Abnormal audio detection method and fault audio sample updating method

Abstract

The application provides an abnormal audio detection method and a fault audio sample updating method. The method comprises the following steps: obtaining to-be-detected audio data; in the case that it is determined that the to-be-detected audio data is abnormal, determining an initial fault type of the to-be-detected audio data; performing secondary detection on the to-be-detected audio data according to positive samples and / or negative samples, and determining a final fault type of the to-be-detected audio data according to a secondary detection result and the initial fault type. The method can improve the accuracy of fault type detection.

Description

Technical Field

[0001] This application relates to the field of data processing, and in particular to an abnormal audio detection method and a method for updating faulty audio samples. Background Technology

[0002] Abnormal sound detection technology, as an important branch of anomaly detection technology, has received widespread attention in recent years. Sound signals have a rich amount of information and have unique advantages in many situations where vision, touch, and smell are not applicable. Detecting the occurrence of abnormal sound signals can provide early warning signals for most accidents and disasters, enabling people to react in advance without seeing the scene.

[0003] For fault detection of various machines in industrial production, such as motors and belt rollers in cement plants and steel plants, analyzing the sounds emitted by these machines during operation can quickly detect whether the machine's operating status is abnormal. This allows for rapid, accurate, and flexible detection of potential machine faults, preventing machine malfunctions and eliminating tedious manual troubleshooting. This is of great significance for the automation and intelligent development of industrial production. Summary of the Invention

[0004] In view of this, this application provides an abnormal audio detection method and a method for updating faulty audio samples.

[0005] Specifically, this application is implemented through the following technical solution:

[0006] According to a first aspect of the embodiments of this application, an abnormal audio detection method is provided, comprising:

[0007] Acquire the audio data to be detected;

[0008] If it is determined that the audio data to be detected is abnormal, the initial fault type of the audio data to be detected shall be determined;

[0009] The audio data to be detected is subjected to secondary detection based on positive and / or negative samples, and the final fault type of the audio data to be detected is determined based on the secondary detection results and the initial fault type.

[0010] According to a second aspect of the embodiments of this application, a method for updating faulty audio samples is provided, comprising:

[0011] The first audio data is acquired using an embedded audio acquisition device;

[0012] When the first audio data is determined to be abnormal by the audio detection algorithm, the first audio data is input to the fault type detection module, wherein the fault type detection module is configured to be presented on the display in the form of a webpage. The webpage form of the fault type detection module includes the input sequence number of the sample audio data, the sample name of the sample audio data, the sample type of the sample audio data, the acquisition time of the sample audio data, and interactive tags that allow users to edit.

[0013] In response to the user's interaction with the first audio data, the sample name of the first audio data is updated so that the sample name of the first audio data includes the fault type input by the user, thereby updating the fault type detection module.

[0014] Acquire second audio data using an embedded audio acquisition device;

[0015] When the second audio data is determined to be abnormal by the audio detection algorithm, the second audio data is input to the updated fault type detection module;

[0016] If a sample audio data matching the second audio data is found in the updated fault type detection module, then the fault type is automatically displayed in the sample name of the second audio data based on the fault type of the matching sample audio data.

[0017] According to a third aspect of the embodiments of this application, an electronic device is provided, the electronic device comprising:

[0018] A processor and a machine-readable storage medium storing machine-executable instructions that can be executed by the processor; the processor is configured to execute the machine-executable instructions to implement the method provided in the first aspect.

[0019] The abnormal audio detection method of this application embodiment, when it is determined that the acquired audio data to be detected is abnormal and the initial fault type of the audio data to be detected is determined, performs a secondary detection on the audio data to be detected based on positive samples and / or negative samples, and determines the final fault type of the audio data to be detected based on the secondary detection results and the initial fault type. By pre-acquiring positive samples and / or negative samples, and performing secondary filtering detection on the audio data to be detected based on the acquired positive samples and / or negative samples, the correction of the initial fault type detection results is achieved, thereby improving the accuracy of fault type detection. Attached Figure Description

[0020] Figure 1 A flowchart illustrating an abnormal audio detection method provided in an embodiment of this application;

[0021] Figure 2 A flowchart illustrating a method for updating faulty audio samples provided in an embodiment of this application;

[0022] Figure 3 A flowchart illustrating a multi-channel abnormal audio detection method provided in an embodiment of this application;

[0023] Figure 4 A schematic diagram illustrating the process of an audio diagnostic device for determining whether audio data is abnormal, provided in an embodiment of this application;

[0024] Figure 5 A schematic diagram of a sample management operation interface provided in an embodiment of this application;

[0025] Figure 6 This application provides a sample adding operation interface in an embodiment of the present application;

[0026] Figures 7-8 A schematic diagram of the secondary filtering detection process provided in an embodiment of this application;

[0027] Figure 9 A schematic diagram illustrating an algorithm update process provided in an embodiment of this application;

[0028] Figure 10 A schematic diagram illustrating an algorithm update process provided in an embodiment of this application;

[0029] Figure 11 A flowchart illustrating an abnormal audio detection method provided in an embodiment of this application;

[0030] Figure 12 This is a schematic diagram of the structure of an abnormal audio detection device provided in an embodiment of this application;

[0031] Figure 13 This is a schematic diagram of another abnormal audio detection device provided in an embodiment of this application;

[0032] Figure 14 This is a schematic diagram of another abnormal audio detection device provided in an embodiment of this application;

[0033] Figure 15 This is a schematic diagram of another abnormal audio detection device provided in an embodiment of this application;

[0034] Figure 16 A schematic diagram of the structure of an update device for faulty audio samples provided in an embodiment of this application;

[0035] Figure 17 A schematic diagram of the structure of another fault audio sample updating device provided in an embodiment of this application;

[0036] Figure 18 An embodiment of this application provides a Figures 11-17 A schematic diagram of the hardware structure of any of the devices shown. Detailed Implementation

[0037] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0038] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0039] To enable those skilled in the art to better understand the technical solutions provided in the embodiments of this application, and to make the above-mentioned objectives, features and advantages of the embodiments of this application more apparent and understandable, the technical solutions in the embodiments of this application will be further described in detail below with reference to the accompanying drawings.

[0040] Please see Figure 1 This is a flowchart illustrating an abnormal audio detection method provided in an embodiment of this application. This abnormal audio detection method can be applied to audio diagnostic devices, such as... Figure 1 As shown, the abnormal audio detection method may include the following steps:

[0041] Step S100: Obtain the audio data to be detected.

[0042] In this embodiment of the application, the audio data to be detected can be audio data acquired at a specified location during the operation of the device to be detected.

[0043] For example, it can be audio data acquired at the head, tail, inner cavity, or outer wall of the device during its operation.

[0044] For example, audio data at different locations during the operation of the device under test can be acquired through multiple channels.

[0045] For example, audio data from the head, tail, inner cavity, and outer wall of the device under test can be acquired through four channels during the device's operation.

[0046] Step S110: If it is determined that there is an abnormality in the audio data to be detected, determine the initial fault type of the audio data to be detected.

[0047] In this embodiment of the application, it can be determined whether the acquired audio data to be detected is abnormal.

[0048] For example, audio anomaly diagnosis algorithms can be used to determine whether there are anomalies in the audio data to be detected.

[0049] For example, the above-mentioned audio anomaly diagnosis algorithm can employ a deep learning algorithm.

[0050] For example, the acquired audio data to be detected can be input into a trained audio anomaly diagnosis algorithm model to obtain an anomaly evaluation score for the audio data. This score is then compared with a threshold used for anomaly determination. If the anomaly evaluation score exceeds the preset threshold, the audio data is determined to be abnormal; if the score does not exceed the threshold, the audio data is determined not to be abnormal.

[0051] If it is determined that there is an anomaly in the audio data to be tested, a fault type detection can be performed on the audio data to determine the fault type of the audio data to be tested (which can be called the initial fault type).

[0052] For example, fault type detection algorithms can be used to detect fault types in the audio data to be tested.

[0053] For example, the above-mentioned fault type detection algorithm can employ a deep learning algorithm.

[0054] It should be noted that in this embodiment of the application, if it is determined that the acquired audio data to be detected is not abnormal, it can be determined to be normal audio data. In this case, it is not necessary to perform subsequent fault type detection.

[0055] Step S120: Perform secondary detection on the audio data to be detected based on positive and / or negative samples, and determine the final fault type of the audio data to be detected based on the secondary detection results and the initial fault type.

[0056] In this embodiment of the application, it is considered that fault type detection algorithms usually cannot guarantee high accuracy for all different scenarios. That is, for any fault type detection algorithm, its accuracy in detecting fault types in some scenarios may be relatively high, but its accuracy in detecting fault types in other scenarios may be relatively low.

[0057] Accordingly, in order to improve the accuracy of fault type detection of audio data, normal audio data samples (which can be called positive samples) and / or abnormal audio data samples (which can be called negative samples) under different scenarios can be obtained in advance, and the audio data to be detected can be subjected to secondary detection based on the obtained positive samples and / or negative samples to obtain secondary detection results.

[0058] For example, any sample (including positive or negative samples) may include a piece of audio data and a model file corresponding to the audio data (a model file obtained by extracting audio features from the audio data).

[0059] For example, the secondary detection result may include the audio data to be detected being normal audio data, the fault type of the audio data to be detected, or the fault type of the audio data to be detected being undetermined (i.e., the audio data to be detected is determined to be abnormal audio data, but the fault type of the audio data to be detected is not determined).

[0060] In this embodiment of the application, the final fault type of the audio data to be detected can be determined based on the secondary detection results and the initial fault type.

[0061] For example, if the secondary detection result is normal audio data or a fault type of the audio data to be detected, the secondary detection result can be taken as the final fault type of the audio data to be detected.

[0062] If the secondary test results show that the fault type of the audio data to be tested cannot be determined, the initial fault type can be taken as the final fault type of the audio data to be tested.

[0063] It can be seen that, in Figure 1 In the method flow shown, for the acquired audio data to be detected, if it is determined that the audio data to be detected is abnormal and the initial fault type of the audio data to be detected is determined, the audio data to be detected is subjected to secondary detection based on positive samples and / or negative samples. Based on the secondary detection results and the initial fault type, the final fault type of the audio data to be detected is determined. By pre-acquiring positive samples and / or negative samples, and performing secondary filtering detection on the audio data to be detected based on the acquired positive samples and / or negative samples, the correction of the initial fault type detection results is achieved, thereby improving the accuracy of fault type detection.

[0064] In some embodiments, step S120, performing secondary detection on the audio data to be detected based on positive and / or negative samples, and determining the final fault type of the audio data to be detected based on the secondary detection results and the initial fault type, may include:

[0065] The audio data to be tested is compared with each positive sample in the positive sample library. If a matching positive sample is found, the audio data to be tested is determined to be normal audio data; otherwise, the final fault type of the audio data to be tested is determined to be the initial fault type.

[0066] For example, in order to reduce the probability of misdetecting normal audio data as abnormal audio data, the audio data to be detected that is determined to be abnormal in step S110 can be subjected to secondary detection based on positive samples.

[0067] For example, the audio data to be detected can be compared with each positive sample in the positive sample library to determine the similarity between the audio data to be detected and the compared positive samples. If it is determined that there is a positive sample (which can be called a matching positive sample) with a similarity higher than a preset similarity threshold (which can be called the first similarity threshold), the audio data to be detected can be determined to be normal audio data. In this case, the final fault type of the audio data to be detected can be determined to be normal (that is, the audio data to be detected is normal audio data).

[0068] If there is no positive sample matching the audio data to be detected in the positive sample library, the audio data to be detected can be determined to be abnormal audio data, and the final fault type of the audio data to be detected can be determined as the initial fault type.

[0069] For example, for the acquired audio data to be detected, audio features can be extracted from the audio data to obtain the audio features of the audio data to be detected, and the audio features can be compared with the model files of each positive sample to determine the similarity between the audio data to be detected and each positive sample.

[0070] It is evident that by maintaining a positive sample library, and performing secondary filtering based on positive samples in the positive sample library for audio data identified as abnormal, the probability of detecting normal audio data as abnormal audio data can be effectively reduced, thereby improving the accuracy of abnormal audio detection.

[0071] In some embodiments, step S120, performing secondary detection on the audio data to be detected based on positive and / or negative samples, and determining the final fault type of the audio data to be detected based on the secondary detection results and the initial fault type, may include:

[0072] The audio data to be tested is compared with each negative sample in the negative sample library. If a matching negative sample exists, the fault type corresponding to the negative sample with the highest matching degree is determined as the final fault type of the audio data to be tested; otherwise, the final fault type of the audio data to be tested is determined as the initial fault type.

[0073] For example, in order to reduce the probability of fault type detection errors, for the audio data to be detected that is determined to be abnormal in step S110, a second detection can be performed on the audio data to be detected based on negative samples.

[0074] For example, the audio data to be detected can be compared with each negative sample in the negative sample library to determine the similarity between the audio data to be detected and the compared negative samples, so as to determine whether there are negative samples with a similarity higher than a preset similarity threshold (which can be called the second similarity threshold).

[0075] If there are negative samples in the negative sample library that have a similarity to the audio data to be detected that is higher than the second similarity threshold, the fault type corresponding to the negative sample with the highest similarity to the audio data to be detected (which can be called the matched negative sample) can be determined as the final fault type of the audio data to be detected.

[0076] If there are no negative samples in the negative sample library that have a similarity higher than the second similarity threshold to the audio data to be detected, that is, if there are no matching negative samples, the final fault type of the audio data to be detected can be determined as the initial fault type.

[0077] For example, for the acquired audio data to be detected, audio features can be extracted from the audio data to obtain the audio features of the audio data to be detected, and the audio features can be compared with the model files of each negative sample to determine the similarity between the audio data to be detected and each negative sample.

[0078] It is evident that by maintaining a negative sample library, for audio data identified as abnormal, secondary filtering and detection can be performed based on the negative samples in the negative sample library to effectively improve the accuracy of determining the fault type of abnormal audio data.

[0079] In some embodiments, step S120, performing secondary detection on the audio data to be detected based on positive and / or negative samples, and determining the final fault type of the audio data to be detected based on the secondary detection results and the initial fault type, may include:

[0080] The audio data to be tested is compared with each positive sample in the positive sample library. If a matching positive sample exists, the audio data to be tested is determined to be normal audio data. If no matching positive sample exists, the audio data to be tested is compared with each negative sample in the negative sample library. If a matching negative sample exists, the fault type corresponding to the negative sample with the highest matching degree is determined as the final fault type of the audio data to be tested. If no matching negative sample exists, the final fault type of the audio data to be tested is determined as the initial fault type.

[0081] For example, in order to reduce the probability of misdetecting normal audio data as abnormal audio data and improve the accuracy of determining the fault type of abnormal audio data, the audio data to be detected that is determined to be abnormal in step S110 can be subjected to secondary detection based on positive and negative samples respectively.

[0082] For example, for the audio data to be detected that is determined to be abnormal in step S110, the audio data to be detected can be compared with each positive sample in the positive sample library to determine the similarity between the audio data to be detected and the compared positive samples. If it is determined that there is a positive sample (which can be called a matching positive sample) with a similarity higher than a preset similarity threshold (such as the first similarity threshold mentioned above), the audio data to be detected is determined to be normal audio data. In this case, the final fault type of the audio data to be detected can be determined to be normal (that is, the audio data to be detected is normal audio data).

[0083] If there is no positive sample matching the audio data to be detected in the positive sample library, the audio data to be detected can be determined to be abnormal audio data. In this case, the audio data to be detected can be compared with each negative sample in the negative sample library to determine the similarity between the audio data to be detected and the compared negative samples, so as to determine whether there are negative samples with a similarity higher than a preset similarity threshold (such as the second similarity threshold mentioned above).

[0084] If there are negative samples in the negative sample library that have a similarity to the audio data to be detected that is higher than the second similarity threshold, the fault type corresponding to the negative sample with the highest similarity to the audio data to be detected (which can be called the matched negative sample) can be determined as the final fault type of the audio data to be detected.

[0085] If there are no negative samples in the negative sample library that have a similarity higher than the second similarity threshold to the audio data to be detected, that is, if there are no matching negative samples, the final fault type of the audio data to be detected can be determined as the initial fault type.

[0086] It is evident that by maintaining positive and negative sample libraries, for audio data identified as anomalous, secondary filtering based on positive samples in the positive sample library can effectively reduce the probability of detecting normal audio data as anomalous audio data and improve the accuracy of anomalous audio detection. For audio data to be detected that does not match any of the positive samples in the positive sample library, the fault type of the audio data to be detected can be filtered again based on the negative samples in the negative sample library, which can effectively improve the accuracy of determining the fault type of anomalous audio data.

[0087] In some embodiments, the abnormal audio detection method provided in this application may further include:

[0088] In response to detected sample management commands, output the sample management operation interface;

[0089] Based on the new sample operation instructions detected through the sample management operation interface, obtain positive samples and / or negative samples, and save the obtained positive samples and / or negative samples to the sample database.

[0090] For example, in order to improve the controllability and flexibility of sample maintenance, users can manage samples through relevant functional interfaces.

[0091] Accordingly, upon detecting a sample management command, the system can respond to the detected sample management command and output a sample management operation interface.

[0092] For example, when adding a new sample through the sample management interface, you can edit / select the sample category and edit the sample name.

[0093] For example, the sample category can include normal or faulty. For normal or faulty samples (i.e., positive or negative samples), the type can be further subdivided, such as faulty samples (i.e., negative samples) can be further subdivided into fault types.

[0094] For example, fault types may include, but are not limited to, mechanical faults, insufficient power, equipment faults, receiving faults, transmission faults, transmission signal faults, signal delays, radio blockages, or mechanical aging.

[0095] For normal samples, the types may include, but are not limited to, signal interference or external interference.

[0096] It should be noted that, for multi-channel audio data acquisition scenarios, considering that different channels acquire audio data from different locations of the device under test, and that normal and abnormal audio data from different locations of the device usually have different characteristics, when adding new samples, samples can be added for different channels.

[0097] For example, the sample data added to the sample database may include audio data and the corresponding model files.

[0098] In one example, the sample database includes a first type of information table and a second type of information table. The first type of information table is used to record the version information of the database, and the second type of information table is used to record sample information.

[0099] For example, when maintaining a sample database (positive sample database and / or negative sample database), at least two types of information tables can be maintained (which can be referred to as the first type information table and the second type information table, respectively).

[0100] For example, the first type of information table can be used to record database version information, and the second type of information table can be used to record sample information.

[0101] For example, database version information can be used to determine whether the database needs to be updated, that is, whether the sample information stored in the database needs to be updated.

[0102] For example, sample information may include, but is not limited to, channel number, sample ID, sample name, sample category, generation time, audio file name, audio file length, model file name, model file length, and a flag indicating whether the sample exists.

[0103] In some embodiments, the abnormal audio detection method provided in this application may further include:

[0104] The audio anomaly diagnosis algorithm and / or fault type detection algorithm are updated and trained based on positive and / or negative samples with a matching rate higher than a preset threshold. The updated audio anomaly diagnosis algorithm is used to determine whether there is an anomaly in the audio data to be detected and / or the updated fault type detection algorithm is used to determine the initial fault type of the audio data to be detected.

[0105] For example, in order to improve the accuracy of audio anomaly diagnosis and the accuracy of fault type determination, the matching rate of the maintained positive and / or negative samples can also be statistically analyzed.

[0106] For example, for any sample (positive or negative), the matching rate of the sample can be the ratio of the number of times the sample successfully matches the sample to be detected to the number of times the sample is compared with the sample to be detected.

[0107] For example, based on the matching rate of each positive sample and / or negative sample, positive samples and / or negative samples with a matching rate higher than a preset threshold can be used for updating and training the audio anomaly diagnosis algorithm and / or fault type detection algorithm.

[0108] For example, an audio anomaly diagnosis algorithm is used to determine whether there is an anomaly in the audio data to be detected; a fault type detection algorithm determines the fault type of the audio data that has an anomaly.

[0109] For example, the audio anomaly diagnosis algorithm and / or fault type detection algorithm can be updated and trained periodically based on positive and / or negative samples with a matching rate higher than a preset threshold.

[0110] For example, when the audio anomaly diagnosis algorithm and / or fault type detection algorithm are updated and trained in the manner described above, the updated audio anomaly diagnosis algorithm can be used to determine whether there is an anomaly in the audio data to be detected and / or the updated fault type detection algorithm can be used to determine the initial fault type of the audio data to be detected, thereby improving the accuracy of audio anomaly diagnosis and / or improving the accuracy of fault type determination.

[0111] In some embodiments, the abnormal audio detection method provided in this application may further include:

[0112] Receive remotely transmitted audio anomaly diagnosis algorithms and / or fault type detection algorithms;

[0113] When it is determined that the local audio anomaly diagnosis algorithm needs to be updated based on the version number of the remotely distributed audio anomaly diagnosis algorithm and the version number of the local audio anomaly diagnosis algorithm, the local audio anomaly diagnosis algorithm is updated to the remotely distributed audio anomaly diagnosis algorithm.

[0114] And / or, when it is determined that the local fault type detection algorithm needs to be updated based on the version number of the fault type detection algorithm remotely issued and the version number of the local fault type detection algorithm, the local fault type detection algorithm is updated to the fault type detection algorithm remotely issued.

[0115] For example, in order to improve the accuracy of audio anomaly diagnosis and fault type determination, the audio anomaly diagnosis algorithm and / or fault type detection algorithm can be updated remotely.

[0116] For example, an updated audio anomaly diagnosis algorithm and / or fault type detection algorithm can be trained through a training platform, and the updated audio anomaly diagnosis algorithm and / or fault type detection algorithm can be remotely sent to the audio diagnostic device.

[0117] For example, the audio anomaly diagnosis algorithm and / or fault type detection algorithm remotely distributed by the training platform may include version number information.

[0118] For example, when an audio diagnostic device receives a remotely sent audio anomaly diagnostic algorithm and / or fault type detection algorithm, it can determine whether an algorithm update is needed based on the version number of the received audio anomaly diagnostic algorithm and / or fault type detection algorithm, as well as the version number of the local audio anomaly diagnostic algorithm and / or fault type detection algorithm.

[0119] For example, an audio diagnostic device can compare the version number of a remotely distributed audio anomaly diagnostic algorithm with the version number of a local audio anomaly diagnostic algorithm. If the two are inconsistent, it is determined that an algorithm update is needed. In this case, the local audio anomaly diagnostic algorithm can be updated to the remotely distributed audio anomaly diagnostic algorithm. If the two are consistent, it is determined that an algorithm update is not needed. In this case, the local audio anomaly diagnostic algorithm can remain unchanged.

[0120] Similarly, audio diagnostic devices can compare the version number of the remotely issued fault type detection algorithm with the version number of the local fault type detection algorithm. If the two are inconsistent, it is determined that an algorithm update is needed. In this case, the local fault type detection algorithm can be updated to the remotely issued fault type detection algorithm. If the two are consistent, it is determined that no algorithm update is needed. In this case, the local fault type detection algorithm can remain unchanged.

[0121] Please see Figure 2 This is a flowchart illustrating a method for updating faulty audio samples provided in an embodiment of this application. Figure 2 As shown, the method for updating the faulty audio sample may include the following steps:

[0122] Step S200: Acquire first audio data using an embedded audio acquisition device.

[0123] In this embodiment of the application, in order to detect device faults, during the operation of the device under test, an embedded audio acquisition device can be used to acquire audio data (referred to as first audio data) at a specified location of the device under test.

[0124] For example, the specified location may include, but is not limited to, the head, tail, inner cavity, or outer wall of the device.

[0125] Step S210: When the first audio data is determined to be abnormal by the audio detection algorithm, the first audio data is input to the fault type detection module. The fault type detection module is configured to be displayed on the screen in the form of a webpage. The webpage format of the fault type detection module includes the input sequence number of the sample audio data, the sample name of the sample audio data, the sample type of the sample audio data, the acquisition time of the sample audio data, and interactive tags that allow users to edit.

[0126] In this embodiment of the application, for the acquired first audio data, an audio detection algorithm can be used to perform anomaly detection in order to determine whether the first audio data is abnormal.

[0127] For example, the audio detection algorithm described above can employ a deep learning algorithm.

[0128] In this embodiment of the application, when it is determined that the first audio data is abnormal, the first audio data can be input to the fault type detection module, which will perform fault type detection on the first audio data and generate a sample audio information record corresponding to the first audio data.

[0129] For example, the fault type detection module can be presented on the display as a webpage.

[0130] For example, the webpage format of the fault type detection module may include, but is not limited to, the input sequence number of the sample audio data, the sample name of the sample audio data, the sample type of the sample audio data, the acquisition time of the sample audio data, and interactive tags that users can edit.

[0131] The input sequence number of the sample audio data can be generated according to a preset strategy. For example, it can be generated sequentially, with the input sequence number of the first recorded audio sample data being 1, the input sequence number of the second recorded audio sample data being 2, and so on.

[0132] The sample name of the sample audio data can be generated based on the fault type detection result of the fault type detection module on the first audio data. For example, the sample name of the sample audio data can include normal or faulty, as well as the fault type of the audio or the cause or source of the audio, etc.

[0133] For example, the sample name of the sample audio data is editable.

[0134] The sample type of the sample audio data can be generated based on the fault type detection result of the fault type detection module on the first audio data, which can include normal or fault.

[0135] The acquisition time of the sample audio data can be determined based on the system time of obtaining the first audio data in step S210.

[0136] Interactive tags that allow users to edit are used to interact with users. Users can use these interactive tags to trigger editable information in the sample audio data, such as editing the sample name of the sample audio data.

[0137] For example, a webpage illustration of the fault type detection module can be shown as follows: Figure 5 As shown.

[0138] Step S220: In response to the user's triggering of the interaction mark corresponding to the first audio data, update the sample name of the first audio data so that the sample name of the first audio data includes the fault type input by the user, and then update the fault type detection module.

[0139] In this embodiment of the application, when a user triggers an interaction marker corresponding to the first audio data, the sample name of the first audio data can be updated in response to the triggering command, and then the fault type detection module can be updated.

[0140] For example, when a user triggers an interaction with the first audio data, the editable portion of the sample name of the first audio data can be displayed in the edit box, allowing the user to edit the editable portion of the sample name of the first audio data.

[0141] Step S230: Acquire second audio data using an embedded audio acquisition device.

[0142] Step S240: When the second audio data is determined to be abnormal by the audio detection algorithm, the second audio data is input to the updated fault type detection module.

[0143] In this embodiment of the application, after the fault type detection module is updated in accordance with the manner described in steps S200 to S210, the audio data (referred to as the second audio data in this document) obtained by the embedded audio acquisition device can be first determined by the audio detection algorithm to determine whether the second audio data is abnormal.

[0144] If the second audio data is determined to be abnormal by the audio detection algorithm, the second audio data can be input to the updated fault type detection module, which will then perform fault type detection on the second audio data.

[0145] Step S250: If a sample audio data matching the second audio data is found in the updated fault type detection module, the fault type will be automatically presented in the sample name of the second audio data based on the fault type of the matching sample audio data.

[0146] In this embodiment of the application, when the second audio data is input into the updated fault type detection module, the sample audio data with information already recorded in the updated fault type detection module can be compared with the second audio data to determine whether there is sample audio data in the updated fault type detection module that matches the second audio data.

[0147] If a sample audio data matching the second audio data is found in the updated fault type detection module, the fault type can be automatically displayed in the sample name of the second audio data based on the fault type of the matching sample audio data.

[0148] For example, a specific implementation for determining whether there is sample audio data matching the second audio data in the updated fault type detection module can be found in [reference needed]. Figure 1The relevant descriptions in the embodiments of the method flow shown are not repeated here.

[0149] It should be noted that, in this embodiment of the application, if no sample audio data matching the second audio data is found in the updated fault type detection module, the sample audio information record corresponding to the second audio data can be generated in the manner described in steps S200 to S220, and the updated fault type detection module can be further updated.

[0150] Furthermore, in steps S200 to S220 above, if the first audio data is input to the fault type detection module, and a sample audio data matching the first audio data is found in the fault type detection module, the fault type can be automatically displayed in the sample name of the first audio data, without requiring the user to manually edit the fault type in the sample name of the first audio data. That is, step S220 can be executed even if no sample audio data matching the first audio data is found in the fault type detection module.

[0151] In some embodiments, both the first audio data and the second audio data are acquired using a first channel of an embedded audio acquisition device, which includes a first channel and at least one other channel; the matched sample audio data is limited to the sample audio data from the first channel in the updated fault type detection module.

[0152] For example, considering that normal and abnormal audio data at different locations of a device usually have different characteristics, in order to improve the accuracy of device fault detection, different channels of an embedded audio acquisition device can be used to acquire audio data from different locations of the device under test during audio acquisition.

[0153] Similarly, when maintaining sample audio data, channels can be distinguished, and sample audio data for different channels can be maintained separately.

[0154] When determining the fault type of audio data based on sample audio data, for any channel, the audio data acquired needs to be compared with the sample audio data from that channel to determine the fault type of the audio data.

[0155] Accordingly, the first audio data acquired in step S200 and the second audio data acquired in step S230 are acquired using the same channel (referred to herein as the first channel) of the embedded audio acquisition device. The embedded audio acquisition device includes this first channel and at least one other channel.

[0156] In step S250, the sample audio data used for comparison with the second audio data in the updated fault type module is also sample audio data from the first channel.

[0157] In one example, in response to a user's interaction with the first audio data, a new webpage is displayed on the screen, so that the fault type detection module can accept the fault type input by the user and does not allow the user to change the first channel where the first audio data is located.

[0158] For example, upon detecting a trigger command for an interactive marker corresponding to the first audio data, a new page can be displayed in response to the trigger command.

[0159] The new page allows users to change the fault type in the sample name of the first audio data, but does not allow users to change the channel where the first audio data is located (i.e., the first channel mentioned above).

[0160] For example, a schematic diagram of the new page can be as follows: Figure 6 As shown.

[0161] For example, based on this new page, users can edit the fault type of the first audio data.

[0162] For example, the fault type detection module can store multiple sample audio data to a non-volatile storage medium.

[0163] For example, in the fault type detection module, the storage area for sample audio data of the normal category (which can be called positive samples) can be called the positive sample library; the storage area for sample audio data of the fault category (which can be called negative samples) can be called the negative sample library.

[0164] As an example, the method for updating faulty audio samples provided in this application embodiment may further include:

[0165] Acquire third-party audio data using an embedded audio acquisition device;

[0166] No alarm is output when the third audio data is determined to be normal by the audio detection algorithm;

[0167] When the third audio data is determined to be abnormal by the audio detection algorithm:

[0168] Input the third audio data into the updated fault type detection module;

[0169] If no sample audio data matching the third audio data is found in the updated fault type detection module, the user is prompted to enter the fault type so that the updated fault type detection module can be updated again based on the third audio data.

[0170] For example, after updating the fault type detection module in accordance with the manner described in the above embodiments, for the audio data (referred to as the third audio data in this document) obtained by the embedded audio acquisition device, the audio detection algorithm can be used to determine whether the third audio data is abnormal.

[0171] If the third audio data is determined to be normal by the audio detection algorithm, no alarm may be output.

[0172] If the third audio data is determined to be abnormal by the audio detection algorithm, the third audio data can be input into the updated fault type detection module to determine whether there is sample audio data that matches the third audio data in the updated fault type detection module.

[0173] It should be noted that when the embedded audio acquisition device is a multi-channel audio acquisition device, it is necessary to determine whether there is sample audio data that matches the third audio data from the sample audio data from the channel where the third audio data is located.

[0174] For example, if no sample audio data matching the third audio data is found in the updated fault type detection module, the user can be prompted to enter the fault type so that the updated fault type detection module can be updated again based on the third audio data.

[0175] For example, if no sample audio data matching the third audio data is found in the updated fault type detection module, a sample audio information record corresponding to the third audio data can be generated, and an editable box can be displayed in the sample name of the sample audio information record, allowing the user to input the fault type of the third audio data.

[0176] In one example, after the number of times a matched sample audio data is successfully matched exceeds a threshold, the matched sample audio data is fed as a negative sample into the audio detection algorithm in order to update the audio detection algorithm.

[0177] For example, for any sample audio data, if the number of times the sample audio data is successfully matched exceeds a threshold, the matched sample audio data can be used as a negative sample input to the audio detection algorithm to update the audio detection algorithm and improve the accuracy of audio anomaly diagnosis.

[0178] To enable those skilled in the art to better understand the technical solutions provided in the embodiments of this application, the technical solutions provided in the embodiments of this application are described below with reference to specific examples.

[0179] In this embodiment, the audio diagnostic device can collect audio data (i.e., the audio data to be tested) from different locations of the device under test through multiple real-time audio data acquisition channels, and perform anomaly diagnosis on the audio data collected from each channel to determine whether there are any abnormalities in the audio data.

[0180] For audio data identified as abnormal, fault type detection can be performed to determine the initial fault type. Then, based on the maintained positive and negative samples, the audio data undergoes secondary filtering to determine the final fault type. A flowchart illustrating this process can be shown below. Figure 3 ( Figure 3 Taking four channels as an example (i.e., real-time data acquisition channels 1 to 4), as shown in the figure.

[0181] For example, the audio diagnostic device can acquire audio data through an embedded device (i.e., the embedded audio acquisition device described above).

[0182] For example, when determining the initial fault type of the collected audio data, if the fault type of the audio data cannot be determined, it can be defaulted to an unknown fault type.

[0183] In this embodiment, a flowchart illustrating the process by which the audio diagnostic device determines whether audio data is abnormal can be found. Figure 4 .

[0184] like Figure 4 As shown, for any channel of audio data collected, an anomaly assessment score can be obtained using a trained audio anomaly diagnosis algorithm model (also known as an audio detection algorithm model). This score is then compared with a preset score threshold. If the anomaly assessment score is lower than the preset score threshold, the audio data is determined to be normal and no further processing is required. If the anomaly assessment score is not lower than the preset score threshold, the audio data is determined to be abnormal. In this case, the initial fault type of the audio data can be further determined, and secondary filtering detection can be performed.

[0185] In this embodiment, for audio data that is still determined to be abnormal audio data after secondary filtering and detection, an alarm file can be generated and uploaded to the platform.

[0186] For example, the alarm file includes abnormal audio data and the fault type.

[0187] In this embodiment, the audio diagnostic device further includes a fault type detection module, which is used to further detect fault types in audio data that are diagnosed as abnormal by the audio anomaly diagnostic algorithm model.

[0188] For example, the fault type detection module can be configured to be displayed on a screen as a webpage (this webpage can also be called a sample management operation interface). A schematic diagram of the webpage format of the fault type detection module can be shown as follows: Figure 5 As shown.

[0189] For example, the sample management interface may include, but is not limited to, the input sequence number of the sample audio data, the sample name of the sample audio data, the sample type of the sample audio data, the acquisition time of the sample audio data, and interactive markers that users can edit.

[0190] Upon detecting a user's trigger command for the interactive marker corresponding to the audio data, the system can respond to the trigger command, update the sample name of the audio data, and then update the fault type detection module.

[0191] For example, when a user triggers an interaction with the corresponding audio data, the editable portion of the audio data sample name can be displayed in the edit box, allowing the user to edit the editable portion of the audio data sample name.

[0192] For example, the sample management interface may also include add options, import options, download options, and delete options.

[0193] like Figure 5 As shown, when the user selects the add option, the user can select audio data as a sample (positive sample or negative sample) from the audio data that has been determined to be normal or abnormal audio data after secondary filtering and detection by the audio diagnostic device; when the user selects the import option, audio files can be imported from external sources as samples (positive samples or negative samples).

[0194] Taking the user's selection of added options as an example, the output can be as follows: Figure 6 The interface shown allows you to add samples, edit / select the category of the added sample, and edit the sample name.

[0195] For example, considering that different channels collect audio data from different locations of the device under test, and that normal and abnormal audio data from different locations of the device usually have different characteristics, samples can be added for different channels when adding samples.

[0196] like Figure 6As shown, when a user needs to add sample data, they can select the category of the sample to be added for the channel to which the sample needs to be added, and edit the name of the added sample.

[0197] For example, sample data may include audio data and the corresponding model file.

[0198] For example, the sample database can maintain two types of information tables: database supplementary information tables (i.e., the first type of information tables mentioned above) and sample information tables (i.e., the second type of information tables mentioned above).

[0199] For example, the database additional information table is used to record the database version, and its format can be as shown in Table 1.

[0200] Table 1

[0201] idx_id db_ver

[0202] Where idx_id is the index identifier, used to uniquely identify a database supplementary information table; db_ver is the database version number.

[0203] For example, the sample information table is used to store various information about the sample, which may include, but is not limited to, channel number, sample ID, sample name, sample type, generation time, audio file name, audio file length, model file name, model file length, sample existence flag, etc., and its format can be as shown in Table 2.

[0204] Table 2

[0205]

[0206] Wherein, idx_id is the index identifier used to uniquely identify a sample information table, chan is the channel number, sample_id is the sample identifier, sample_name is the sample name, sample_type is the sample type, time is the generation time, audio_file is the audio file name, audio_file_len is the audio file length, model_file is the model file name, model_file_len is the model file length, and exist_flag is the sample existence flag.

[0207] like Figure 7 and Figure 8 As shown in this embodiment, for audio data diagnosed as abnormal, it can be compared with each positive sample in the positive sample library one by one. If there is a positive sample with a similarity exceeding the threshold (such as the first similarity threshold mentioned above), then the audio data is determined to be normal audio data and there is no need to report it to the platform.

[0208] If there are no positive samples with similarity exceeding the threshold, the audio data is further compared with each negative sample in the negative sample library. If there are negative samples with similarity exceeding the threshold (such as the second similarity threshold mentioned above), that is, there are negative samples that match the audio data, the fault type corresponding to the negative sample with the highest similarity is determined as the final fault type of the audio data.

[0209] If there are no negative samples with similarity exceeding the threshold, the initial fault type of the audio data is determined as the final fault type of the audio data.

[0210] like Figure 9 As shown, in this embodiment, the audio diagnostic device can also record the number of positive sample matches and the number of negative sample matches, calculate the matching rate of each positive sample and negative sample, and periodically update and train the audio anomaly diagnosis algorithm and the fault type detection algorithm based on positive samples and negative samples with matching rates higher than a preset threshold. It can also determine whether there is an anomaly in the audio data to be detected based on the updated audio anomaly diagnosis algorithm, and determine the initial fault type of the audio data to be detected based on the updated fault type detection algorithm.

[0211] like Figure 10 As shown, in this embodiment, relevant personnel can also remotely update the audio anomaly diagnosis algorithm and fault type detection algorithm used by the audio diagnostic device through the training platform. When the audio diagnostic device receives the audio anomaly diagnosis algorithm and fault type detection algorithm remotely issued by the training platform, it can perform algorithm verification, that is, compare the version number of the received audio anomaly diagnosis algorithm with the version number of the local audio anomaly diagnosis algorithm, and compare the version number of the received fault type detection algorithm with the version number of the local fault type detection algorithm, to determine whether the local audio anomaly diagnosis algorithm and fault type detection algorithm need to be updated.

[0212] If it is determined that the local audio anomaly diagnosis algorithm and fault type detection algorithm need to be updated, the audio diagnostic device can update the sample library, that is, update the local audio anomaly diagnosis algorithm and fault type detection algorithm to the audio anomaly diagnosis algorithm and fault type detection algorithm issued by the training platform.

[0213] The complete abnormal audio detection process can be as follows: Figure 11 As shown.

[0214] The method provided in this application has been described above. The apparatus provided in this application is described below:

[0215] Please see Figure 12 This is a schematic diagram of the structure of an abnormal audio detection device provided in an embodiment of this application, as shown below. Figure 12 As shown, the abnormal audio detection device may include:

[0216] The acquisition unit is used to acquire the audio data to be detected;

[0217] The first detection unit is used to determine the initial fault type of the audio data to be detected when it is determined that there is an abnormality in the audio data to be detected.

[0218] The second detection unit is used to perform secondary detection on the audio data to be detected based on positive samples and / or negative samples; and to determine the final fault type of the audio data to be detected based on the secondary detection results and the initial fault type.

[0219] In one embodiment, the second detection unit performs secondary detection on the audio data to be detected based on positive and / or negative samples, and determines the final fault type of the audio data to be detected based on the secondary detection results and the initial fault type, including:

[0220] The audio data to be detected is compared with each positive sample in the positive sample library. If a matching positive sample is found, the audio data to be detected is determined to be normal audio data; otherwise, the final fault type of the audio data to be detected is determined to be the initial fault type.

[0221] or,

[0222] The audio data to be detected is compared with each negative sample in the negative sample library. If a matching negative sample exists, the fault type corresponding to the negative sample with the highest matching degree is determined as the final fault type of the audio data to be detected; otherwise, the final fault type of the audio data to be detected is determined as the initial fault type.

[0223] or,

[0224] The audio data to be detected is compared with each positive sample in the positive sample library. If a matching positive sample exists, the audio data to be detected is determined to be normal audio data. If no matching positive sample exists, the audio data to be detected is compared with each negative sample in the negative sample library. If a matching negative sample exists, the fault type corresponding to the negative sample with the highest matching degree is determined as the final fault type of the audio data to be detected. If no matching negative sample exists, the final fault type of the audio data to be detected is determined as the initial fault type.

[0225] In one embodiment, such as Figure 13 As shown, the device further includes:

[0226] The sample maintenance unit is used to respond to detected sample management instructions and output a sample management operation interface; based on the new sample operation instructions detected through the sample management operation interface, it acquires positive samples and / or negative samples and saves the acquired positive samples and / or negative samples to the sample database.

[0227] In one embodiment, such as Figure 14 As shown, the device further includes:

[0228] The first algorithm update module is used to update and train the audio anomaly diagnosis algorithm and / or fault type detection algorithm based on positive samples and / or negative samples with a matching rate higher than a preset threshold.

[0229] The first detection module is further configured to determine whether the audio data to be detected is abnormal based on the updated audio anomaly diagnosis algorithm and / or to determine the initial fault type of the audio data to be detected based on the updated fault type detection algorithm.

[0230] In one embodiment, such as Figure 15 As shown, the device further includes:

[0231] The second algorithm update unit is used to receive the audio anomaly diagnosis algorithm and / or fault type detection algorithm remotely sent; when it is determined that the local audio anomaly diagnosis algorithm needs to be updated based on the version number of the remotely sent audio anomaly diagnosis algorithm and the version number of the local audio anomaly diagnosis algorithm, the local audio anomaly diagnosis algorithm is updated to the audio anomaly diagnosis algorithm remotely sent.

[0232] And / or, when it is determined that the local fault type detection algorithm needs to be updated based on the version number of the fault type detection algorithm remotely issued and the version number of the local fault type detection algorithm, the local fault type detection algorithm is updated to the fault type detection algorithm remotely issued.

[0233] Please see Figure 16 This is a schematic diagram of the structure of a fault audio sample updating device provided in an embodiment of this application, as shown below. Figure 16 As shown, the device for updating the faulty audio sample may include:

[0234] The acquisition unit is used to acquire first audio data using an embedded audio acquisition device;

[0235] An input unit is used to input the first audio data to a fault type detection module when the first audio data is determined to be abnormal by an audio detection algorithm. The fault type detection module is configured to be presented on a display as a webpage. The webpage format of the fault type detection module includes the input sequence number of the sample audio data, the sample name of the sample audio data, the sample type of the sample audio data, the acquisition time of the sample audio data, and interactive tags that allow users to edit.

[0236] The update unit is used to update the sample name of the first audio data in response to the user's triggering of the interaction mark corresponding to the first audio data, so that the sample name of the first audio data includes the fault type input by the user, and then update the fault type detection module.

[0237] The acquisition unit is also used to acquire second audio data using an embedded audio acquisition device;

[0238] The input unit is further configured to input the second audio data to the updated fault type detection module when the second audio data is determined to be abnormal by the audio detection algorithm;

[0239] The display unit is used to automatically display the fault type in the sample name of the second audio data based on the fault type of the matching sample audio data when a sample audio data matching the second audio data is found in the updated fault type detection module.

[0240] In some embodiments, both the first audio data and the second audio data are acquired using a first channel of the embedded audio acquisition device, which includes the first channel and at least one other channel; the matched sample audio data is limited to sample audio data from the first channel in the updated fault type detection module.

[0241] In some embodiments, the display unit is further configured to present a new webpage on the display in response to a user's triggering of an interaction marker corresponding to the first audio data, so that the fault type detection module can accept the fault type input by the user and does not allow the user to change the first channel where the first audio data is located.

[0242] In some embodiments, the fault type detection module stores multiple sample audio data, and the fault type detection module stores the data to a non-volatile storage medium.

[0243] In some embodiments, such as Figure 17 As shown, the device further includes: an output unit and a prompting unit;

[0244] The acquisition unit is also used to acquire third audio data using the embedded audio acquisition device;

[0245] The output unit is configured not to output an alarm when the third audio data is determined to be normal by the audio detection algorithm.

[0246] The input unit is also used to input the third audio data to the updated fault type detection module when the third audio data is determined to be abnormal by the audio detection algorithm.

[0247] The prompting unit is used to prompt the user to input a fault type if no sample audio data matching the third audio data is found in the updated fault type detection module, so that the updated fault type detection module can be updated again based on the third audio data.

[0248] In some embodiments, the input unit is further configured to, after the number of times the matched sample audio data has been successfully matched exceeds a threshold, input the matched sample audio data as a negative sample into the audio detection algorithm in order to update the audio detection algorithm.

[0249] Correspondingly, this application also provides Figure 12-17 Hardware structure of any of the illustrated devices. See also Figure 18 The hardware structure may include: a processor and a machine-readable storage medium, the machine-readable storage medium storing machine-executable instructions that can be executed by the processor; the processor is used to execute the machine-executable instructions to implement the method disclosed in the above example of this application.

[0250] Based on the same application concept as the above method, this application embodiment also provides a machine-readable storage medium storing a plurality of computer instructions, which, when executed by a processor, can implement the method disclosed in the above examples of this application.

[0251] For example, the aforementioned machine-readable storage medium can be any electronic, magnetic, optical, or other physical storage device that can contain or store information such as executable instructions, data, etc. For instance, machine-readable storage media can be: RAM (Random Access Memory), volatile memory, non-volatile memory, flash memory, storage drives (such as hard disk drives), solid-state drives, any type of storage disk (such as optical discs, DVDs, etc.), or similar storage media, or combinations thereof.

[0252] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0253] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. An abnormal audio detection method, characterized by, The method comprises the following steps: acquiring to-be-detected audio data; in a case where it is determined that the to-be-detected audio data is abnormal, determining an initial fault type of the to-be-detected audio data; performing secondary detection on the to-be-detected audio data according to positive samples and / or negative samples, and determining a final fault type of the to-be-detected audio data according to a secondary detection result and the initial fault type; wherein the method further comprises: receiving an audio anomaly diagnosis algorithm and / or a fault type detection algorithm remotely issued; when it is determined that a local audio anomaly diagnosis algorithm needs to be updated according to a version number of the audio anomaly diagnosis algorithm remotely issued and a version number of the local audio anomaly diagnosis algorithm, updating the local audio anomaly diagnosis algorithm to the audio anomaly diagnosis algorithm remotely issued; and / or, when it is determined that a local fault type detection algorithm needs to be updated according to a version number of the fault type detection algorithm remotely issued and a version number of the local fault type detection algorithm, updating the local fault type detection algorithm to the fault type detection algorithm remotely issued.

2. The method of claim 1, wherein, The method further comprises: respectively comparing the to-be-detected audio data with each positive sample in a positive sample library, and if there is a matching positive sample, determining that the to-be-detected audio data is normal audio data; otherwise, determining that the final fault type of the to-be-detected audio data is the initial fault type; or, respectively comparing the to-be-detected audio data with each negative sample in a negative sample library, and if there is a matching negative sample, determining that the final fault type of the to-be-detected audio data is the fault type corresponding to the negative sample with the highest matching degree; otherwise, determining that the final fault type of the to-be-detected audio data is the initial fault type; or, respectively comparing the to-be-detected audio data with each positive sample in a positive sample library, and if there is a matching positive sample, determining that the to-be-detected audio data is normal audio data; if there is no matching positive sample, respectively comparing the to-be-detected audio data with each negative sample in a negative sample library, and if there is a matching negative sample, determining that the final fault type of the to-be-detected audio data is the fault type corresponding to the negative sample with the highest matching degree; if there is no matching negative sample, determining that the final fault type of the to-be-detected audio data is the initial fault type.

3. The method of claim 1, wherein, The method further comprises: in response to a detected sample management instruction, outputting a sample management operation interface; acquiring positive samples and / or negative samples according to a detected new sample operation instruction through the sample management operation interface, and saving the acquired positive samples and / or negative samples to a sample database.

4. The method of claim 2, wherein, The method further comprises: performing update training on an audio anomaly diagnosis algorithm and / or a fault type detection algorithm according to positive samples and / or negative samples with a matching rate higher than a preset threshold, and determining whether to-be-detected audio data is abnormal according to the updated audio anomaly diagnosis algorithm and / or determining an initial fault type of the to-be-detected audio data according to the updated fault type detection algorithm.

5. A method of updating a failure audio sample, characterized by, The method comprises the following steps: acquiring first audio data by using an embedded audio acquisition device; when the first audio data is determined to be abnormal via an audio detection algorithm, inputting the first audio data to a fault type detection module, wherein the fault type detection module is configured to be presented in the form of a webpage on a display, the webpage form of the fault type detection module including an input serial number of sample audio data, a sample name of the sample audio data, a sample type of the sample audio data, a collection time of the sample audio data, and an interactive marker allowing user editing; in response to a user triggering the interactive marker corresponding to the first audio data, updating the sample name of the first audio data so that the sample name of the first audio data includes a fault type input by the user, and then updating the fault type detection module; acquiring second audio data by using the embedded audio acquisition device; when the second audio data is determined to be abnormal via the audio detection algorithm, inputting the second audio data to the updated fault type detection module; when sample audio data matching the second audio data is found in the updated fault type detection module, automatically presenting the fault type in the sample name of the second audio data based on the fault type of the matched sample audio data.

6. The method of claim 5, wherein, The first audio data and the second audio data are both acquired by using a first channel of the embedded audio acquisition device, the embedded audio acquisition device including the first channel and at least one other channel; the matched sample audio data is limited to sample audio data from the first channel in the updated fault type detection module.

7. The method of claim 6, wherein, in response to a user triggering the interactive marker corresponding to the first audio data, presenting a new webpage on the display so that the fault type detection module can accept a fault type input by the user and does not allow the user to change the first channel where the first audio data is located.

8. The method of claim 7, wherein, The fault type detection module stores a plurality of sample audio data, and the fault type detection module is stored in a non-volatile storage medium.

9. The method of claim 8, wherein, The method further includes: acquiring third audio data by using the embedded audio acquisition device; when the third audio data is determined to be normal via the audio detection algorithm, not outputting an alarm; when the third audio data is determined to be abnormal via the audio detection algorithm: inputting the third audio data to the updated fault type detection module; when no sample audio data matching the third audio data is found in the updated fault type detection module, prompting the user to input a fault type so as to update the updated fault type detection module again based on the third audio data.

10. The method of claim 9, wherein, When the number of times that the matched sample audio data is successfully matched exceeds a threshold, the matched sample audio data is input to the audio detection algorithm as a negative sample so as to update the audio detection algorithm.

Images