CN-116312562-B - Equipment noise detection model construction method and device based on industrial noise analysis technology

CN116312562BCN 116312562 BCN116312562 BCN 116312562BCN-116312562-B

Abstract

The application relates to a device noise detection model construction method and device based on an industrial noise analysis technology, and relates to the technical field of device noise detection; the method comprises the steps of processing sound sampling data to obtain sound restoration data, obtaining first volume information and first density information of sound of each frequency band of the sound restoration data in any time period, and constructing a device noise detection model according to the first volume information and the first density information. The application has the effect of finding the frequency band of the abnormal sound and the volume and density distribution relation of each frequency band.

Inventors

Sun Binyong
SUN BINQIANG
LIU LEI

Assignees

杭州云音超算智能科技有限责任公司

Dates

Publication Date: 20260505
Application Date: 20230327

Claims (7)

1. The method for constructing the equipment noise detection model based on the industrial noise analysis technology is characterized by comprising the following steps of: acquiring sound sample data in response to the request; Processing the sound sampling data to obtain sound restoration data; Acquiring first volume information and first density information of sound of each frequency band of the sound restoration data in any time period; constructing a device noise detection model according to the first volume information and the first density information; Processing the sound sampling data to obtain sound restoration data, wherein the processing comprises the following steps: Dividing a plurality of frequency segments in a preset frequency range; Acquiring second volume information and second density information of the sound sampling data in each frequency segment; Obtaining the sound restoration data according to the second volume information and the second density information; The acquiring the first volume information and the first density information of the sound of each frequency band of the sound restoration data in any time period specifically includes: Collecting attribute information of sound waves in the sound restoration data within a preset time range; Analyzing the attribute information to obtain first volume information and first density information of sound of each frequency band of sound of the sound wave in a second preset frequency range; The constructing a device noise detection model according to the first volume information and the first density information specifically includes: Classifying the sound restoration data according to a preset rule to obtain a classification result, wherein the classification result comprises periodic sound, continuous sound and sudden sound; and respectively constructing corresponding mathematical models for the periodic sound, the continuous sound and the sudden sound according to the first volume information and the first density information.
2. The method for constructing a device noise detection model based on an industrial noise analysis technique according to claim 1, wherein the mathematical model of the periodic sound includes: P_S=(T,f,v,s) wherein P_S is a periodic sound, T is a sounding period time, f is a sounding frequency, v is a volume, and S is a period stability.
3. The method for constructing a device noise detection model based on an industrial noise analysis technique according to claim 1, wherein the mathematical model of the continuous sound comprises: L_S=(f1,f2,T,t0,v,d) Wherein l_s is continuous sound, f1, f2 are pronunciation frequency, T is pronunciation period time, T0 is extraction interval time, v is volume, and d is density.
4. The method for constructing a device noise detection model based on an industrial noise analysis technique according to claim 1, wherein the mathematical model of the sudden sound comprises: S_S=(f1,f2,v,d,T,t0,v0,d0) Wherein s_s is a sudden sound, f1, f2 are sound frequencies, v is volume, d is density, T is sound period time, T0 is extraction interval time, v0 is volume threshold, and d0 is density threshold.
5. An apparatus noise detection model construction device based on an industrial noise analysis technology, which is characterized by comprising: a first acquisition module for acquiring sound sampling data in response to a request; the data processing module is used for processing the sound sampling data to obtain sound restoration data; The second acquisition module is used for acquiring first volume information and first density information of each frequency band sound of the sound restoration data in any time period; the model construction module is used for constructing a noise detection model according to the first volume information and the first density information; The method for processing the sound sampling data to obtain sound restoration data specifically comprises the following steps: Dividing a plurality of frequency segments in a preset frequency range; acquiring second volume information and second density information of sound sampling data in each frequency segment; Obtaining sound restoration data according to the second volume information and the second density information; The method for acquiring the first volume information and the first density information of the sound of each frequency band of the sound restoration data in any time period specifically comprises the following steps: collecting attribute information of sound waves in sound restoration data within a preset time range; Analyzing the attribute information to obtain first volume information and first density information of sound of each frequency band of sound of the sound wave in a second preset frequency range; Constructing a device noise detection model according to the first volume information and the first density information, wherein the device noise detection model specifically comprises the following steps: Classifying the sound restoration data according to a preset rule to obtain a classification result, wherein the classification result comprises periodic sound, continuous sound and sudden sound; And respectively constructing corresponding mathematical models for the periodic sound, the continuous sound and the sudden sound according to the first volume information and the first density information.
6. A terminal comprising a memory, a processor and a computer program stored in the memory and capable of running on the processor, characterized in that the processor performs the method of any of claims 1-4 when the computer program is loaded by the processor.
7. A computer readable storage medium having a computer program stored therein, characterized in that the computer program, when loaded by a processor, performs the method of any of claims 1-4.

Description

Equipment noise detection model construction method and device based on industrial noise analysis technology Technical Field The application relates to the technical field of equipment noise detection, in particular to an intelligent equipment noise detection method and device based on an industrial noise analysis technology. Background The common equipment operation condition can be distinguished by listening to the sound, and the vibration or noise generated in the equipment operation is mainly caused by the fact that abnormal conditions occur, so that the condition of some equipment can be inspected by listening to the sound, and whether the equipment fails or not is distinguished. The traditional Fourier analysis is based on the calculation that the sound in a certain frequency band has good periodic rule in analysis time, and the sound is completely inconsistent with the actual environmental sound because the industrial noise generally bursts, the attenuation is rapid, the duration is very short and the periodic rule is not generated. The traditional wavelet analysis is that the sound signals in the noisy environment are very complex, the sound signals have no consistent attenuation rule, and the wavelet analysis cannot accurately analyze all the sound signals by using a proper wavelet operator. Machine learning, namely, the generalization capability of machine learning is relatively poor, a large amount of abnormal sample data is required, the recognition effect is also not ideal, periodic sound rules cannot be found, indexes such as volume density of each frequency band cannot be accurately calculated, and the voice signals with relatively obvious characteristics are intelligently recognized at present. Therefore, the conventional analysis method cannot accurately analyze the frequency band of the abnormal sound and the distribution relation of the volume and density existing in each frequency band. Disclosure of Invention The application aims to provide a device noise detection model construction method and device based on an industrial noise analysis technology, wherein the device noise detection model construction method and device can find out frequency bands of abnormal sounds and the volume and density distribution relation of each frequency band. In a first aspect, the present application provides a method for constructing an equipment noise detection model based on an industrial noise analysis technology, which adopts the following technical scheme: A method for constructing a device noise detection model of an industrial noise analysis technology comprises the following steps: acquiring sound sample data in response to the request; Processing the sound sampling data to obtain sound restoration data; Acquiring first volume information and first density information of sound of each frequency band of the sound restoration data in any time period; And constructing a device noise detection model according to the first volume information and the first density information. By adopting the technical scheme, the noise detection is carried out on the equipment according to the constructed equipment noise detection model, so that the frequency band of the abnormal sound and the volume and density distribution relation of the sound in each frequency band can be found. Optionally, the processing the sound sampling data to obtain sound restoration data specifically includes: Dividing a plurality of frequency segments in a preset frequency range; Acquiring second volume information and second density information of the sound sampling data in each frequency segment; and obtaining the sound restoration data according to the second volume information and the second density information. By adopting the technical scheme, a plurality of frequency segments are divided, and the second volume information and the second density information of the sound sampling data are collected in each frequency segment, so that sound restoration data can be obtained according to the second volume information and the second density information. Optionally, the acquiring the first volume information and the first density information of the sound of each frequency band of the sound restoration data in any time period specifically includes: collecting attribute information of sound in the sound restoration data within a preset time range; And analyzing the attribute information to obtain first volume information and first density information of the sound of each frequency band in a preset frequency range. By adopting the technical scheme, the attribute information of the sound in the sound restoration data is acquired within the preset time range, and the first volume information and the first density information of the sound in each frequency range within the preset frequency range can be obtained according to the attribute information of the sound in the sound restoration data, so that the first volume information and the first den