Equipment abnormity sound-light alarm method and system, medium and processor

Abstract

The invention discloses an equipment abnormity sound-light alarm method and system, a medium and a processor, and relates to the technical field of equipment fault sound-light alarm. Comprising the steps of obtaining operation sound signals of to-be-identified equipment; performing feature extraction on the operation sound signal to obtain feature values at a plurality of different moments; processing the characteristic value at the current moment through a preset voice recognition model to obtain a voice category of the to-be-recognized equipment, and determining an operation state of the to-be-recognized equipment according to the voice category; typical equipment fault states are selected to be matched with different alarm sounds, the different alarm sounds are arranged at different positions, and alarm lamps with different colors are emitted. According to the invention, the type of the fault is distinguished by acquiring the sound of the equipment, then different fault types are matched with different alarm sounds, and the voice alarm module gives an alarm, so that a worker can immediately distinguish the type of the fault of the equipment after hearing the corresponding voice alarm sound and take corresponding measures in time.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment fault sound and light alarm, and in particular to an equipment abnormal sound and light alarm method, system, medium and processor. Background Art

[0002] Many devices, such as engines, propellers, speakers, etc., do not have obvious changes in vision, touch, and smell when internal parts fail. However, there are usually obvious changes in operating sounds. Therefore, when industrial equipment is working, operating sounds are one of the most important comprehensive representations of the health status of industrial equipment.

[0003] In the related art, equipment failure is usually judged based on manual experience. However, the above method has the problems of strong subjectivity, low recognition accuracy caused by manual fatigue, and scarcity of experienced personnel.

[0004] Traditional sound and light alarms make a "beep" or similar sound when they sound an alarm. The volume and frequency of the sound vary depending on the fault. However, this is not intuitive enough, and some staff members cannot distinguish the type of fault. Therefore, how to identify the type of fault from the various faults of the equipment, how to match it with various voices after identification, and finally select the appropriate voice before broadcasting it are relatively serious problems.

[0005] In view of this, a method, system, medium and processor for abnormal sound and light alarm of equipment are needed. Summary of the invention

[0006] In view of the problems existing in the prior art, the present invention provides a method, system, medium and processor for abnormal sound and light alarm of equipment, which can identify equipment failure. The specific technical solution is as follows:

[0007] A device abnormal sound and light alarm method, comprising:

[0008] Obtaining the operating sound signal of the device to be identified;

[0009] Extracting features of the running sound signal to obtain feature values ​​at multiple different times;

[0010] Processing the feature value at the current moment by using a preset sound recognition model to obtain the sound category of the device to be identified, so as to determine the operating state of the device to be identified according to the sound category;

[0011] Select typical equipment failure states to match different alarm sounds, set different alarm sounds at different locations, and emit alarm lights of different colors.

[0012] Preferably, the running sound signal is subjected to feature extraction to obtain feature values ​​at multiple different moments, including:

[0013] The running sound signal is divided into frames according to a preset rule to obtain sound frames at different times;

[0014] Information is extracted from the sound frame at each moment to obtain feature values ​​at the multiple different moments.

[0015] Preferably, the preset sound recognition model is obtained by training based on historical operating sound signals and corresponding abnormal sound category labels and abnormal confidence labels. The sound recognition model includes a first network unit, a threshold prediction layer, a classification layer and at least one second network unit. The first network unit is respectively connected to the threshold prediction layer and the classification layer, and the at least one second network unit is connected between the first network unit and the classification layer. When there are multiple second network units, multiple second network units are connected in sequence, and each second network unit is respectively connected to the classification layer. The sound recognition model uses cross entropy as the loss function.

[0016] Preferably, the preset sound recognition model is obtained by training based on historical operating sound signals and corresponding abnormal sound category labels and abnormal confidence labels, and includes:

[0017] Obtain historical operating sound signals of the device to be identified;

[0018] Extracting features of the historical operation sound signal to obtain feature values ​​at multiple different historical moments, as well as abnormal sound category labels and abnormal confidence labels;

[0019] Training a classification model to be trained according to the feature values ​​of the multiple different historical moments and the abnormal sound category labels at the corresponding moments, wherein the classification model to be trained includes the first network unit, the classification layer and the at least one second network unit;

[0020] If the first training stop condition is met, training the confidence prediction model to be trained according to the feature values ​​at the multiple different historical moments and the corresponding abnormal confidence labels, wherein the confidence prediction model to be trained includes the first network unit and the threshold prediction layer;

[0021] If the second training stop condition is met, the preset sound recognition model is obtained.

[0022] Preferably, the process of processing the feature value at the current moment by using a preset sound recognition model to obtain the sound category of the device to be identified includes:

[0023] Processing the characteristic value at the current moment according to the first network unit to obtain a first processing result;

[0024] Classifying the first processing result through the classification layer to obtain the sound type of the device to be identified at the current moment;

[0025] The first processing result is processed through the threshold prediction layer to obtain the confidence at the current moment, and the number of frame skips or the number of activations of the second network unit at the next moment is determined based on the confidence at the current moment.

[0026] An equipment abnormal sound and light alarm system, applied to the above method, comprises a sound acquisition module, a feature extraction module, a sound recognition module, a sound alarm module and a light alarm module;

[0027] The sound acquisition module is used to acquire the operation sound signal of the device to be identified;

[0028] The feature extraction module extracts features from the running sound signal to obtain feature values ​​at multiple different moments;

[0029] The sound recognition module is used to process the feature value at the current moment through a preset sound recognition model to obtain the sound category of the device to be identified, so as to determine the operating state of the device to be identified according to the sound category, wherein the preset sound recognition model is obtained by training according to historical operating sound signals and corresponding sound category labels, and the sound recognition model includes a first network unit, a threshold prediction layer, a classification layer and at least one second network unit, the first network unit is respectively connected to the threshold prediction layer and the classification layer, the at least one second network unit is connected between the first network unit and the classification layer, when there are multiple second network units, the multiple second network units are connected in sequence, and each second network unit is respectively connected to the classification layer, and the sound recognition model uses cross entropy as a loss function;

[0030] The sound alarm module is used to match different alarm sounds to different equipment failures;

[0031] The optical alarm module is used to match different alarm sounds with different light beam alarms.

[0032] A computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the above method.

[0033] A processor is used to run a program, wherein the program executes the above method when it is run.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] 1. In the present invention, the type of fault is identified by acquiring the sound of the equipment, and then different fault types are matched with different sound alarm sounds, and an alarm is issued through a voice alarm module, so that the staff can immediately identify the type of equipment fault after hearing the corresponding voice alarm sound and take corresponding measures in time.

[0036] 2. In the present invention, while different fault voices are broadcast for different faults, different light alarm modules are matched to further alarm the faults, and for different fault types, the voice alarm module is matched with the corresponding light alarm module and placed in the same place, so that the staff can intuitively find the fault type of the equipment through voice and light alarms. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0038] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0041] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0042] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0043] Example 1 See also Figure 1 , a device abnormal sound and light alarm method, comprising: Obtaining the operating sound signal of the device to be identified; Extracting features of the running sound signal to obtain feature values ​​at multiple different times; Processing the feature value at the current moment by using a preset sound recognition model to obtain the sound category of the device to be identified, so as to determine the operating state of the device to be identified according to the sound category; Select typical equipment failure states to match different alarm sounds, set different alarm sounds at different locations, and emit alarm lights of different colors.

[0049] In the above scheme, the equipment terminal may include various devices that can make sounds, such as engines, propellers, speakers, etc., including but not limited to fans, industrial equipment, and desktop computers. The equipment operation sound signal can be collected by a sound collection device, such as a pickup, a recorder, a microphone, etc. The sound collection device can collect the operation sound of the terminal equipment, and the operation sound is used as a research object. The sound collection device can send the collected operation sound to the feature extraction module and the sound recognition module to process the operation sound to obtain the sound type and identify the abnormality of the equipment. Finally, different alarm sounds and alarm beams are matched to different types of equipment and equipment fault types, and placed in different places to sound alarms, so that the staff can quickly identify the fault information of the equipment.

[0050] As a preferred embodiment, the running sound signal is subjected to feature extraction to obtain feature values ​​at multiple different moments, including:

[0051] The running sound signal is divided into frames according to a preset rule to obtain sound frames at different times;

[0052] Information is extracted from the sound frame at each moment to obtain feature values ​​at the multiple different moments.

[0053] As a preferred embodiment, the preset sound recognition model is obtained by training based on historical operating sound signals and corresponding abnormal sound category labels and abnormal confidence labels. The sound recognition model includes a first network unit, a threshold prediction layer, a classification layer and at least one second network unit. The first network unit is respectively connected to the threshold prediction layer and the classification layer, and the at least one second network unit is connected between the first network unit and the classification layer. When there are multiple second network units, multiple second network units are connected in sequence, and each second network unit is respectively connected to the classification layer. The sound recognition model uses cross entropy as the loss function.

[0054] As a preferred embodiment, the preset sound recognition model is obtained by training based on historical operating sound signals and corresponding abnormal sound category labels and abnormal confidence labels, and includes:

[0055] Obtain historical operating sound signals of the device to be identified;

[0056] Extracting features of the historical operation sound signal to obtain feature values ​​at multiple different historical moments, as well as abnormal sound category labels and abnormal confidence labels;

[0057] Training a classification model to be trained according to the feature values ​​of the multiple different historical moments and the abnormal sound category labels at the corresponding moments, wherein the classification model to be trained includes the first network unit, the classification layer and the at least one second network unit;

[0058] If the first training stop condition is met, training the confidence prediction model to be trained according to the feature values ​​at the multiple different historical moments and the corresponding abnormal confidence labels, wherein the confidence prediction model to be trained includes the first network unit and the threshold prediction layer;

[0059] If the second training stop condition is met, the preset sound recognition model is obtained.

[0060] As a preferred embodiment, the process of processing the feature value at the current moment by using a preset sound recognition model to obtain the sound category of the device to be identified includes:

[0061] Processing the characteristic value at the current moment according to the first network unit to obtain a first processing result;

[0062] Classifying the first processing result through the classification layer to obtain the sound type of the device to be identified at the current moment;

[0063] The first processing result is processed through the threshold prediction layer to obtain the confidence at the current moment, and the number of frame skips or the number of activations of the second network unit at the next moment is determined based on the confidence at the current moment.

[0064] Example 2

[0065] An audible and visual alarm system based on equipment operation signals, applied to the above method, comprises a sound acquisition module, a feature extraction module, a sound recognition module, a sound alarm module and a light alarm module;

[0066] The sound acquisition module is used to acquire the operation sound signal of the device to be identified;

[0067] The feature extraction module extracts features from the running sound signal to obtain feature values ​​at multiple different moments;

[0068] The sound recognition module is used to process the feature value at the current moment through a preset sound recognition model to obtain the sound category of the device to be identified, so as to determine the operating state of the device to be identified according to the sound category, wherein the preset sound recognition model is obtained by training according to historical operating sound signals and corresponding sound category labels, and the sound recognition model includes a first network unit, a threshold prediction layer, a classification layer and at least one second network unit, the first network unit is respectively connected to the threshold prediction layer and the classification layer, the at least one second network unit is connected between the first network unit and the classification layer, when there are multiple second network units, the multiple second network units are connected in sequence, and each second network unit is respectively connected to the classification layer, and the sound recognition model uses cross entropy as a loss function;

[0069] The sound alarm module is used to match different alarm sounds to different equipment failures;

[0070] The optical alarm module is used to match different alarm sounds with different light beam alarms.

[0071] Example 3

[0072] A computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the above method.

[0073] Example 4

[0074] A processor is used to run a program, wherein the program executes the above method when it is run.

[0075] Those of ordinary skill in the art will appreciate that the units of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0076] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0077] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0078] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nlyMemory), random access memory (RAM, RandomAccessMemory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A device abnormal sound and light alarm method, characterized in that: include: Obtaining the operating sound signal of the device to be identified; Extracting features of the running sound signal to obtain feature values ​​at multiple different times; Processing the feature value at the current moment by using a preset sound recognition model to obtain the sound category of the device to be identified, so as to determine the operating state of the device to be identified according to the sound category; Select typical equipment failure states to match different alarm sounds, set different alarm sounds at different locations, and emit alarm lights of different colors.

2. The device abnormal sound and light alarm method according to claim 1, characterized in that: The running sound signal is subjected to feature extraction to obtain feature values ​​at multiple different times, including: The running sound signal is divided into frames according to a preset rule to obtain sound frames at different times; Information is extracted from the sound frame at each moment to obtain feature values ​​at the multiple different moments.

3. The device abnormal sound and light alarm method according to claim 1, characterized in that: The preset sound recognition model is trained based on historical operating sound signals and corresponding abnormal sound category labels and abnormal confidence labels. The sound recognition model includes a first network unit, a threshold prediction layer, a classification layer and at least one second network unit. The first network unit is respectively connected to the threshold prediction layer and the classification layer, and the at least one second network unit is connected between the first network unit and the classification layer. When there are multiple second network units, multiple second network units are connected in sequence, and each second network unit is respectively connected to the classification layer. The sound recognition model uses cross entropy as the loss function.

4. The device abnormal sound and light alarm method according to claim 3 is characterized in that: The preset sound recognition model is trained based on historical operating sound signals and corresponding abnormal sound category labels and abnormal confidence labels, and includes: Obtain historical operating sound signals of the device to be identified; Extracting features of the historical operation sound signal to obtain feature values ​​at multiple different historical moments, as well as abnormal sound category labels and abnormal confidence labels; Training a classification model to be trained according to the feature values ​​of the multiple different historical moments and the abnormal sound category labels at the corresponding moments, wherein the classification model to be trained includes the first network unit, the classification layer and the at least one second network unit; If the first training stop condition is met, training the confidence prediction model to be trained according to the feature values ​​at the multiple different historical moments and the corresponding abnormal confidence labels, wherein the confidence prediction model to be trained includes the first network unit and the threshold prediction layer; If the second training stop condition is met, the preset sound recognition model is obtained.

5. The device abnormal sound and light alarm method according to claim 3 is characterized in that: The process of processing the feature value at the current moment by using a preset sound recognition model to obtain the sound category of the device to be recognized includes: Processing the characteristic value at the current moment according to the first network unit to obtain a first processing result; The first processing result is classified by the classification layer to obtain the sound type of the device to be identified at the current moment; the first processing result is processed by the threshold prediction layer to obtain the confidence at the current moment, and according to the confidence at the current moment, the number of frame skips is determined or the number of activations of the second network unit at the next moment is determined.

6. An equipment abnormal sound and light alarm system, characterized in that: The method applied to any one of claims 1 to 5, comprising a sound acquisition module, a feature extraction module, a sound recognition module, a sound alarm module and a light alarm module; The sound acquisition module is used to acquire the operation sound signal of the device to be identified; The feature extraction module extracts features from the running sound signal to obtain feature values ​​at multiple different moments; The sound recognition module is used to process the feature value at the current moment through a preset sound recognition model to obtain the sound category of the device to be identified, so as to determine the operating state of the device to be identified according to the sound category, wherein the preset sound recognition model is obtained by training according to historical operating sound signals and corresponding sound category labels, and the sound recognition model includes a first network unit, a threshold prediction layer, a classification layer and at least one second network unit, the first network unit is respectively connected to the threshold prediction layer and the classification layer, the at least one second network unit is connected between the first network unit and the classification layer, when there are multiple second network units, the multiple second network units are connected in sequence, and each second network unit is respectively connected to the classification layer, and the sound recognition model uses cross entropy as a loss function; The sound alarm module is used to match different alarm sounds to different equipment failures; The optical alarm module is used to match different alarm sounds with different light beam alarms.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the methods in claims 1 to 5.

8. A processor, characterized in that: The processor is used to run a program, wherein the program executes the method according to any one of claims 1 to 5 when running.

Images