Aero-engine bearing fault diagnosis method based on domain generalization

Abstract

The invention discloses an aero-engine bearing fault diagnosis method based on domain generalization. The aero-engine bearing fault diagnosis method comprises the steps of obtaining to-be-diagnosed bearing data; inputting the to-be-diagnosed bearing data into a fault diagnosis model to obtain a fault diagnosis result; wherein the fault diagnosis model comprises a feature extraction module and a condition comparison feature module, training the feature extraction module and the condition comparison feature module by using a training set, updating the feature extraction module and the condition comparison feature module in combination with a corresponding loss function, and obtaining the fault diagnosis model. The training set comprises bearing working data. According to the invention, fault diagnosis is carried out on the aero-engine bearing by using a domain generalization technology, independent sample collection and labeling operation for each working condition of the aero-engine are not needed, and a lot of manpower and material resources are saved.

Description

Technical Field

[0001] The invention belongs to the field of fault diagnosis, and in particular relates to an aircraft engine bearing fault diagnosis method based on domain generalization. Background Art

[0002] As the core component of an aircraft, the smooth operation of an aircraft engine is crucial to ensure flight safety. During the operation of the engine, rolling bearings are key components that bear huge loads. Their working environment is harsh and they are prone to failure. This may lead to serious flight accidents. Therefore, in-depth research on the fault diagnosis method of aircraft engine bearings will help to promptly discover the characteristics of early bearing failures, thereby effectively preventing accidents. Therefore, it is necessary to provide an aircraft engine bearing fault diagnosis method to solve the above technical problems;

[0003] The existing technologies have the following defects: 1. In the existing technologies, the basic assumption that the test data and the training data are independent and identically distributed is followed. Once this assumption is broken, the diagnostic accuracy of the existing methods will drop significantly; 2. The existing methods all need to collect a sufficient number of health status labels, but in the actual working environment of aircraft engines, it is impossible to collect enough data and mark enough labels for each task and each working condition; 3. There are methods based on domain adaptation to alleviate the above problems, but the domain adaptation method can only solve the problem of one source domain to one target domain, and the working conditions of aircraft engines vary a lot; 4. The application of domain adaptation technology can only perform offline fault diagnosis, and online fault diagnosis cannot be performed for current operating data; 5. The existing domain generalization technology does not extract domain discrimination information well, so that the generalization performance cannot meet the requirements; 6. The existing domain generalization technology extracts domain invariant features based on the feature distribution difference measurement method, and this method sometimes extracts information irrelevant to the task, ignores the differences between different classes, resulting in poor quality of domain invariant features and affecting domain generalization. Therefore, there is an urgent need for a method for fault diagnosis of aircraft engine bearings based on domain generalization. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes an aircraft engine bearing fault diagnosis method based on domain generalization, which can use domain generalization technology to diagnose faults of aircraft engine bearings. There is no need to collect samples separately for each operating condition of the aircraft engine and perform labeling operations, thus saving a lot of manpower and material resources.

[0005] To achieve the above objectives, the present invention provides an aircraft engine bearing fault diagnosis method based on domain generalization, comprising:

[0006] Obtain the bearing data to be diagnosed;

[0007] Inputting the bearing data to be diagnosed into a fault diagnosis model to obtain a fault diagnosis result;

[0008] Among them, the fault diagnosis model includes a feature extraction module and a condition comparison feature module. The feature extraction module and the condition comparison module are trained respectively using a training set. In combination with a corresponding loss function, the feature extraction module and the condition comparison module are updated to obtain the fault diagnosis model. The training set includes bearing working data.

[0009] Optionally, the bearing operating data includes: bearing speed, bearing load and acceleration signal.

[0010] Optionally, before using the training set to train the fault diagnosis model, the method further includes: labeling the training set to obtain a labeling category;

[0011] The marking categories are: fault type and fault size.

[0012] Optionally, the feature extraction module includes: a first feature extraction layer and a first classifier;

[0013] The first feature extraction layer is used to obtain multi-source domain features using a feature extractor;

[0014] The first classifier is used to adjust the multi-source domain features by using an improved cross entropy loss function to obtain a cross entropy loss.

[0015] Optionally, the improved cross entropy loss function is:

[0016]

[0017] Among them, L CCE is the total cross entropy loss in cross-domain learning, is the jth sample obtained from the i-th source domain, For its corresponding label, is the output of the sample observation after passing through the feature extraction layer and the classification layer, L2 is the cross entropy loss function, M is the number of different source domains, N i is the number of samples in the i-th source domain.

[0018] Optionally, the condition comparison module includes: a second feature extraction layer and a second classifier;

[0019] The second feature extraction layer is used to obtain sample features using a feature extractor;

[0020] The second classifier is used to adjust the sample features by using a conditional contrast function to obtain a conditional contrast loss.

[0021] Optionally, the conditional comparison function is:

[0022]

[0023] in, is the conditional contrast loss, is the specific loss for a single sample u, M is the number of different source domains, and N i is the number of samples in the i-th source domain.

[0024] Optionally, updating the feature extraction module and the condition comparison module includes:

[0025] Based on the cross entropy loss and conditional contrast loss and the SGD optimization algorithm, the feature extractor and the classifier are updated to obtain the updated feature extraction layer and classifier.

[0026] Optionally, the updated feature extraction layer and classifier are:

[0027]

[0028] Among them, f θ is the special diagnosis extraction layer, ε is the learning rate, L cc is the conditional contrast loss, α is the weight, L CCE is the cross entropy loss, g θ is the classifier, f θ ← is the updated feature extraction layer, g θ ← is the updated classifier.

[0029] Compared with the prior art, the present invention has the following advantages and technical effects:

[0030] (1) When an aircraft engine is running at different speeds and loads, the data generated will have different degrees of domain shift. Traditional deep learning fault diagnosis algorithms are not competent for fault diagnosis in this case. However, this method can perform high-accuracy fault diagnosis on aircraft engine bearing operation data under various working conditions with only one training.

[0031] (2) Currently, there are methods based on domain adaptation that can alleviate the problem of missing labels for some data. However, domain adaptation can usually only solve the problem of missing labels for some samples and cannot solve the problem of lack of training.

[0032] (3) The present invention can perform intelligent fault diagnosis in an online manner, which can not only collect data from sensors but also reflect the health status of aircraft engine bearings in real time.

[0033] (4) The present invention utilizes the improved softmax loss and compares it with the healthy state condition to achieve the optimal effect in fault diagnosis under unknown working conditions.

[0034] (5) The fault diagnosis method of the present invention does not require human intervention and has high processing efficiency. While improving the diagnostic accuracy, it reduces the data collection cost, thereby greatly improving the safety and reliability of the engine bearing. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0036] Figure 1 It is a flow chart of an aircraft engine bearing fault diagnosis method based on domain generalization according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0038] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0039] The present invention provides an aircraft engine bearing fault diagnosis method based on domain generalization, such as Figure 1 As shown, specifically including:

[0040] Obtain the bearing data to be diagnosed;

[0041] Input the bearing data to be diagnosed into the fault diagnosis model to obtain the fault diagnosis result;

[0042] Among them, the fault diagnosis model includes a feature extraction module and a condition comparison feature module. The feature extraction module and the condition comparison module are trained respectively using the training set. Combined with the corresponding loss function, the feature extraction module and the condition comparison module are updated to obtain the fault diagnosis model. The training set includes bearing working data.

[0043] Specifically, the trained model can be applied to many new working conditions and can be identified online. Traditional manual methods can only be used for regular inspections or manual intervention inspections when obvious faults occur in aircraft engine bearings, and their fault diagnosis has a lag. Traditional deep learning methods cannot perform high-precision fault diagnosis when working conditions change. The network trained based on domain generalization technology can detect operating signals in real time, greatly improving the working safety of aircraft engines.

[0044] Furthermore, the bearing operating data includes: bearing speed, bearing load and acceleration signal.

[0045] Furthermore, before using the training set to train the fault diagnosis model, the method further includes: labeling the training set to obtain the labeling category;

[0046] Among them, the annotation categories are: fault type and fault size.

[0047] Specifically, the use of domain generalization technology to diagnose faults in aircraft engine bearings eliminates the need to collect samples for each operating condition of the aircraft engine and perform labeling operations, saving a lot of manpower and material resources.

[0048] Further, the feature extraction module includes: a first feature extraction layer and a first classifier;

[0049] A first feature extraction layer, used to obtain multi-source domain features using a feature extractor;

[0050] The first classifier is used to adjust the multi-source domain features by using the improved cross entropy loss function to obtain the cross entropy loss.

[0051] Specifically, the feature extraction layer uses a one-dimensional convolutional neural network to extract features from multiple source domains, encodes the aircraft engine bearing data into feature representation, and the feature extractor is represented as f θ The classifier uses an improved cross entropy loss function to readjust the extracted features to make the intra-class distance closer and the inter-class distance farther, obtain discriminative features, and convert the features into prediction scores, and select the highest score as the prediction label. The classifier is represented by g θ , calculate the task-specific cross entropy loss L between the prediction and the true label CCE Using conditional contrast loss L cc Maximize the intra-class similarity and inter-class separability between different domains. At the same time, optimize L CCE With L cc To train the feature extractor.

[0052] Among them, the one-dimensional convolutional neural network contains five layers of convolutional modules. The first four layers of convolutional modules include convolutional layer, pooling layer, and weight normalization layer. The last two layers are fully connected layers, which are represented by L1 and L2.

[0053] The improved cross entropy loss function is NormFace, and its formula is shown in (1):

[0054]

[0055] Among them, L2 is the feature of the last fully connected layer, s is the scaling factor, cos(θ yi ) is the angle between the feature and the class it belongs to. Compared with the traditional Softmax loss, this loss function can significantly reduce the intra-class distance and expand the inter-class distance, so that there is a significant margin between different health states, which is conducive to the recognition of domain generalization. The final improved cross entropy loss function is shown in formula (2):

[0056]

[0057] Among them, L CCE is the cross entropy loss, is the jth sample obtained from the i-th source domain, For its corresponding label, is the output of the sample observation after the feature extraction layer and the classification layer, L2 is the feature of the last fully connected layer, M is the number of source domains, and N is i is the number of samples in the i-th source domain.

[0058] Further, the condition comparison module includes: a second feature extraction layer and a second classifier;

[0059] The second feature extraction layer is used to obtain sample features using a feature extractor;

[0060] The second classifier is used to adjust the sample features using the conditional contrast function to obtain the conditional contrast loss.

[0061] Specifically, samples with the same health status label in different source domains are set as positive samples, and other samples are set as negative samples. The purpose of this model is to maximize the mutual information between positive samples and minimize the similarity between negative samples.

[0062] The advantage of this setting is that it avoids the problem that even if the feature distribution differences are well aligned, samples belonging to different classes are too close, resulting in poor performance of the domain generalization network. The proposed method only pulls samples belonging to the same health state together, allowing the model to find class-conditional invariant representations in multiple source domains. Its loss function is defined as formula (3):

[0063]

[0064] Among them, |pos(u)| is the number of positive samples, σ(a,b)=(aT b / |a||b|) is the similarity function of given vectors a and b, and τ is the temperature function used to scale the comparison results. Among them, g θ ⊙f θ represents the combination of feature extractor and classifier, Represents the anchor sample x u The contrast loss of all samples is calculated as follows:

[0065]

[0066] in, is the conditional contrast loss. is the specific loss function for a single sample u, M is the number of source domains, N i is the number of samples in the i-th source domain. By minimizing The lower bound of the mutual information between positive samples can be maximized, where the theoretical formula of mutual information can be described by formula (5):

[0067]

[0068] Among them, p(a,b) is the joint distribution of (a,b), and p(a)p(b) is the product of marginal distributions, which can be deduced as follows:

[0069]

[0070] The Xavier initialization method is used to initialize the normal distribution to avoid the problem of gradient vanishing or gradient exploding during the training process. At the same time, the signal can be transmitted deeper in the neural network, and the value remains in a reasonable range after passing through multiple layers of neurons. As shown in formula (7):

[0071]

[0072] Furthermore, updating the feature extraction module and the condition comparison module includes:

[0073] Based on the cross entropy loss and conditional contrast loss and the SGD optimization algorithm, the feature extractor and the classifier are updated to obtain the updated feature extraction layer and classifier.

[0074] Specifically, as mentioned above, the overall optimization goal of the model is to and Joint optimization is performed to improve the generalization diagnostic performance of the model on unknown operating condition data. Therefore, the overall optimization goal is as follows:

[0075]

[0076] Furthermore, the updated feature extraction layer and classifier are:

[0077]

[0078] Among them, f θ is the special diagnosis extraction layer, ε is the learning rate, L cc is the conditional contrast loss, α is the weight balancing factor, and L CCE is the cross entropy loss, g θ is the classifier, f θ ← is the updated feature extraction layer, g θ ← is the updated classifier.

[0079] Input the operating data under the new working condition into the above-mentioned trained model, adjust the model to the test mode, set the gradient update to 0, and judge the prediction result of the model.

[0080] The innovations obtained through this embodiment include:

[0081] (1) Traditional manual identification methods require a lot of prior knowledge and manpower, and identification usually has a lag. Using traditional deep learning methods, it is necessary to collect operating data under all working conditions and label them. Although it does not require prior knowledge and a lot of manpower, the cost of collecting labels and operating data is still expensive. Using domain generalization technology to diagnose aircraft engine bearing faults does not require separate sample collection and labeling operations for each operating condition of the aircraft engine, saving a lot of manpower and material resources.

[0082] (2) The trained model can be applied to many new working conditions and can be identified online. Traditional manual methods can only be used for periodic inspections or manual intervention inspections when obvious faults occur in aircraft engine bearings, and their fault diagnosis is lagging. Traditional deep learning methods cannot perform high-precision fault diagnosis when working conditions change. The network trained based on domain generalization technology can detect operating signals in real time, greatly improving the working safety of aircraft engines.

[0083] (3) Using the improved softmax loss function, NormFace adjusts the classification features, ensuring that there is a certain margin between the classification boundaries, which clearly promotes the compact representation of features and is conducive to improving the domain generalization performance.

[0084] (4) Introducing the health status conditional contrast loss to measure the similarity between classes and avoiding the use of maximum mean difference, deep correlation contrast or adversarial methods for feature difference contrast can ensure that similar category features between different domains are closer, further improving the generalization performance of aircraft engine bearing fault diagnosis.

[0085] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for fault diagnosis of aircraft engine bearings based on domain generalization, characterized in that: include: Obtain the bearing data to be diagnosed; Inputting the bearing data to be diagnosed into a fault diagnosis model to obtain a fault diagnosis result; Among them, the fault diagnosis model includes a feature extraction module and a condition comparison feature module. The feature extraction module and the condition comparison module are trained respectively using a training set. In combination with a corresponding loss function, the feature extraction module and the condition comparison module are updated to obtain the fault diagnosis model. The training set includes bearing working data.

2. The method for diagnosing aircraft engine bearing faults based on domain generalization according to claim 1, characterized in that: The bearing operating data includes: bearing speed, bearing load and acceleration signal.

3. The method for diagnosing aircraft engine bearing faults based on domain generalization according to claim 2, characterized in that: Before using the training set to train the fault diagnosis model, the method further includes: labeling the training set to obtain a labeling category; The marking categories are: fault type and fault size.

4. The method for diagnosing aircraft engine bearing faults based on domain generalization according to claim 1, characterized in that: The feature extraction module includes: a first feature extraction layer and a first classifier; The first feature extraction layer is used to obtain multi-source domain features using a feature extractor; The first classifier is used to adjust the multi-source domain features by using an improved cross entropy loss function to obtain a cross entropy loss.

5. The method for diagnosing aircraft engine bearing faults based on domain generalization according to claim 4, characterized in that: The improved cross entropy loss function is: Among them, L CCE is the cross entropy loss function in cross-domain fault diagnosis, is the jth sample obtained from the i-th source domain, For its corresponding label, is the output of the sample observation after the feature extraction layer and the classification layer, L2 is the feature of the last fully connected layer, M is the number of different source domains, representing M source domains, N i is the number of samples in the i-th source domain.

6. The method for diagnosing aircraft engine bearing faults based on domain generalization according to claim 1, characterized in that: The condition comparison module includes: a second feature extraction layer and a second classifier; The second feature extraction layer is used to obtain sample features using a feature extractor; The second classifier is used to adjust the sample features by using a conditional contrast function to obtain a conditional contrast loss.

7. The method for diagnosing aircraft engine bearing faults based on domain generalization according to claim 6, characterized in that: The conditional comparison function is: in, is the conditional contrast loss, is the specific loss function for a single sample u, M is the number of source domains, N i is the number of samples in the i-th source domain.

8. The method for diagnosing aircraft engine bearing faults based on domain generalization according to claim 1, characterized in that: Updating the feature extraction module and the condition comparison module includes: Based on the cross entropy loss and conditional contrast loss and the SGD optimization algorithm, the feature extractor and the classifier are updated to obtain the updated feature extraction layer and classifier.

9. The method for diagnosing aircraft engine bearing faults based on domain generalization according to claim 8, characterized in that: The updated feature extraction layer and classifier are: Among them, f θ is the special diagnosis extraction layer, ε is the learning rate, L cc is the conditional contrast loss, α is the weight balancing factor, and L CCE is the cross entropy loss, g θ is the classifier, f θ ← is the updated feature extraction layer, g θ ← is the updated classifier.

Images