer of the target domain deep autoencoder transfer learning model to obtain a classification result.

[0018] An aero-engine fault diagnosis system, comprising:

[0019] The establishment module is configured to establish a source domain data set and a target domain data set;

[0020] The first processing module is configured to establish a source domain deep autoencoder basic fault classifier model through the source domain data set, and perform fault classification on the source domain deep autoencoder basic fault classifier model to obtain a classification result;

[0021] The second processing module is configured to train the source domain deep auto-encoder basic fault classifier model with the target domain data set to obtain a target domain deep auto-encoder transfer learning model, and perform fault classification on the target domain deep auto-encoder transfer learning model to obtain a classification result.

[0022] The third processing module is configured to extract deep features of the target domain deep auto-encoder transfer learning model, perform fault classification on the deep features, and output the deep features to complete fault diagnosis.

[0023] The fourth processing module is configured to input different data models corresponding to the same fault diagnosis object into the target domain deep auto-encoder transfer learning model with the best performance to train and perform fault classification, obtain abnormal data in the classification result, and complete fault diagnosis.

[0024] A mobile terminal comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the aero-engine fault diagnosis method when executing the computer program.

[0025] A computer readable storage medium stores a computer program, and the computer program implements the steps of the aero-engine fault diagnosis method when executed by a processor.

[0026] Compared with the prior art, the present application has the following beneficial technical effects:

[0027] The present application provides an aero-engine fault diagnosis method, which trains a source domain basic model with a data set of a certain type of object, and on this basis, when diagnosing faults of different types of the same object, only a small amount of target domain data set is needed, combined with transfer learning technology, the source domain basic model can be fine-tuned to obtain a target domain deep auto-encoder transfer learning model, deep features of the target domain data can be extracted, and finally, an SVM classifier is used for classification to realize fault diagnosis.

[0028] Further, the model transfer can be conveniently extended to the target domain objects of the same type but different types, avoiding the complexity of retraining the model on the target domain objects, and meeting the demand of rapid fault diagnosis. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1The aviation engine fault diagnosis method flowchart in the application;

[0030] Figure 2 The aviation engine fault diagnosis system structure diagram in the application. DETAILED DESCRIPTION

[0031] In order to make the personnel in the art better understand the application scheme, the technical solutions in the embodiments of the application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the application.

[0032] It should be noted that the terms "first", "second" and the like in the specification and claims of the application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0033] The application will be further described in detail below in conjunction with the accompanying drawings:

[0034] The application provides an aviation engine fault diagnosis method, system, terminal and medium, which can be extended to similar different types of target domain objects, diagnose other object data, and effectively improve the diagnosis accuracy.

[0035] Specifically, according to Figure 1 As shown in the figure, the aviation engine fault diagnosis method comprises the following steps:

[0036] S1, establishing a source domain data set and a target domain data set;

[0037] Two sets of data sets, called source domain data set and target domain data set, are established in the health management data set by using two types of aviation engine objects. A deep auto-encoder base model is established through the source domain data set, and a deep auto-encoder transfer learning model suitable for the target domain object is established by fine-tuning the base model using part of the target domain data set.

[0038] S2, establishing a source domain deep auto-encoder basic fault classifier model through the source domain data set, and performing fault classification on the source domain deep auto-encoder basic fault classifier model to obtain a classification result;

[0039] The source domain data set is divided into a source domain training set and a source domain test set, the deep auto-encoder model is trained by using the source domain training set, and the source domain test set is used for testing and verification, and a source domain deep model with good performance is obtained.

[0040] The deep auto-encoder basic model is stacked by three hidden layers, wherein the output of the i-th layer is used as the input of the i+1-th layer. The training mode is layer-by-layer training, and a SVM classifier is added after the output of the third hidden layer to realize fault category classification and obtain a classification result.

[0041] S3, training the source domain deep auto-encoder basic fault classifier model by using the target domain data set to obtain a target domain deep auto-encoder transfer learning model, and performing fault classification on the target domain deep auto-encoder transfer learning model to obtain a classification result;

[0042] S4, comparing the classification result obtained by the source domain deep auto-encoder basic fault classifier model with the classification result obtained by the target domain deep auto-encoder transfer learning model, and inputting the comparison error as error data into the target domain deep auto-encoder transfer learning model for training and verification to obtain a target domain deep auto-encoder transfer learning model with the best performance;

[0043] S5, inputting different data models corresponding to the same fault diagnosis object into the target domain deep auto-encoder transfer learning model with the best performance to train and classify the classification result corresponding to the target domain, extracting abnormal data in the classification result, and completing fault diagnosis work.

[0044] Embodiment

[0045] Taking a sensor in an aero-engine as an example, the fault categories include six categories of open circuit fault, short circuit fault, bias fault, impact fault, drift fault and periodic interference fault. Through the aero-engine fault diagnosis method, the steps are as follows:

[0046] Step 1: Create a health management data set. Models of the six fault types are established, and measurement signals under normal state and the six fault states of the sensor are collected. Seven states including the normal state are collected for 500 samples, and the training set and the test set are divided according to the ratio of 4:1. Similarly, another sensor of the same type is collected under the seven states, and 200 samples are collected for each state, and the training set and the test set are divided according to the ratio of 1:9.

[0047] Step 2: constructing a source domain deep auto-encoder basic fault classifier model. The deep auto-encoder basic model is stacked by three hidden layers, wherein the output of the i-th layer is used as the input of the i+1-th layer. The training mode is layer-by-layer training, and a SVM classifier is added after the output of the third hidden layer to realize fault class classification, i.e. fault diagnosis, and the diagnosis result is only used for further training of the target domain deep auto-encoder transfer learning model in step 3.

[0048] Step 3: constructing a target domain deep auto-encoder transfer learning model. The specific steps are as follows:

[0049] Step 3-1: migrating the model parameters such as weights and biases of the source domain deep auto-encoder basic model to the target domain deep auto-encoder transfer learning model.

[0050] Step 3-2: training the migration model obtained in step 3-1 using the target domain training set to improve the generalization performance of the model, and obtaining a target domain deep auto-encoder transfer learning model with better performance. The classification results obtained in step 2 and step 3 are compared, and the errors of the two can be used as data input to the target domain deep auto-encoder transfer learning model for further training.

[0051] Step 3-3: verifying the performance of the target domain deep auto-encoder transfer learning model using the target domain test set.

[0052] Step 4: according to the basic fault classifier model established in step 2, when a new same type fault diagnosis object appears, i.e. a new target domain appears, one third of the target domain data set can be used to fine-tune the source domain model to obtain a transfer learning model for fault diagnosis.

[0053] In summary, an aero-engine fault diagnosis method is provided, which trains a source domain basic model using a data set of an object of a certain type, and on this basis, when a fault of a same type object of a different type needs to be diagnosed, only a small amount of target domain data set is needed, combined with transfer learning technology, to fine-tune the source domain basic model to obtain a target domain deep auto-encoder transfer learning model, realize the extraction of deep features of the target domain data, and finally use a SVM classifier for classification to realize fault diagnosis. Compared with the traditional engine fault diagnosis method, the method provided by the present application has the advantages of easy migration of diagnosis object, low requirement for fault sample amount, high diagnosis precision, etc.

[0054] As shown in Figure 2 The present application also provides an aero-engine fault diagnosis system, which comprises an establishing module, a first processing module, a second processing module, a third processing module and a fourth processing module.

[0055] The establishing module is used for establishing a source domain data set and a target domain data set.

[0056] The first processing module is configured to establish a source domain deep auto-encoder basic fault classifier model through a source domain data set, and perform fault classification on the source domain deep auto-encoder basic fault classifier model to obtain a classification result.

[0057] The second processing module is configured to train the source domain deep auto-encoder basic fault classifier model through a target domain data set to obtain a target domain deep auto-encoder transfer learning model, and perform fault classification on the target domain deep auto-encoder transfer learning model to obtain a classification result.

[0058] The third processing module is configured to extract deep features of the target domain deep auto-encoder transfer learning model, perform fault classification on the deep features, and output the deep features to complete fault diagnosis.

[0059] The fourth processing module is configured to input different data models corresponding to a same fault diagnosis object into a target domain deep auto-encoder transfer learning model with the best performance, train the target domain deep auto-encoder transfer learning model, and perform fault classification on the target domain deep auto-encoder transfer learning model to obtain a classification result corresponding to the target domain, extract abnormal data in the classification result, and complete fault diagnosis.

[0060] The present application also provides a mobile terminal comprising a memory, a processor, and a computer program, such as an aero-engine fault diagnosis program, stored in the memory and executable on the processor.

[0061] The processor executes the computer program to implement the steps of the aero-engine fault diagnosis method, for example:

[0062] The source domain data set and the target domain data set are established.

[0063] The source domain deep auto-encoder basic fault classifier model is established through the source domain data set, and fault classification is performed on the source domain deep auto-encoder basic fault classifier model to obtain a classification result.

[0064] The source domain deep auto-encoder basic fault classifier model is trained through the target domain data set to obtain a target domain deep auto-encoder transfer learning model, and fault classification is performed on the target domain deep auto-encoder transfer learning model to obtain a classification result.

[0065] The classification result obtained by the source domain deep auto-encoder basic fault classifier model is compared with the classification result obtained by the target domain deep auto-encoder transfer learning model, and the comparison error is input as error data into the target domain deep auto-encoder transfer learning model for training and verification to obtain a target domain deep auto-encoder transfer learning model with the best performance.

[0066] The target domain corresponding to different data types of the same fault diagnosis object is input into the target domain deep auto-encoder transfer learning model with the best performance to train and obtain the classification result corresponding to the target domain to which the fault classification is performed, abnormal data in the classification result is extracted, and the fault diagnosis work is completed.

[0067] Alternatively, the processor implements the functions of the modules in the above system when executing the computer program, for example: the establishing module is configured to establish the source domain data set and the target domain data set;

[0068] The first processing module is configured to establish a source domain deep auto-encoder basic fault classifier model through the source domain data set, and perform fault classification on the source domain deep auto-encoder basic fault classifier model to obtain a classification result;

[0069] The second processing module is configured to train the source domain deep auto-encoder basic fault classifier model with the target domain data set to obtain a target domain deep auto-encoder transfer learning model, and perform fault classification on the target domain deep auto-encoder transfer learning model to obtain a classification result;

[0070] The third processing module is configured to extract deep features of the target domain deep auto-encoder transfer learning model, perform fault classification on the deep features, and output the deep features to complete the fault diagnosis work.

[0071] The fourth processing module is configured to input the target domain corresponding to different data types of the same fault diagnosis object into the target domain deep auto-encoder transfer learning model with the best performance to train and obtain the classification result corresponding to the target domain to which the fault classification is performed, extract abnormal data in the classification result, and complete the fault diagnosis work.

[0072] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present application. The one or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the mobile terminal. For example,

[0073] The computer program can be divided into an establishing module, a first processing module, a second processing module, a third processing module, and a fourth processing module, and the specific functions of each module are as follows:

[0074] The establishing module is configured to establish the source domain data set and the target domain data set;

[0075] The first processing module is configured to establish a source domain deep auto-encoder basic fault classifier model through the source domain data set, and perform fault classification on the source domain deep auto-encoder basic fault classifier model to obtain a classification result;

[0076] The second processing module is configured to train the source domain deep auto-encoder basic fault classifier model with the target domain data set to obtain a target domain deep auto-encoder transfer learning model, and perform fault classification on the target domain deep auto-encoder transfer learning model to obtain a classification result.

[0077] The third processing module is configured to extract deep features of the target domain deep auto-encoder transfer learning model, perform fault classification on the deep features, and output the deep features to complete fault diagnosis.

[0078] The fourth processing module is configured to input different data models of the same fault diagnosis object into a target domain deep auto-encoder transfer learning model with the best performance, train the target domain deep auto-encoder transfer learning model, and perform fault classification on the target domain deep auto-encoder transfer learning model to obtain a classification result corresponding to the target domain, extract abnormal data in the classification result, and complete fault diagnosis.

[0079] The mobile terminal can be a desktop computer, a notebook computer, a palm computer, a cloud server, and the like. The mobile terminal can include, but is not limited to, a processor and a memory.

[0080] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, or the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The processor is a control center of the mobile terminal, and connects all parts of the mobile terminal through various interfaces and lines.

[0081] The memory can be used to store computer programs and / or modules. The processor realizes various functions of the mobile terminal by running or executing the computer programs and / or modules stored in the memory, and calling data stored in the memory.

[0082] The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, application programs required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; and the data storage area can store data (such as audio data, a phone book, etc.) created according to the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a SmartMediaCard (SMC), a Secure Digital (SD) card, a Flash Card, at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0083] The application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the aero-engine fault diagnosis method.

[0084] The modules / units integrated in the mobile terminal, if realized in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium.

[0085] Based on such understanding, the application implements all or part of the processes in the above method, and can also be completed by a computer program instructing related hardware, the computer program can be stored in a computer readable storage medium, and the computer program can implement the steps of the aero-engine fault diagnosis method when executed by a processor. The computer program includes computer program codes, and the computer program codes can be in the form of source codes, object codes, executable files or some intermediate forms, etc.

[0086] The computer readable medium can include any entity or device, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. that can carry the computer program codes.

[0087] It should be noted that the content contained in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in a jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0088] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application but not to limit it. Although the present application has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that the specific embodiments of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and any modification or equivalent replacement should be covered in the protection scope of the claims of the present application.

Claims

1. A method for diagnosing faults in an aircraft engine, characterized in that, Includes the following steps: Establish source domain datasets and target domain datasets; A source domain deep autoencoder basic fault classifier model is established using the source domain dataset, and the source domain deep autoencoder basic fault classifier model is used to classify faults to obtain classification results. The source domain deep autoencoder basic fault classifier model is trained using the target domain dataset to obtain the target domain deep autoencoder transfer learning model, and the target domain deep autoencoder transfer learning model is used to classify faults to obtain the classification results. The classification results obtained by the source domain deep autoencoder basic fault classifier model are compared with the classification results obtained by the target domain deep autoencoder transfer learning model. The comparison error is used as error data and input into the target domain deep autoencoder transfer learning model for training and validation to obtain the target domain deep autoencoder transfer learning model with the best performance. The target domains corresponding to different data models of the same fault diagnosis object are input into the target domain deep autoencoder transfer learning model with the best performance, and the classification results corresponding to the target domain are obtained by classifying the fault. Abnormal data in the classification results are extracted to complete the fault diagnosis work. The source domain deep autoencoder basic fault classifier model classifies faults using an SVM classifier. The SVM classifier classifies the hidden layer of the last autoencoder in the source domain deep autoencoder basic fault classifier model to obtain the classification result. The process of establishing a target domain deep autoencoder transfer learning model is as follows: The target domain dataset is divided into a target domain training set and a target domain test set. The target domain training set is input into the source domain deep autoencoder basic fault classifier model for training. The source domain deep autoencoder basic fault classifier model trained with the target domain test set is tested to obtain the target domain deep autoencoder transfer learning model. The target domain deep autoencoder transfer learning model classifies faults using an SVM classifier. The SVM classifier classifies the hidden layer of the last autoencoder layer of the target domain deep autoencoder transfer learning model to obtain the classification result.

2. The method for diagnosing aircraft engine faults according to claim 1, characterized in that, In the health management dataset, source domain datasets and target domain datasets are created for aero-engine objects.

3. The method for diagnosing aircraft engine faults according to claim 1, characterized in that, The process of establishing the source domain deep autoencoder basic fault classifier model is as follows: The source domain dataset is divided into a source domain training set and a source domain test set. The deep autoencoder base model is trained on the source domain training set, and the deep autoencoder base model is validated using the source domain test set.

4. An aircraft engine fault diagnosis system, characterized in that, A method for diagnosing aero-engine faults according to any one of claims 1-3 includes: The module is used to create source domain datasets and target domain datasets; The first processing module is used to establish a source domain deep autoencoder basic fault classifier model through the source domain dataset, and to perform fault classification on the source domain deep autoencoder basic fault classifier model to obtain the classification result. The second processing module is used to train the source domain deep autoencoder basic fault classifier model with the target domain dataset to obtain the target domain deep autoencoder transfer learning model, and to perform fault classification on the target domain deep autoencoder transfer learning model to obtain the classification result. The third processing module is used to extract deep features from the target domain deep autoencoder transfer learning model, classify the deep features for faults and output them to complete the fault diagnosis work. The fourth processing module is used to input the target domains corresponding to different data models of the same fault diagnosis object into the target domain deep autoencoder transfer learning model with the best performance, train it, classify the fault into the target domain, extract abnormal data from the classification results, and complete the fault diagnosis work.

5. A mobile terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the aircraft engine fault diagnosis method as described in any one of claims 1-3.

6. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the aircraft engine fault diagnosis method as described in any one of claims 1-3.

Images