Rolling bearing composite fault diagnosis method, device, equipment and storage medium

Abstract

The present invention discloses a rolling bearing composite fault diagnosis method, device, equipment and storage medium based on a generative adversarial network; the method comprises: collecting single fault vibration signals and composite fault vibration signals of a rolling bearing as a training set and a test set respectively; extracting time domain features and frequency domain features from each sample in the training set and the test set to generate real fault samples; establishing a rolling bearing composite fault diagnosis model based on a generative adversarial network; training the composite fault diagnosis model with real single fault samples, random noise and single fault labels to achieve Nash equilibrium; fixing the parameters of a discriminator module, and training a generator with randomly generated virtual composite fault labels and noise; generating false fault samples with the trained generator, and outputting the rolling bearing composite fault type with the false composite fault samples and the real composite fault samples; the present invention can accurately identify single-point and multi-point composite fault problems of rolling bearings under zero-sample cross-process conditions.

Description

Technical Field

[0001] The present invention relates to the field of rolling bearing fault diagnosis, and in particular to a rolling bearing composite fault diagnosis method, device, equipment and storage medium based on a generative adversarial network. Background Art

[0002] With the continuous progress of industrial production in automation, informatization, and intelligence, enterprises' requirements for equipment reliability have become increasingly stringent and sophisticated. To meet this demand, implementing predictive maintenance has become one of the effective strategies to improve equipment reliability. Predictive maintenance not only provides a forward-looking assessment of the health status of equipment, but also can take preventive measures before failure occurs, thereby ensuring the continuous and stable operation of equipment and significantly reducing the economic losses caused by equipment failure. This strategy helps to optimize equipment maintenance plans, improve production efficiency, and reduce downtime.

[0003] Deep learning models have achieved remarkable results in diagnosing single-fault rolling element bearings. These models demonstrate high accuracy and robustness in identifying and classifying single fault types. However, single-point or multi-point combined faults in rolling element bearings remain a challenging problem in fault diagnosis. While numerous researchers have conducted in-depth research on single-point or multi-point combined faults in rolling element bearings, most of these studies suffer from a problem: they treat combined rolling element bearing faults as a new type of fault distinct from single faults, ignoring the coupling relationship between combined and single faults.

[0004] Existing deep learning-based rolling bearing composite fault diagnosis models are mostly based on supervised or unsupervised learning, requiring a large number of labeled or unlabeled composite fault samples to train the model. However, in real-world scenarios, composite fault samples are often unavailable. Therefore, a zero-shot learning method for composite fault diagnosis of rolling bearings is urgently needed.

[0005] The information disclosed in this background technology section is only intended to enhance understanding of the overall background of the invention and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art. Summary of the Invention

[0006] The purpose of the present invention is to realize zero-sample identification of single-point and multi-point composite fault problems of rolling bearings across working conditions.

[0007] The present invention provides a rolling bearing composite fault diagnosis method based on a generative adversarial network, comprising the steps of:

[0008] S11, collecting single fault vibration signals and composite fault vibration signals of rolling bearings; using the single fault vibration signals as a training set and the composite fault vibration signals as a test set, wherein each sample in the training set and the test set includes a corresponding fault type label; the fault type label includes a single fault label and a composite fault label;

[0009] S12. Extracting time domain features and frequency domain features from each sample in the training set to generate a real single fault sample; extracting time domain features and frequency domain features from each sample in the test set to generate a real compound fault sample;

[0010] S13, establishing a rolling bearing composite fault diagnosis model based on a generative adversarial network; the rolling bearing composite fault diagnosis model based on a generative adversarial network includes a generator module and a discriminator module;

[0011] S14, training the rolling bearing composite fault diagnosis model using the real single fault samples, random noise, and the single fault labels to achieve Nash equilibrium;

[0012] S15, fixing the parameters of the discriminator module, and training the generator using randomly generated virtual composite fault labels and the random noise;

[0013] S16. Generate a false composite fault sample through the trained generator; compare the false composite fault sample with the real composite fault sample, and output the rolling bearing composite fault type.

[0014] Preferably, in the embodiment of the present invention,

[0015] The generator module takes random noise and the fault type label as input to generate a false fault sample corresponding to the fault type label;

[0016] The discriminator module includes: a true-false discriminator and multiple fault type discriminators; the discriminator module uses the real single fault sample, the false fault sample and the corresponding fault type label as input to discriminate the authenticity of the sample and the fault type of the sample.

[0017] Preferably, in an embodiment of the present invention, the fault type label includes:

[0018] The coupling relationship between the composite fault type and the single fault type is identified using a unique hot label.

[0019] Preferably, in the embodiment of the present invention,

[0020] The number of the fault type discriminators is three, which are used to identify the rolling element fault, inner ring fault and outer ring fault of the rolling bearing respectively.

[0021] Preferably, in an embodiment of the present invention, training the rolling bearing composite fault diagnosis model using the real single fault samples, random noise and the single fault labels includes:

[0022] False single fault samples are generated by the generator module, and the discriminator module is trained according to the false single fault samples and the real single fault samples under different working conditions.

[0023] Preferably, in an embodiment of the present invention, training the rolling bearing composite fault diagnosis model using the real single fault samples, random noise and the single fault labels includes:

[0024] Before extracting the fault features of the real single fault sample, the fault feature map output by the previous layer is spliced ​​with the corresponding fault type label.

[0025] Preferably, in the embodiment of the present invention,

[0026] The discriminator module includes the following discriminator module loss function:

[0027]

[0028] Where, L D is the loss function of the discriminator module; D D is the true and false sample discriminator; D B is the rolling element fault discriminator; D I is the inner ring fault discriminator; D O is the outer race fault discriminator; P G is the sample space of false samples; P data is the sample space of real samples; P x P G and P data Coupled sample space; y B Roller fault label for unique hot label; y I is the inner circle fault label; y O L is the outer ring fault label; R is the loss of true or false discrimination; L B is the rolling element fault discrimination loss function; L I is the inner ring fault discrimination loss function; L O is the outer ring fault discrimination loss function; L Coral is the correlation alignment loss function; is the sample space P G The sample input discriminator D in D The mean of the results obtained after

[0029] The generator module includes the following generator module loss function:

[0030]

[0031] Where, L G is the loss function of the generator module; L F is the true and false discrimination loss function.

[0032] In another aspect of the present invention, a rolling bearing composite fault diagnosis device based on a generative adversarial network is provided, comprising:

[0033] A vibration signal acquisition unit is configured to acquire single fault vibration signals and composite fault vibration signals of a rolling bearing; the single fault vibration signals are used as a training set, and the composite fault vibration signals are used as a test set, wherein each sample in the training set and the test set includes a corresponding fault type label; the fault type label includes a single fault label and a composite fault label;

[0034] A fault sample generating unit is configured to extract time domain features and frequency domain features from each sample in the training set to generate a real single fault sample; and extract time domain features and frequency domain features from each sample in the test set to generate a real compound fault sample;

[0035] A composite fault diagnosis model building unit is used to establish a rolling bearing composite fault diagnosis model based on a generative adversarial network; the rolling bearing composite fault diagnosis model based on a generative adversarial network includes a generator module and a discriminator module;

[0036] A composite fault diagnosis model training unit, configured to train the rolling bearing composite fault diagnosis model using the real single fault samples, random noise, and the single fault labels to achieve Nash equilibrium;

[0037] a generator module training unit, configured to fix the parameters of the discriminator module and train the generator using randomly generated virtual composite fault labels and the random noise;

[0038] The composite fault type output unit is used to generate a false composite fault sample through a trained generator; compare the false composite fault sample with the real composite fault sample, and output the composite fault type of the rolling bearing.

[0039] On the other hand, an embodiment of the present invention provides a composite fault diagnosis device for rolling bearings based on a generative adversarial network. The composite fault diagnosis device for rolling bearings based on a generative adversarial network includes a computer program stored on a medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer executes the methods described in the above aspects and achieves the same technical effects.

[0040] On the other hand, a storage medium is provided on which a computer program is stored. When the computer program is executed by a processor, the steps of the rolling bearing composite fault diagnosis method based on a generative adversarial network as described above are implemented.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] In the present invention, firstly, the single fault vibration signal and the composite fault vibration signal of the rolling bearing are collected; the single fault vibration signal is used as the training set, and the composite fault vibration signal is used as the test set, and each sample in the training set and the test set includes a corresponding fault type label; the coupling relationship between the single fault and the composite fault is reflected by the fault type label; the time domain features and frequency domain features of each sample in the training set are extracted to generate a real single fault sample; the time domain features and frequency domain features of each sample in the test set are extracted to generate a real composite fault sample; a rolling bearing composite fault diagnosis model based on a generative adversarial network is established; the real single fault samples, random noise and single fault samples are used to identify the composite faults. The rolling bearing composite fault diagnosis model is trained with fault labels to achieve Nash equilibrium, so that the discriminator can distinguish the authenticity of the input samples and the fault type to which they belong; the parameters of the discriminator module are fixed, and the generator is trained with randomly generated virtual composite fault labels and random noise; the generator is enabled to use random noise and fault labels to generate false fault samples, so that the false fault samples are as similar as possible to the real samples, and finally the rolling bearing composite fault type is output by comparing the false composite fault samples with the real composite fault samples; the present invention can accurately identify single-point and multi-point composite fault problems of rolling bearings under zero-sample cross-working conditions.

[0043] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and to implement it according to the contents of the specification, and to make the above and other purposes, technical features and advantages of the present invention easier to understand, one or more preferred embodiments are listed below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1 This is a step diagram of the rolling bearing composite fault diagnosis method based on the generative adversarial network described in the present invention;

[0046] Figure 2It is a flow chart of the rolling bearing composite fault diagnosis method based on generative adversarial network described in the present invention;

[0047] Figure 3 The time domain features and frequency domain features described in the present invention;

[0048] Figure 4 It is a schematic diagram of the structure of the generator module of the present invention;

[0049] Figure 5 Schematic diagram of the structure of the discriminator module in the present invention;

[0050] Figure 6 It is a structural schematic diagram of the rolling bearing composite fault diagnosis device based on the generative adversarial network described in the present invention;

[0051] Figure 7 It is a structural schematic diagram of the rolling bearing composite fault diagnosis equipment based on the generative adversarial network described in the present invention. DETAILED DESCRIPTION

[0052] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific embodiments.

[0053] Unless expressly stated otherwise, throughout the specification and claims, the term “comprise” or variations such as “include” or “comprising” will be understood to include the stated elements or components but not to exclude other elements or components.

[0054] In this document, the terms "first", "second", etc. are used to distinguish two different elements or parts, and are not used to limit specific positions or relative relationships. In other words, in some embodiments, the terms "first", "second", etc. can also be interchangeable with each other.

[0055] Example 1

[0056] In order to achieve zero-sample identification of single-point and multi-point composite faults of rolling bearings across working conditions, such as Figure 1 and Figure 2 As shown, in an embodiment of the present invention, a rolling bearing composite fault diagnosis method based on a generative adversarial network is provided, comprising the steps of:

[0057] S11, collecting single fault vibration signals and composite fault vibration signals of rolling bearings; using the single fault vibration signals as a training set and the composite fault vibration signals as a test set, wherein each sample in the training set and the test set includes a corresponding fault type label; the fault type label includes a single fault label and a composite fault label;

[0058] In the embodiment of the present invention, an acceleration sensor is used to collect vibration signals when a rolling bearing has a single fault or a compound fault, and a fault type label is marked on each sample in the training set and the test set.

[0059] Furthermore, in order to reflect the coupling relationship between the composite fault and single fault of the rolling bearing, in an embodiment of the present invention, the coupling relationship between the composite fault type and the single fault type is identified by a unique hot label; in actual application, the unique hot label of a specific fault type is represented as: rolling element fault label [1,0,0]; inner ring fault label [0,1,0]; outer ring fault label [0,0,1]; rolling element and inner ring composite fault label [1,1,0]; and other composite fault types can be deduced accordingly.

[0060] S12. Extracting time domain features and frequency domain features from each sample in the training set to generate a real single fault sample; extracting time domain features and frequency domain features from each sample in the test set to generate a real compound fault sample;

[0061] In the embodiment of the present invention, the time domain features and frequency domain features of all samples of the collected training set and test set are extracted to obtain the following Figure 3 The 27 time domain features and frequency domain features are shown.

[0062] S13, establishing a rolling bearing composite fault diagnosis model based on a generative adversarial network; the rolling bearing composite fault diagnosis model based on a generative adversarial network includes a generator module and a discriminator module;

[0063] In the embodiment of the present invention, the generator module takes random noise and a fault type label as input and generates a false fault sample corresponding to the fault type label;

[0064] The discriminator module includes: a true-false discriminator and multiple fault type discriminators; the discriminator module takes real single fault samples, false fault samples and corresponding fault type labels as input to discriminate the authenticity of samples and the fault type of samples.

[0065] Preferably, in the embodiment of the present invention, the number of fault type discriminators is three, which are respectively used to identify rolling element fault, inner ring fault and outer ring fault of the rolling bearing.

[0066] S14, training the rolling bearing composite fault diagnosis model using the real single fault samples, random noise, and the single fault labels to achieve Nash equilibrium;

[0067] In an embodiment of the present invention, a generator module is used to generate false single fault samples, and the false single fault samples and real single fault samples under different working conditions are used to train the discriminator module, so that the discriminator can distinguish the authenticity and fault type of the input samples.

[0068] During the generator training process, before each operation of extracting fault features of a real single-fault sample, the output feature map of the previous layer is concatenated with the corresponding fault type label. The fault type label is directly introduced into the feature extraction process, thereby improving the accuracy of the generator in generating specific fault features. During the training process of the generative adversarial network, in order to balance the adversarial nature between the generator and the discriminator, a strategy of multiple iterative optimization of the generator after each round of discriminator update is adopted, that is, the generator is updated five times for each discriminator update. At the same time, the Coral loss function is added to reduce the distribution differences between data under different working conditions. The Coral loss function is:

[0069]

[0070] Where C1 is the feature covariance matrix of the first source domain samples, C2 is the feature covariance matrix of the second source domain samples, d is the length of the input feature, ||·|| F is the Frobenius norm.

[0071] The discriminator module includes the following discriminator module loss function:

[0072]

[0073] Where, L D is the loss function of the discriminator module; D D is the true and false sample discriminator; D B is the rolling element fault discriminator; D I is the inner ring fault discriminator; D O is the outer race fault discriminator; P G is the sample space of false samples; P data is the sample space of real samples; P x P G and P data Coupled sample space; y B Roller fault label for unique hot label; y I is the inner circle fault label; y O L is the outer ring fault label; R is the loss of true or false discrimination; L B is the rolling element fault discrimination loss function; L I is the inner ring fault discrimination loss function; L O is the outer ring fault discrimination loss function; L Coral is the correlation alignment loss function; is the sample space P G The sample input discriminator D in D The mean of the results obtained after

[0074] Generator module, including the following generator module loss function:

[0075]

[0076] Where, L G is the loss function of the generator module; L F is the true and false discrimination loss function.

[0077] Furthermore, in order to solve the class imbalance problem involved in the training process, the implementation of the present invention introduces the CB (Class-Balanced) loss function, that is, by adding a weight to multiply the original loss function, the corresponding class imbalance loss function is:

[0078]

[0079] Where CB is the class balance loss function; β is a hyperparameter, β∈[0,1); n y is the number of categories y in the training set; L is the loss function for distinguishing each single fault.

[0080] S15, fixing the parameters of the discriminator module, and training the generator using randomly generated virtual composite fault labels and the random noise;

[0081] In an embodiment of the present invention, the parameters of the discriminator module trained in step S14 are frozen, and the generator module is further trained by randomly generated virtual composite fault labels and random noise, so that the generator can generate false composite fault samples; in this process, the loss function of the generator module is defined as the generator module loss function in step S14.

[0082] S16. Generate a false composite fault sample through the trained generator; compare the false composite fault sample with the real composite fault sample, and output the rolling bearing composite fault type.

[0083] In the embodiment of the present invention, randomly generated virtual composite fault labels and random noise are input into a trained generator module, and false composite fault samples are generated by the trained generator.

[0084] Subsequently, the T-SNE dimensionality reduction algorithm is used to reduce the dimensionality of the real compound fault samples and the generated false compound fault samples in the test set; in the space after dimensionality reduction, the mean Euclidean distance between the generated false compound fault samples and the compound fault types in each real compound fault sample is calculated to identify the type of compound fault.

[0085] In summary, the embodiment of the present invention first collects the single fault vibration signal and the composite fault vibration signal of the rolling bearing; uses the single fault vibration signal as the training set and the composite fault vibration signal as the test set, and each sample in the training set and the test set includes a corresponding fault type label; the coupling relationship between the single fault and the composite fault is reflected by the fault type label; extracts the time domain features and frequency domain features of each sample in the training set to generate a real single fault sample; extracts the time domain features and frequency domain features of each sample in the test set to generate a real composite fault sample; establishes a rolling bearing composite fault diagnosis model based on a generative adversarial network; and uses real single fault samples, random noise The rolling bearing composite fault diagnosis model is trained with single fault labels to achieve Nash equilibrium, so that the discriminator can distinguish the authenticity of input samples and the fault type to which they belong; the parameters of the discriminator module are fixed, and the generator is trained with randomly generated virtual composite fault labels and random noise; the generator is enabled to generate false fault samples using random noise and fault labels, so that the false fault samples are as similar as possible to the real samples; finally, the false composite fault samples are compared with the real composite fault samples to output the rolling bearing composite fault type; the present invention can accurately identify single-point and multi-point composite fault problems of rolling bearings under zero-sample cross-working conditions.

[0086] Example 2

[0087] Corresponding to the method embodiment, another aspect of the embodiment of the present invention further provides a rolling bearing composite fault diagnosis device based on a generative adversarial network. Figure 6 The schematic diagram of the structure of the rolling bearing composite fault diagnosis device based on the generative adversarial network provided by the embodiment of the present invention is shown. The rolling bearing composite fault diagnosis device based on the generative adversarial network is Figure 1 The device corresponding to the rolling bearing composite fault diagnosis method based on generative adversarial network described in the corresponding embodiment, that is, is realized by means of a virtual device Figure 1 In the corresponding embodiment of the rolling bearing composite fault diagnosis method based on a generative adversarial network, each virtual module constituting the rolling bearing composite fault diagnosis device based on a generative adversarial network can be executed by an electronic device, such as a network device, a terminal device, or a server. Specifically, the rolling bearing composite fault diagnosis device based on a generative adversarial network in the embodiment of the present invention includes:

[0088] The vibration signal acquisition unit 01 is used to acquire single fault vibration signals and composite fault vibration signals of rolling bearings; the single fault vibration signals are used as a training set, and the composite fault vibration signals are used as a test set, wherein each sample in the training set and the test set includes a corresponding fault type label; the fault type label includes a single fault label and a composite fault label;

[0089] In the embodiment of the present invention, an acceleration sensor is used to collect vibration signals when a rolling bearing has a single fault or a compound fault, and a fault type label is marked on each sample in the training set and the test set.

[0090] In order to reflect the coupling relationship between the composite fault and single fault of the rolling bearing, in the embodiment of the present invention, the coupling relationship between the composite fault type and the single fault type is identified by a unique hot label. In practical applications, the unique hot labels of specific fault types are represented as follows: rolling element fault label [1,0,0]; inner ring fault label [0,1,0]; outer ring fault label [0,0,1]; rolling element and inner ring composite fault label [1,1,0]; and other composite fault types can be deduced in the same way.

[0091] Fault sample generating unit 02, configured to extract time domain features and frequency domain features from each sample in the training set to generate a real single fault sample; and extract time domain features and frequency domain features from each sample in the test set to generate a real composite fault sample;

[0092] In the embodiment of the present invention, the time domain features and frequency domain features of all samples of the collected training set and test set are extracted to obtain the following Figure 3 The 27 time domain features and frequency domain features are shown.

[0093] The composite fault diagnosis model construction unit 03 is used to establish a composite fault diagnosis model for rolling bearings based on a generative adversarial network; the composite fault diagnosis model for rolling bearings based on a generative adversarial network includes a generator module and a discriminator module;

[0094] In the embodiment of the present invention, the generator module takes random noise and a fault type label as input and generates a false fault sample corresponding to the fault type label;

[0095] The discriminator module includes: a true-false discriminator and multiple fault type discriminators; the discriminator module takes real single fault samples, false fault samples and corresponding fault type labels as input to discriminate the authenticity of samples and the fault type of samples.

[0096] Preferably, in the embodiment of the present invention, the number of fault type discriminators is three, which are respectively used to identify rolling element fault, inner ring fault and outer ring fault of the rolling bearing.

[0097] The composite fault diagnosis model training unit 04 is used to train the rolling bearing composite fault diagnosis model using the real single fault samples, random noise and the single fault labels to achieve Nash equilibrium;

[0098] In an embodiment of the present invention, a generator module is used to generate false single fault samples, and the false single fault samples and real single fault samples under different working conditions are used to train the discriminator module, so that the discriminator can distinguish the authenticity and fault type of the input samples.

[0099] During the generator training process, before each operation of extracting the fault features of a real single fault sample, the output feature map of the previous layer is spliced ​​with the corresponding fault type label, and the fault type label is directly introduced into the feature extraction process, thereby improving the accuracy of the generator in generating specific fault features; during the training process of the generative adversarial network, in order to balance the adversarial nature between the generator and the discriminator, a strategy of iteratively optimizing the generator multiple times after each round of discriminator update is adopted, that is, the generator is updated five times for each discriminator update.

[0100] A generator module training unit 05 is used to fix the parameters of the discriminator module and train the generator using the randomly generated virtual composite fault label and the random noise;

[0101] In an embodiment of the present invention, the parameters of the discriminator module trained in step S14 are frozen, and the generator module is further trained by randomly generated virtual composite fault labels and random noise so that the generator can generate false composite fault samples; in this process, the loss function of the generator module is defined as the generator module loss function in step S14.

[0102] The composite fault type output unit 06 is used to generate a false composite fault sample through a trained generator; compare the false composite fault sample with the real composite fault sample, and output the composite fault type of the rolling bearing.

[0103] In the embodiment of the present invention, randomly generated virtual composite fault labels and random noise are input into a trained generator module, and false composite fault samples are generated by the trained generator.

[0104] The T-SNE dimensionality reduction algorithm is used to reduce the dimensionality of the real compound fault samples and the generated false compound fault samples in the test set. In the space after dimensionality reduction, the mean Euclidean distance between the generated false compound fault samples and the compound fault types in each real compound fault sample is calculated to identify the type of compound fault.

[0105] It should be noted that the specific implementation and technical effects of the rolling bearing composite fault diagnosis device based on the generative adversarial network in the embodiment of the present invention can be referred to Figure 1 The corresponding rolling bearing composite fault diagnosis method based on generative adversarial network will not be described here.

[0106] Example 3

[0107] Corresponding to the method embodiments, embodiments of the present invention also provide a rolling bearing composite fault diagnosis device based on a generative adversarial network, such as a terminal or server. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be, but is not limited to, a smartphone, tablet computer, laptop computer, or desktop computer.

[0108] An example diagram of the hardware structure block diagram of the rolling bearing composite fault diagnosis device based on the generative adversarial network provided by the embodiment of the present invention is shown in FIG. Figure 7 As shown, this may include:

[0109] Processor 1, communication interface 2, memory 3 and communication bus 4;

[0110] The processor 1, the communication interface 2, and the memory 3 communicate with each other via the communication bus 4;

[0111] Optionally, the communication interface 2 may be an interface of a communication module, such as an interface of a GSM module;

[0112] The processor 1 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0113] The memory 3 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.

[0114] The processor 1 is specifically configured to execute the computer program stored in the memory 3 to perform the following steps:

[0115] S11, collecting single fault vibration signals and composite fault vibration signals of rolling bearings; using the single fault vibration signals as a training set and the composite fault vibration signals as a test set, wherein each sample in the training set and the test set includes a corresponding fault type label; the fault type label includes a single fault label and a composite fault label;

[0116] S12. Extracting time domain features and frequency domain features from each sample in the training set to generate a real single fault sample; extracting time domain features and frequency domain features from each sample in the test set to generate a real compound fault sample;

[0117] S13, establishing a rolling bearing composite fault diagnosis model based on a generative adversarial network; the rolling bearing composite fault diagnosis model based on a generative adversarial network includes a generator module and a discriminator module;

[0118] S14, training the rolling bearing composite fault diagnosis model using the real single fault samples, random noise, and the single fault labels to achieve Nash equilibrium;

[0119] S15, fixing the parameters of the discriminator module, and training the generator using the randomly generated virtual composite fault label and the random noise;

[0120] S16. Generate a false composite fault sample through the trained generator; compare the false composite fault sample with the real composite fault sample, and output the rolling bearing composite fault type.

[0121] The above-mentioned product can execute the method provided by the embodiment of the present invention and has the corresponding functional modules and beneficial effects of the execution method. For technical details not fully described in this embodiment, please refer to the rolling bearing composite fault diagnosis method based on generative adversarial network provided by the embodiment of the present invention.

[0122] Example 4

[0123] In an embodiment of the present invention, a storage medium is further provided. The storage medium may store a program suitable for execution by a processor, wherein the program is used to:

[0124] S11, collecting single fault vibration signals and composite fault vibration signals of rolling bearings; using the single fault vibration signals as a training set and the composite fault vibration signals as a test set, wherein each sample in the training set and the test set includes a corresponding fault type label; the fault type label includes a single fault label and a composite fault label;

[0125] S12. Extracting time domain features and frequency domain features from each sample in the training set to generate a real single fault sample; extracting time domain features and frequency domain features from each sample in the test set to generate a real compound fault sample;

[0126] S13, establishing a rolling bearing composite fault diagnosis model based on a generative adversarial network; the rolling bearing composite fault diagnosis model based on a generative adversarial network includes a generator module and a discriminator module;

[0127] S14, training the rolling bearing composite fault diagnosis model using the real single fault samples, random noise, and the single fault labels to achieve Nash equilibrium;

[0128] S15, fixing the parameters of the discriminator module, and training the generator using the randomly generated virtual composite fault label and the random noise;

[0129] S16. Generate a false composite fault sample through the trained generator; compare the false composite fault sample with the real composite fault sample, and output the rolling bearing composite fault type.

[0130] Optionally, the detailed functions and extended functions of the program may refer to the above description.

[0131] The above-mentioned product can execute the method provided by the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not fully described in this embodiment, please refer to the methods provided by other embodiments of the present invention.

[0132] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0133] In the several embodiments provided herein, it should be understood that the disclosed systems, devices, and methods may be implemented in other ways. Furthermore, the couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0134] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0135] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0136] It should be understood that in the embodiments of the present application, the various embodiments and features can be combined with each other to solve the aforementioned technical problems.

[0137] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0138] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A rolling bearing composite fault diagnosis method based on generative adversarial network, characterized in that: Including steps: S11, collecting single fault vibration signals and composite fault vibration signals of rolling bearings; using the single fault vibration signals as a training set and the composite fault vibration signals as a test set, wherein each sample in the training set and the test set includes a corresponding fault type label; the fault type label includes a single fault label and a composite fault label; S12. Extracting time domain features and frequency domain features from each sample in the training set to generate a real single fault sample; extracting time domain features and frequency domain features from each sample in the test set to generate a real compound fault sample; S13, establishing a rolling bearing composite fault diagnosis model based on a generative adversarial network; the rolling bearing composite fault diagnosis model based on a generative adversarial network includes a generator module and a discriminator module; S14, training the rolling bearing composite fault diagnosis model using the real single fault samples, random noise, and the single fault labels to achieve Nash equilibrium; S15, fixing the parameters of the discriminator module, and training the generator using randomly generated virtual composite fault labels and the random noise; S16. Generate a false composite fault sample through the trained generator; compare the false composite fault sample with the real composite fault sample, and output the rolling bearing composite fault type.

2. The rolling bearing composite fault diagnosis method based on generative adversarial network according to claim 1 is characterized in that: The generator module takes random noise and the fault type label as input to generate a false fault sample corresponding to the fault type label; The discriminator module includes: a true-false discriminator and multiple fault type discriminators; The discriminator module uses the real single fault sample, the false fault sample and the corresponding fault type label as input to discriminate the authenticity of the sample and the fault type of the sample.

3. The rolling bearing composite fault diagnosis method based on generative adversarial network according to claim 2 is characterized in that: The fault type label includes: The coupling relationship between the composite fault type and the single fault type is identified using a unique hot label.

4. The rolling bearing composite fault diagnosis method based on generative adversarial network according to claim 3 is characterized in that: The number of the fault type discriminators is three, which are used to identify the rolling element fault, inner ring fault and outer ring fault of the rolling bearing respectively.

5. The rolling bearing composite fault diagnosis method based on generative adversarial network according to claim 4 is characterized in that: The rolling bearing composite fault diagnosis model is trained using the real single fault sample, random noise, and the single fault label, including: False single fault samples are generated by the generator module, and the discriminator module is trained according to the false single fault samples and the real single fault samples under different working conditions.

6. The rolling bearing composite fault diagnosis method based on generative adversarial network according to claim 5 is characterized in that: The rolling bearing composite fault diagnosis model is trained using the real single fault sample, random noise, and the single fault label, including: Before extracting the fault features of the real single fault sample, the fault feature map output by the previous layer is spliced ​​with the corresponding fault type label.

7. The rolling bearing composite fault diagnosis method based on generative adversarial network according to claim 6 is characterized in that: The discriminator module includes the following discriminator module loss function: Where, L D is the loss function of the discriminator module; D D is the true and false sample discriminator; D B is the rolling element fault discriminator; D I is the inner ring fault discriminator; D O is the outer race fault discriminator; P G is the sample space of false samples; P data is the sample space of real samples; P x P G and P data Coupled sample space; y B Roller fault label for unique hot label; y I is the inner circle fault label; y O L is the outer ring fault label; R is the loss of true or false discrimination; L B is the rolling element fault discrimination loss function; L I is the inner ring fault discrimination loss function; L O is the outer ring fault discrimination loss function; L Coral is the correlation alignment loss function; is the sample space P G The sample input discriminator D in D The mean of the results obtained after The generator module includes the following generator module loss function: Where, L G is the loss function of the generator module; L F is the true and false discrimination loss function.

8. A rolling bearing composite fault diagnosis device based on generative adversarial network, characterized in that: include: A vibration signal acquisition unit is configured to acquire single fault vibration signals and composite fault vibration signals of a rolling bearing; the single fault vibration signals are used as a training set, and the composite fault vibration signals are used as a test set, wherein each sample in the training set and the test set includes a corresponding fault type label; the fault type label includes a single fault label and a composite fault label; A fault sample generating unit is configured to extract time domain features and frequency domain features from each sample in the training set to generate a real single fault sample; and extract time domain features and frequency domain features from each sample in the test set to generate a real compound fault sample; A composite fault diagnosis model building unit is used to establish a rolling bearing composite fault diagnosis model based on a generative adversarial network; the rolling bearing composite fault diagnosis model based on a generative adversarial network includes a generator module and a discriminator module; A composite fault diagnosis model training unit, configured to train the rolling bearing composite fault diagnosis model using the real single fault samples, random noise, and the single fault labels to achieve Nash equilibrium; a generator module training unit, configured to fix the parameters of the discriminator module and train the generator using randomly generated virtual composite fault labels and the random noise; The composite fault type output unit is used to generate a false composite fault sample through a trained generator; compare the false composite fault sample with the real composite fault sample, and output the composite fault type of the rolling bearing.

9. A rolling bearing composite fault diagnosis device based on a generative adversarial network, characterized in that: include: Memory for storing computer programs; A processor is used to call and execute the computer program to implement the steps of the rolling bearing composite fault diagnosis method based on a generative adversarial network as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The method comprises a software program, wherein the software program is suitable for a processor to execute the steps of the rolling bearing composite fault diagnosis method based on a generative adversarial network as claimed in any one of claims 1 to 7.

Images