Method, apparatus, edge device and storage medium for monitoring AGV status

Abstract

The present invention relates to a method, an apparatus, a computing device and a storage medium for monitoring the state of an AGV. The method comprises: a first signal acquisition step: acquiring first signal data from an AGV at a first sampling rate, wherein the first signal data at least comprises a vibration signal; a feature extraction step: extracting features from the first signal data; an abnormal state judgment step: determining whether the AGV is in an abnormal state based on the extracted features, and executing a storage step if the AGV is determined to be in a normal state, and executing a second signal acquisition step if the AGV is determined to be in an abnormal state; a second signal acquisition step: acquiring second signal data from the AGV at a second sampling rate, wherein the second sampling rate is greater than the first sampling rate, and the type of the second signal data is equal to or greater than the type of the first signal data; and a storage step: storing the acquired first signal data or second signal data.

Description

Technical Field

[0001] The present disclosure generally relates to the field of factory digitalization technology, and more specifically, to a method, apparatus, computing device, and storage medium for monitoring AGV status. Background Art

[0002] At present, with the trend of flexible production, AGV (Automatic Guided Vehicle) is widely used. Almost all AGVs have a control system, and if an electronic failure occurs, it is easy to get the error code. But most AGVs cannot diagnose mechanical failures by themselves. According to the requirements of equipment safety and reliability, the health status of AGV needs to be monitored and analyzed in real time. Some commercial solutions are to set up an edge computing device on the AGV to collect relevant data and analyze it. Due to the limitations of edge device performance, and in order to ensure real-time diagnosis, the data sampling rate is low, and some simple algorithms are more suitable. Therefore, it is difficult to get accurate fault location.

[0003] In the current common AGV edge computing solutions, due to the limitations of edge devices, raw data is usually collected at a lower frequency, and useful features are extracted through knowledge of AGV such as root mean square and kurtosis; then some simple algorithms are used to process the features, such as threshold detection or linear regression; finally, the features and algorithm results are stored or published to a third-party system.

[0004] Take the assembly AGV in the automobile production line as an example. Its task is to transport automobile parts and at the same time become an assembly workstation. In this scenario, there are the following problems:

[0005] 1. There are many valuable data sources in AGV, such as temperature, vibration, moving path, etc. If all the data is processed and stored, it will be a huge challenge for edge devices.

[0006] 2. It is difficult to obtain accurate results using insufficient data and simple algorithms. If some complex algorithms, such as deep learning, are used, a large amount of data storage and better edge device performance are required, which will greatly increase costs. Summary of the invention

[0007] A brief overview of the present invention is given below in order to provide a basic understanding of certain aspects of the present invention. It should be understood that this overview is not an exhaustive overview of the present invention. It is not intended to identify key or important parts of the present invention, nor is it intended to limit the scope of the present invention. Its purpose is merely to give certain concepts in a simplified form as a prelude to a more detailed description discussed later.

[0008] In view of this, the present invention proposes a method for monitoring the status of AGV, in which an abnormal trigger mechanism is introduced to solve the cost and performance problems in the prior art. When the AGV is in a normal state, the trigger mechanism is in an inactive state, and only a small amount of conventional data needs to be collected and stored, and the edge device resources consumed are also very small; when the AGV fails, the abnormal trigger mechanism will be activated, in which case as much information as possible will be collected from the AGV. Using complex algorithms to analyze the collected information for fault diagnosis can obtain more accurate fault diagnosis results.

[0009] According to one aspect of the present disclosure, a method for monitoring the status of an AGV is provided, comprising: a first signal acquisition step: acquiring first signal data from the AGV at a first sampling rate, the first signal data including at least a vibration signal; a feature extraction step: extracting features from the first signal data; an abnormal state judgment step: determining whether the AGV is in an abnormal state based on the extracted features, and executing a storage step if the AGV is determined to be in a normal state, and executing a second signal acquisition step if the AGV is determined to be in an abnormal state; a second signal acquisition step: acquiring second signal data from the AGV at a second sampling rate, wherein the second sampling rate is greater than the first sampling rate, and the types of the second signal data are equal to or more than the types of the first signal data; and a storage step: storing the acquired first signal data or second signal data.

[0010] In this way, the efficiency of data collection and data storage can be improved.

[0011] Optionally, in an example of the above aspect, the first signal data also includes at least one of a rotation speed signal, a current signal, a temperature signal and a position signal.

[0012] In this way, not only the vibration signal can be analyzed, but also other signals can be selected for analysis as needed.

[0013] Optionally, in an example of the above aspect, the feature includes at least one of the following: a time domain feature, a frequency domain feature, and a time-frequency domain feature.

[0014] In this way, the frequency domain features can be used to determine the fault location, and the time domain features can be used to determine the severity of the fault.

[0015] Optionally, in an example of the above aspect, after the second signal acquisition step, the method further includes: a fault diagnosis step: analyzing the second signal data using a pre-stored machine learning model to perform fault diagnosis on the AGV, wherein the storage step further includes: storing the second signal data and a fault diagnosis result obtained based on the second signal data.

[0016] In this way, more accurate fault diagnosis of the AGV can be performed by analyzing the collected second signal data.

[0017] Optionally, in an example of the above aspect, the machine learning model is learned using a machine learning algorithm using pre-collected AGV signal history data and fault diagnosis history data as training samples.

[0018] In this way, a large amount of historical data can be used to generate a machine learning model, so that the collected second signal data can be processed in a complex manner.

[0019] Optionally, in an example of the above aspect, the machine learning algorithm is a deep neural network algorithm.

[0020] Optionally, in an example of the above aspect, the stored second signal data and the fault diagnosis result are used as new training samples to update the machine learning model.

[0021] In this way, the accuracy of the machine learning model can be improved, further improving the accuracy of fault diagnosis.

[0022] According to another aspect of the present disclosure, a device for monitoring the status of an AGV is provided, which is installed on the AGV and includes: a first signal acquisition unit, configured to acquire first signal data from the AGV at a first sampling rate, wherein the first signal data includes at least a vibration signal; a feature extraction unit, configured to extract features from the first signal data; an abnormal state judgment unit, configured to determine whether the AGV is in an abnormal state based on the extracted features; a second signal acquisition unit, configured to acquire a second signal from the AGV at a second sampling rate, wherein the second sampling rate is greater than the first sampling rate, and the type of the second signal is equal to or more than the type of the first signal; and a storage unit, configured to store the acquired first signal data or second signal data.

[0023] Optionally, in an example of the above aspect, the first signal data also includes at least one of a rotation speed signal, a current signal, a temperature signal and a position signal.

[0024] Optionally, in an example of the above aspect, the feature includes at least one of the following: a time domain feature, a frequency domain feature, and a time-frequency domain feature.

[0025] Optionally, in an example of the above aspect, the device for monitoring the status of the AGV also includes: a fault diagnosis unit, configured to analyze the second signal data using a pre-stored machine learning model to perform fault diagnosis on the AGV, wherein the storage unit is further configured to: store the second signal data and a fault diagnosis result obtained based on the second signal data.

[0026] Optionally, in an example of the above aspect, the machine learning model is learned using a machine learning algorithm using pre-collected AGV signal history data and fault diagnosis history data as training samples.

[0027] Optionally, in an example of the above aspect, the machine learning algorithm is a deep neural network algorithm.

[0028] Optionally, in an example of the above aspect, the device for monitoring the status of an AGV further includes: an updating unit configured to update the machine learning model using the stored second signal data and the fault diagnosis result as new training samples.

[0029] According to another aspect of the present disclosure, a computing device is provided, including: at least one processor; and a memory coupled to the at least one processor, the memory being used to store instructions, which, when executed by the at least one processor, enable the processor to perform the method described above.

[0030] According to another aspect of the present disclosure, a non-transitory machine-readable storage medium is provided, which stores executable instructions. When the instructions are executed, the machine performs the method as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The above and other purposes, features and advantages of the present invention will be more easily understood by referring to the following description of the embodiments of the present invention in combination with the accompanying drawings. The components in the accompanying drawings are only for illustrating the principles of the present invention. In the accompanying drawings, the same or similar technical features or components will be represented by the same or similar reference numerals. In the accompanying drawings:

[0032] Figure 1 is a flow chart showing an exemplary process of a method 100 for monitoring the status of an AGV according to one embodiment of the present disclosure.

[0033] Figure 2 is a block diagram showing an exemplary configuration of an apparatus 200 for monitoring an AGV status according to an embodiment of the present disclosure.

[0034] Figure 3 A block diagram of an edge device 300 for AGV status monitoring according to an embodiment of the present disclosure is shown, wherein the reference numerals are as follows:

[0035] 100: Method steps S102, S104, S106, S108, S110, S112 for monitoring AGV status:

[0036] 200: Device for monitoring AGV status 202: First signal acquisition unit

[0037] 204: Feature extraction unit 206: Abnormal state determination unit

[0038] 208: second signal acquisition unit 210: storage unit

[0039] 212: Fault diagnosis unit 214: Update unit

[0040] 300: computing device 302: processor

[0041] 304: Memory DETAILED DESCRIPTION

[0042] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and is not a limitation of the scope of protection, applicability or examples set forth in the claims. The function and arrangement of the elements discussed can be changed without departing from the scope of protection of the present disclosure. Various examples can omit, replace or add various processes or components as needed. For example, the described method can be performed in an order different from the described order, and various steps can be added, omitted or combined. In addition, the features described relative to some examples can also be combined in other examples.

[0043] As used herein, the term "including" and its variations represent open terms, meaning "including but not limited to". The term "based on" means "based at least in part on". The terms "one embodiment" and "an embodiment" mean "at least one embodiment". The term "another embodiment" means "at least one other embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other definitions may be included below, whether explicit or implicit. Unless the context clearly indicates otherwise, the definition of a term is consistent throughout the specification.

[0044] The present disclosure proposes a method for monitoring the status of an AGV, in which an abnormal trigger mechanism is introduced to solve the cost and performance problems in the prior art. When the AGV is in a normal state, the trigger mechanism is in an inactive state, and only a small amount of conventional data needs to be collected and stored, and very few edge device resources are consumed; when the AGV fails, the abnormal trigger mechanism will be activated, in which case as much information as possible will be collected from the AGV. Using complex algorithms to analyze the collected information for fault diagnosis can obtain more accurate fault diagnosis results.

[0045] The method and device for monitoring the status of an AGV according to an embodiment of the present disclosure will be described below in conjunction with the accompanying drawings.

[0046] Figure 1 is a flow chart showing an exemplary process of a method 100 for monitoring the status of an AGV according to one embodiment of the present disclosure.

[0047] exist Figure 1 In the process, the first signal collection step S102 is first performed to collect first signal data from the AGV at a first sampling rate, wherein the first signal data at least includes a vibration signal.

[0048] When the AGV is running in a normal state, in order to avoid occupying too many resources, for conventional monitoring, the first signal data is collected from the AGV at a lower sampling rate. Since the mechanical failure of the AGV is more easily detected through vibration data, the collected first signal data at least includes the vibration signal.

[0049] In one embodiment, the first signal data may include other signals such as a rotation speed signal, a current signal, a temperature signal, and a position signal in addition to the vibration signal.

[0050] It is understandable that multiple data adapters, such as analog input adapters, Modbus adapters, and various other adapters, can be set in the edge device to collect data from different data sources, such as vibration signals, speed signals, current signals, temperature signals, position signals, etc. from PLC (Programmable Logic Controller). The edge device can manage these adapters in a unified manner, and the adapters of each protocol use standard interfaces, so that they can be easily integrated and expanded.

[0051] Preferably, a configuration file can be pre-set in the edge device, which stores the data sampling rate, the type of data source to be collected, the data address, etc., and then data collection can be performed according to the relevant parameters set in the configuration file.

[0052] Next, a feature extraction step S104 is performed to extract features from the first signal data.

[0053] Specifically, one or more of the time domain features, frequency domain features, and time-frequency domain features of the signal can be extracted. Among them, the time domain features can include RMS (root mean square), kurtosis and other features; the frequency domain features can include FFT-energy (Fourier transform energy), peak and other features; the time-frequency domain features can include STFT (Short time Fourier Transform, short time Fourier transform) and the like.

[0054] Typically, in vibration analysis, frequency domain features can be used to determine the fault location, and time domain features can be used to determine the severity of the fault. In theory, most mechanical faults can be detected by monitoring these features.

[0055] The flexibility of the method of the present invention can be improved by setting and modifying specific feature extraction methods and parameters in a configuration file.

[0056] Next, the abnormal state judgment step S106 is executed to determine whether the AGV is in an abnormal state based on the extracted features. If the AGV is determined to be in a normal state, the storage step in box S110 is directly executed, and if the AGV is determined to be in an abnormal state, the second signal collection step in box S108 is executed.

[0057] Specifically, a commonly used abnormal state detection method in the art can be used to determine whether the AGV is in an abnormal state based on the extracted features. The present invention does not limit the abnormal state detection method used. Those skilled in the art can understand the specific process of using the abnormal state detection method in the prior art to determine whether the AGV is in an abnormal state, which will not be repeated here.

[0058] In the second signal collection step S108, a second signal is collected from the AGV at a second sampling rate, wherein the second sampling rate is greater than the first sampling rate, and the types of the second signal are equal to or greater than the types of the first signal.

[0059] In the method of the present invention, when the AGV is in a normal state, first signal data is collected from the AGV at a first sampling rate, and this first sampling rate is usually low to avoid occupying too many resources; when it is determined that the AGV is in an abnormal state, the signal is collected at a second sampling rate, and the second sampling rate is a higher rate than the first sampling rate. In this way, a large amount of signal data can be collected, thereby performing fault diagnosis more accurately.

[0060] In addition, when the AGV is in a normal state, only the vibration signal can be collected to reduce the amount of signal data collected and stored; when it is determined that the AGV is in an abnormal state, in order to perform a more accurate fault diagnosis on the AGV, as many types of signals as possible are collected from the AGV. Therefore, the second signal data can include vibration signals, speed signals, current signals, temperature signals, position signals, etc.

[0061] It can be understood that the higher the sampling rate, the more types of signals collected, the larger the amount of data collected, and thus the more resources occupied. Those skilled in the art can balance the amount of data collected and the resources occupied as needed, and select appropriate sampling rates and signal types for the first signal data and the second signal data. As described above, the first signal data may include other signals such as speed signals, current signals, temperature signals, and position signals in addition to vibration signals.

[0062] In the present invention, a more complex algorithm may be used to analyze the second signal data collected when the AGV is in an abnormal state.

[0063] Specifically, after executing the second signal collection step S108, the fault diagnosis step S112 can be executed to analyze the collected second signal data using a machine learning model to perform fault diagnosis on the AGV.

[0064] The machine learning model uses the pre-collected AGV signal history data and fault diagnosis history data as training samples and is learned using a machine learning algorithm.

[0065] Preferably, a deep neural network algorithm can be used for machine learning. When using a deep neural network algorithm for machine learning, data obtained from a variety of different types of data sources and corresponding fault diagnosis results can be used as training samples, and a multi-layer autoencoder can be used to obtain effective features.

[0066] By analyzing the collected second signal data using the learned machine learning model, the specific details of the fault can be obtained, which can help maintenance personnel conduct on-site inspections.

[0067] After the operation of step S112 is performed, in step S110 , both the collected second signal data and the fault diagnosis result obtained based on the analysis of the second signal data may be stored.

[0068] In one example, an updating step S114 may be further performed to update the machine learning model using the stored second signal data and fault diagnosis results as new training samples. In this way, self-optimization of the machine learning model may be achieved, thereby further improving the accuracy of AGV fault diagnosis using the method according to the present invention.

[0069] Figure 2 is a block diagram showing an exemplary configuration of an apparatus 200 for monitoring an AGV status according to an embodiment of the present disclosure.

[0070] The device 200 for monitoring the status of the AGV is installed on the AGV, which is equivalent to an edge device to monitor the status of the AGV.

[0071] like Figure 2 As shown, the device 200 for monitoring the state of an AGV includes: a first signal acquisition unit 202 , a feature extraction unit 204 , an abnormal state judgment unit 206 , a second signal acquisition unit 208 and a storage unit 210 .

[0072] The first signal acquisition unit 202 is configured to acquire first signal data from the AGV at a first sampling rate, and the first signal data at least includes a vibration signal.

[0073] The feature extraction unit 204 is configured to extract features from the first signal data.

[0074] The abnormal state determination unit 206 is configured to determine whether the AGV is in an abnormal state based on the extracted features.

[0075] When it is determined that the AGV is in a normal state, the storage unit 210 performs a storage operation; when it is determined that the AGV is in an abnormal state, the operation of the second signal acquisition unit 208 is triggered.

[0076] The second signal acquisition unit 208 is configured to acquire a second signal from the AGV at a second sampling rate, wherein the second sampling rate is greater than the first sampling rate, and the types of the second signal are equal to or more than the types of the first signal.

[0077] The storage unit 210 is configured to store the collected first signal data or the second signal data.

[0078] The first signal data further includes at least one of a rotation speed signal, a current signal, a temperature signal and a position signal.

[0079] The features include at least one of the following: time domain features, frequency domain features and time-frequency domain features.

[0080] In one example, the device 200 for monitoring the status of an AGV further includes: a fault diagnosis unit 212, configured to analyze the second signal data using a pre-stored machine learning model to perform fault diagnosis on the AGV, wherein the storage unit 210 is further configured to store the second signal data and a fault diagnosis result obtained based on the second signal data.

[0081] The machine learning model is learned by using the pre-collected AGV signal history data and fault diagnosis history data as training samples using a machine learning algorithm.

[0082] Wherein, the machine learning algorithm is a deep neural network algorithm.

[0083] In one example, the device 200 for monitoring the state of an AGV further includes: an updating unit 214, which is configured to use the stored second signal data and the fault diagnosis result as new training samples to update the machine learning model. The updating unit 214 can provide the updated machine learning model to the fault diagnosis unit 212, so that the fault diagnosis unit 212 can use the updated machine learning model to analyze the second signal data and perform fault diagnosis on the AGV, which can further improve the accuracy of fault diagnosis for the AGV.

[0084] It should be noted that Figure 2 The structure of the device 200 for monitoring the status of an AGV and its components is merely exemplary, and those skilled in the art may modify the device 200 as needed. Figure 2 Modify the structural block diagram shown.

[0085] The details of the operation and functions of the various parts of the apparatus 200 for monitoring the status of a device may be described in conjunction with reference to Figure 1 The relevant parts of the embodiment of the method 100 for monitoring the status of an AGV described in the present disclosure are the same or similar and will not be described in detail here.

[0086] According to the method and apparatus of the present invention, data storage on edge devices can be optimized. When the AGV is operating normally, only a small amount of conventional data is collected, thereby avoiding occupying too many storage resources and realizing the maximum potential of the edge devices.

[0087] According to the method and device of the present invention, by using configuration files for configuration, data sources, features, algorithm libraries, etc. can be flexibly expanded, and a unified data interface is provided, so that global management can be conveniently performed and can be applied to other AGVs.

[0088] According to the method and device of the present invention, when it is determined that an AGV is abnormal, the data collection rate can be increased, the data collection categories can be increased, and the accuracy of fault diagnosis can be improved.

[0089] According to the method and device of the present invention, the machine learning model adopted can be updated with the collected data and analysis results to improve the accuracy of the machine learning model, thereby further improving the accuracy of fault diagnosis.

[0090] As above Figure 1 and Figure 2 , the embodiments of the method and device for monitoring the state of AGV according to the embodiments of the present disclosure are described. The above-mentioned device for monitoring the state of the equipment can be implemented by hardware, or by software or a combination of hardware and software.

[0091] Figure 3 A block diagram of an edge device 300 for performing AGV status monitoring according to an embodiment of the present disclosure is shown. According to one embodiment, the edge device 300 may include at least one processor 302, which executes at least one computer-readable instruction (i.e., the above-mentioned element implemented in software form) stored or encoded in a computer-readable storage medium (i.e., memory 304).

[0092] In one embodiment, computer executable instructions are stored in the memory 304, which, when executed, cause at least one processor 302 to perform the following actions: collect first signal data from the AGV at a first sampling rate, wherein the first signal data includes at least a vibration signal; extract features from the first signal data; determine whether the AGV is in an abnormal state based on the extracted features, and directly perform a storage action when it is determined that the AGV is in a normal state, and perform a second signal collection action when it is determined that the AGV is in an abnormal state; collect second signal data from the AGV at a second sampling rate, wherein the second sampling rate is greater than the first sampling rate, and the type of the second signal data is equal to or more than the type of the first signal data; store the collected first signal data or second signal data.

[0093] It should be understood that the computer executable instructions stored in the memory 304, when executed, cause the at least one processor 302 to perform the above combined operations in various embodiments of the present disclosure. Figure 1-2 Describes the various operations and functions.

[0094] According to one embodiment, a non-transitory machine-readable medium is provided. The non-transitory machine-readable medium may have machine-executable instructions (i.e., the above-mentioned elements implemented in the form of software), which, when executed by a machine, causes the machine to execute the above-mentioned combination of various embodiments of the present disclosure. Figure 1-2 Describes the various operations and functions.

[0095] The specific embodiments described above in conjunction with the accompanying drawings describe exemplary embodiments, but do not represent all embodiments that can be implemented or fall within the scope of protection of the claims. The term "exemplary" used throughout this specification means "used as an example, instance or illustration" and does not mean "preferred" or "having advantages" over other embodiments. For the purpose of providing an understanding of the described technology, the specific embodiments include specific details. However, these technologies can be implemented without these specific details. In some instances, in order to avoid making the concepts of the described embodiments difficult to understand, well-known structures and devices are shown in block diagram form.

[0096] The above description of the present disclosure is provided to enable any person of ordinary skill in the art to implement or use the present disclosure. Various modifications to the present disclosure will be apparent to those of ordinary skill in the art, and the general principles defined herein may be applied to other variations without departing from the scope of protection of the present disclosure. Therefore, the present disclosure is not limited to the examples and designs described herein, but is consistent with the widest range of principles and novel features disclosed herein.

[0097] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method (100) for monitoring an AGV status, comprising: A first signal collection step (S102): collecting first signal data from the AGV at a first sampling rate, wherein the first signal data is a vibration signal; Feature extraction step (S104): extracting features from the first signal data; Abnormal state judgment step (S106): based on the extracted features, determining whether the AGV is in an abnormal state, if it is determined that the AGV is in a normal state, executing a storage step, if it is determined that the AGV is in an abnormal state, executing a second signal acquisition step; A second signal collection step (S108): collecting second signal data from the AGV at a second sampling rate, wherein the second sampling rate is greater than the first sampling rate, and the types of the second signal data are more than the types of the first signal data; as well as Storing step (S110): storing the collected first signal data or second signal data; The method (100) is executed by an edge device installed on an AGV; wherein when it is determined that the AGV is in an abnormal state, the second signal data includes a vibration signal, a rotation speed signal, a current signal, a temperature signal and a position signal.

2. The method (100) of claim 1, wherein: The feature includes at least one of the following: a time domain feature, a frequency domain feature, and a time-frequency domain feature.

3. The method (100) according to claim 1 or 2, after the second signal acquisition step, the method further comprises: Fault diagnosis step (S112): Analyzing the second signal data using a pre-stored machine learning model to perform fault diagnosis on the AGV. Wherein, the storing step further includes: storing the second signal data and a fault diagnosis result obtained based on the second signal data.

4. The method (100) of claim 3, wherein: The machine learning model is learned using a machine learning algorithm using pre-collected AGV signal history data and fault diagnosis history data as training samples.

5. The method (100) of claim 4, wherein: The machine learning algorithm is a deep neural network algorithm.

6. The method (100) as claimed in claim 3 further includes an updating step (S114): using the stored second signal data and the fault diagnosis results as new training samples to update the machine learning model.

7. A device (200) for monitoring the status of an AGV, the device being installed on an edge device of the AGV, comprising: A first signal acquisition unit (202) is configured to acquire first signal data from the AGV at a first sampling rate, wherein the first signal data is a vibration signal; A feature extraction unit (204), configured to extract features from the first signal data; an abnormal state judgment unit (206), configured to determine whether the AGV is in an abnormal state based on the extracted features; A second signal acquisition unit (208) is configured to acquire a second signal from the AGV at a second sampling rate when it is determined that the AGV is in an abnormal state, wherein the second sampling rate is greater than the first sampling rate, and the types of the second signal are more than the types of the first signal; as well as A storage unit (210), configured to store the collected first signal data or second signal data; When it is determined that the AGV is in an abnormal state, the second signal data includes a vibration signal, a rotation speed signal, a current signal, a temperature signal and a position signal.

8. The device (200) of claim 7, wherein: The feature includes at least one of the following: a time domain feature, a frequency domain feature, and a time-frequency domain feature.

9. The device (200) according to claim 7 or 8, further comprising: A fault diagnosis unit (212) is configured to analyze the second signal data using a pre-stored machine learning model to perform fault diagnosis on the AGV, Wherein, the storage unit (210) is further configured to: store the second signal data and a fault diagnosis result obtained based on the second signal data.

10. The device (200) of claim 9, wherein: The machine learning model is learned using a machine learning algorithm using pre-collected AGV signal history data and fault diagnosis history data as training samples.

11. The device (200) of claim 10, wherein: The machine learning algorithm is a deep neural network algorithm.

12. The apparatus (200) of claim 9, further comprising: An updating unit (214) is configured to update the machine learning model using the stored second signal data and the fault diagnosis result as new training samples.

13. An edge device (300), comprising: at least one processor (302); as well as A memory (304) coupled to the at least one processor (302), the memory being used to store instructions, which, when executed by the at least one processor (302), causes the processor (302) to perform the method according to any one of claims 1 to 6.

14. A non-transitory machine-readable storage medium storing executable instructions, which, when executed, cause the machine to perform the method of any one of claims 1 to 6.

Images