Machine learning-based industrial equipment noise diagnosis method and diagnosis equipment
Through the noise diagnosis method of industrial equipment based on machine learning, the subjective deviation and high cost of noise detection in the existing technology are solved, and intelligent analysis without manual participation is realized. It is suitable for complex industrial environments and the objective diagnosis results are improved. and productivity.
Patent Information
- Application Number
- CN202411403934.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-09
- Publication Date
- 2025-01-07
AI Technical Summary
The existing industrial equipment noise detection methods rely on artificial auditory judgments and have subjective deviations, and the existing scientific measurement technology requires expensive equipment and high processing capabilities, which cannot be adapted to a narrow environment and are costly, and cannot be widely promoted.
Using machine learning-based industrial equipment noise diagnosis method, it uses machine learning models to collect and divide noise data, build a machine learning model for analysis, realizes intelligent diagnosis without manual participation, reduces equipment costs and adapts to complex environments.
It realizes simple and fast noise diagnosis, significantly improves the objectivity and consistency of diagnostic results, reduces equipment costs, is suitable for complex industrial environments, and improves production efficiency.
Smart Images
Figure CN119271967A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to noise diagnosis technology, and specifically to an industrial equipment noise diagnosis method and diagnostic equipment based on machine learning. Background Art
[0002] In traditional industrial environments, the judgment of industrial equipment noise mainly relies on human auditory judgment. Although this method is intuitive and convenient, it inevitably has a large subjective bias. This subjective bias may be caused by various factors such as individual hearing differences, different tolerance to noise, the influence of environmental factors, and the psychological state of the judge, resulting in the inability to maintain a high consistency in the judgment results, and the labor cost is high, and the effect is not ideal.
[0003] In order to overcome these subjective biases and improve the accuracy and objectivity of noise judgment, modern industrial noise detection has gradually introduced more scientific and accurate measurement technologies, mainly including the following technical means:
[0004] 1. Sound level meter measurement: A sound level meter is an instrument used to measure the strength of sound, which can provide relatively objective noise data. The sound level meter can be used to quantitatively measure the noise of industrial equipment, such as A sound level, C sound level, and octave band sound pressure level, so as to understand the characteristics of noise more accurately.
[0005] 2. Spectrum analysis: Spectrum analysis is a method of evaluating the nature of noise by analyzing the frequency components of noise. Through spectrum analysis, the main frequency components in the noise can be identified, and then the source and type of noise can be determined, providing a scientific basis for noise control.
[0006] 3. Sound intensity method: The sound intensity method is a method of evaluating noise energy by measuring the sound intensity value. Compared with the sound pressure method, the sound intensity method can more directly reflect the energy distribution and propagation characteristics of noise, so it has higher accuracy in certain specific occasions.
[0007] 4. Vibration method: The vibration method is to determine the amount of noise radiated by the machine by measuring the amount of vibration on the machine surface. This method is suitable for situations where the influence of environmental noise and reflected sound is large. It can eliminate the interference of some environmental factors and improve the accuracy of noise measurement.
[0008] 5. Noise source identification technology: In addition to the above measurement methods, modern industrial noise detection also uses a variety of noise source identification technologies, such as subjective identification method, separate operation method, selective isolation method, near-field sound intensity method, surface vibration velocity and sound pressure method, time domain analysis method, frequency domain analysis method and acoustic holography analysis method, etc. These technologies can analyze and identify noise sources from different angles, providing more comprehensive and accurate information for noise control.
[0009] However, all of the above methods require the use of terminal collection and host computer analysis modes, and require a host computer with high processing power and expensive sensors to obtain accurate diagnostic results. This makes the overall system complicated and has high requirements for the installation environment. It cannot be adapted to a relatively small industrial environment, and the cost is very high, so it cannot be widely promoted. Summary of the invention
[0010] In response to the problems existing in the prior art, this application proposes an industrial equipment noise diagnosis method and diagnostic equipment based on machine learning, which can make fault judgments on the noise of equipment running in complex industrial environments, realize simple, fast, and intelligent analysis without human intervention, and improve the industrial production rhythm and production line efficiency.
[0011] In a first aspect, the present application provides an industrial equipment noise diagnosis method based on machine learning, comprising the following contents:
[0012] S1, collects the original noise data of the working environment of industrial equipment;
[0013] S2, performing gear segmentation on the original noise data according to the gear segmentation time point;
[0014] S3, construct a machine learning model, input the noise data after gear segmentation into the machine learning model, and the machine learning model is used to analyze the noise data, extract noise characteristic values and output noise diagnosis results.
[0015] Furthermore, step S2 specifically includes the following contents:
[0016] S20, cutting out the first level of noise data;
[0017] S21, setting cutting time;
[0018] S22, cutting out the subsequent noise data gear according to the set cutting time.
[0019] Furthermore, in step S20, the first level of noise data is cut out according to the sudden change in sound pressure.
[0020] Furthermore, step S3 specifically includes the following contents:
[0021] S30, perform basic acoustic analysis on existing noise data, extract multiple eigenvalues, and build an initial machine learning model;
[0022] S31, perform feature value traversal and model traversal on the original noise data with OK and NG labels;
[0023] S32, fault diagnosis prediction is performed on the original noise data with OK and NG labels through the initial machine learning model;
[0024] S33, marking the fault diagnosis results predicted by the initial machine learning model with OK and NG labels;
[0025] S34, comparing the fault diagnosis results marked with OK and NG labels predicted by the initial machine learning model with the OK and NG labels carried by the original data itself, calculating the error rate between the two, and obtaining the model accuracy;
[0026] S35, for the noise data with errors, repeat steps S31-S35 and perform model prediction again;
[0027] S36, finally a machine learning model that meets the accuracy requirements is obtained and used for equipment fault diagnosis calculations in actual environments.
[0028] Furthermore, the method also includes the step of displaying the diagnosis result.
[0029] Furthermore, the method further includes the step of providing an alarm using sound and / or light based on the diagnosis result.
[0030] A diagnostic device for executing the industrial equipment noise diagnostic method, comprising:
[0031] Data acquisition module, used to collect raw noise data of the working environment of industrial equipment;
[0032] A data processing module is used to perform gear segmentation on the original noise data according to the gear segmentation time point to obtain noise data segments of each gear;
[0033] The fault judgment module is used to diagnose the noise data segment of each gear and obtain the diagnosis result.
[0034] Furthermore, the device also includes a fault display module, which is electrically / communicatively connected to the fault judgment module and is used to display the diagnosis result in the form of OK and NG.
[0035] Furthermore, the device also includes an alarm module, which is electrically / communicatively connected to the fault judgment module and is used to alarm the diagnosis result in the form of sound and / or light.
[0036] A computer program is stored in the diagnostic device, and the computer program is configured to execute the industrial equipment noise diagnostic method when running.
[0037] Compared with the existing methods, this application has the following beneficial effects:
[0038] This application achieves simple, fast, and human-free intelligent analysis through collection, judgment, analysis, and optimization, effectively saving labor costs and improving industrial production rhythm and production line efficiency; it analyzes and trains noise data based on machine learning models, and adjusts the accuracy of the results according to the errors, which significantly improves the objectivity and consistency of the diagnostic results, avoids subjective bias caused by human participation, and does not require expensive collection and analysis equipment, effectively reducing equipment costs and facilitating promotion and application. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 A schematic diagram of a process for performing gear segmentation on raw noise data in a method for diagnosing industrial equipment noise based on machine learning provided in an embodiment of the present application;
[0040] Figure 2 Schematic diagram of various noise coupling in industrial environment. DETAILED DESCRIPTION
[0041] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings.
[0042] It should be clear that the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.
[0043] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.
[0044] It should be understood that the term "and / or" used in this article is only a description of the same field of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.
[0045] In a first aspect, the present application provides an industrial equipment noise diagnosis method based on machine learning, comprising the following steps:
[0046] S1, collects the original noise data of the working environment of industrial equipment;
[0047] S2, performing gear segmentation on the original noise data according to the gear segmentation time point to obtain noise data segments for each gear;
[0048] S3, build a machine learning model, input each level of noise data into the machine learning model for diagnosis, and output the noise diagnosis result.
[0049] like Figure 1 As shown, in a preferred embodiment, step S2 specifically includes the following contents:
[0050] S20, cutting out the first level of noise data;
[0051] Specifically, the first data segment is cut out according to the sudden change of sound pressure; the place where the noise data waveform has an obvious upward change is defined as the sudden change of sound pressure, and the critical point between the first and second gears is obtained according to the difference in the front and rear noise amplitudes; preferably, the place where the front and rear noise amplitude difference>2DB is selected as the critical point between the first and second gears ( Figure 1 Red frame);
[0052] S21, setting the cutting time of each gear;
[0053] S22, cut out the gear of subsequent noise data according to the set cutting time; conventionally, the default time of each gear is 5S, and the user can set the corresponding time according to the actual test environment;
[0054] like Figure 2 The figure shows a schematic diagram of the coupling of multiple noises in an industrial environment (different colors represent different types of noise in the current environment). Affected by on-site noise, equipment noise, etc., the collected original noise data has multiple environmental noise couplings. The low gear is mainly wind noise, and the high gear has some order characteristics, which can easily lead to the failure of acoustic-based segmentation. Secondly, industrial equipment needs to work in multiple speed gears due to different speed requirements, so fault analysis needs to be performed for each gear.
[0055] When segmenting a large amount of data with low signal-to-noise ratio or high signal-to-noise ratio but accompanied by a certain segment of signal NG, the above steps S20-S22 can effectively avoid segmentation failure and have a more independent and robust segmentation effect.
[0056] In a preferred embodiment, step S3 inputs the noise data after gear segmentation into a machine learning model, and the machine learning model is used to analyze the noise data, extract noise feature values and output noise diagnosis results; specifically, it includes the following contents:
[0057] S30, model building: perform basic acoustic analysis on existing noise data, extract multiple eigenvalues, and build an initial machine learning model;
[0058] S31, label data traversal: perform feature value traversal and model traversal on some of the original noise data with OK and NG labels, that is, extract the feature value of each original noise data, and input the extracted feature value into the initial machine learning model;
[0059] S32, model training and result prediction: Generate a PKL model file for fault diagnosis prediction through the initial machine learning model, and perform fault diagnosis prediction on the original noise data with OK and NG labels;
[0060] S33, prediction result labeling: labeling the fault diagnosis results predicted by the initial machine learning model with OK and NG labels;
[0061] S34, model error analysis: comparing the fault diagnosis results marked with OK and NG labels predicted by the initial machine learning model with the OK and NG labels of the original data itself, and obtaining the current model accuracy through error analysis;
[0062] S35, model accuracy optimization: perform model training and result prediction again for the erroneous noise data, i.e. repeat steps S31-S35;
[0063] S36, finally obtain the model parameters that meet the accuracy requirements, deploy the corresponding PKL model file in the corresponding diagnostic equipment, and use it for equipment fault diagnosis calculation in the actual environment.
[0064] In the above process, an accuracy threshold can be set, and the proportion of inconsistent label data to total training data can be calculated to characterize the model accuracy. If the current model accuracy is within the accuracy threshold, it means that the model accuracy meets the requirements. If the current model accuracy is not within the accuracy threshold, it means that the model accuracy does not meet the requirements.
[0065] Furthermore, the industrial equipment noise diagnosis method of the present application also includes the step of displaying the diagnosis result and / or giving an alarm. Preferably, the alarm can be given in the form of sound and light.
[0066] In a second aspect, the present application provides a diagnostic device for executing the above industrial equipment noise diagnostic method, comprising:
[0067] Data acquisition module, used to collect raw noise data of the working environment of industrial equipment;
[0068] A data processing module is used to perform gear segmentation on the original noise data according to the gear segmentation time point to obtain data segments of each gear;
[0069] The fault judgment module is used to diagnose the data segment of each file and obtain the diagnosis result.
[0070] Furthermore, the diagnostic device also includes a fault display module, which is electrically / communicatively connected to the fault judgment module and is used to display the diagnostic results in the form of OK and NG.
[0071] Furthermore, the diagnostic device also includes an alarm module, which is electrically / communicatively connected to the fault judgment module and is used to inform the user of the diagnosis result by sound and light (such as a warning light).
[0072] This equipment uses the acquisition module to collect raw noise data, gives the diagnosis results through the machine learning module, and displays the results in the form of OK and NG in the fault judgment module. It can also inform on-site staff through methods such as warning lights so that the staff can make timely judgments and repairs. It can cope with equipment noise diagnosis in various complex industrial environments and support real-time monitoring of equipment working status.
[0073] The diagnostic device of this embodiment stores a computer program, and the computer program is configured to execute the above industrial equipment noise diagnostic method when it is run. The computer readable storage medium can be a tangible device that can hold and store instructions used by the instruction execution device. The computer readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the above.
[0074] The computer program instructions used to perform the operations of the present application may be source code or object code written in any combination of one or more programming languages, and the computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server.
[0075] The above embodiments are only for illustrating the technical concept and features of the present application, and their purpose is to enable technicians familiar with the technical field to understand the content of the present application and implement it accordingly, and they cannot be used to limit the protection scope of the present application. For ordinary technicians in the technical field, they can make several improvements and modifications without departing from the principles of the present application, and these improvements and modifications should also be regarded as the protection scope of the present application.
Claims
1. A method for diagnosing industrial equipment noise based on machine learning, characterized in that: It includes the following: S1, collects the original noise data of the working environment of industrial equipment; S2, performing gear segmentation on the original noise data according to the gear segmentation time point; S3, construct a machine learning model, input the noise data after gear segmentation into the machine learning model, and the machine learning model is used to analyze the noise data, extract noise characteristic values and output noise diagnosis results.
2. The method for diagnosing industrial equipment noise based on machine learning according to claim 1, characterized in that: Step S2 specifically includes the following contents: S20, cut out the first level of noise data; S21, setting cutting time; S22, cutting out the subsequent noise data gear according to the set cutting time.
3. The method for diagnosing industrial equipment noise based on machine learning according to claim 2, characterized in that: In step S20, the first level of noise data is cut out according to the sudden change in sound pressure.
4. The method for diagnosing industrial equipment noise based on machine learning according to claim 1, characterized in that: Step S3 specifically includes the following contents: S30, perform basic acoustic analysis on existing noise data, extract multiple eigenvalues, and build an initial machine learning model; S31, perform feature value traversal and model traversal on the original noise data with OK and NG labels; S32, fault diagnosis prediction is performed on the original noise data with OK and NG labels through the initial machine learning model; S33, marking the fault diagnosis results predicted by the initial machine learning model with OK and NG labels; S34, comparing the fault diagnosis results marked with OK and NG labels predicted by the initial machine learning model with the OK and NG labels carried by the original data itself, and obtaining the model accuracy according to the errors between the two; S35, for the noise data with errors, repeat steps S31-S35 and perform model prediction again; S36, finally a machine learning model that meets the accuracy requirements is obtained and used for equipment fault diagnosis calculations in actual environments.
5. The method for industrial equipment noise diagnosis based on machine learning according to claim 1, characterized in that: The method also includes the step of displaying the diagnosis result.
6. The method for diagnosing industrial equipment noise based on machine learning according to claim 1, characterized in that: The method also includes the step of providing an alarm with sound and / or light based on the diagnosis result.
7. A diagnostic device for executing the industrial equipment noise diagnostic method according to any one of claims 1 to 6, characterized in that: include: Data acquisition module, used to collect raw noise data of the working environment of industrial equipment; A data processing module is used to perform gear segmentation on the original noise data according to the gear segmentation time point to obtain noise data segments of each gear; The fault judgment module is used to diagnose the noise data segment of each gear and obtain the diagnosis result.
8. The diagnostic device according to claim 7, characterized in that Also includes: The fault display module is electrically connected / communicationally connected to the fault judgment module, and is used to display the diagnosis result in the form of OK and NG.
9. The diagnostic device according to claim 7, characterized in that Also includes: The alarm module is electrically / communicatively connected to the fault judgment module and is used to alarm the diagnosis result in the form of sound and / or light.
10. The diagnostic device according to claim 7, characterized in that A computer program is stored in the diagnostic device, and the computer program is configured to execute the industrial equipment noise diagnostic method according to any one of claims 1 to 5 when running.