Unsupervised machine abnormal sound detection method and device based on single classification algorithm

A detection method, an unsupervised technology, applied in computer parts, neural learning methods, instruments, etc., can solve problems such as unavailability, few abnormal sound samples of equipment, failure types of mechanical equipment failure rates, etc. The detection accuracy and sensitivity are improved, the experimental effect is improved, the mechanical and personal safety and the economic cost are guaranteed.

Pending Publication Date: 2022-05-10
NANJING UNIV OF POSTS & TELECOMM
View PDF1 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this invention does not take into account the low failure rate and variety of failures of mechanical equipment in actual factories, and does not solve the problem of few or even unavailable equipment abnormal sound samples during the classifier training process

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Unsupervised machine abnormal sound detection method and device based on single classification algorithm
  • Unsupervised machine abnormal sound detection method and device based on single classification algorithm
  • Unsupervised machine abnormal sound detection method and device based on single classification algorithm

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0044] figure 1 It is a schematic flowchart of an unsupervised machine abnormal sound detection method based on a single classification algorithm in Embodiment 1 of the present invention. see figure 1 , the detection method includes:

[0045] S1. Place a sound collection device next to the equipment to be tested in the operating state, and collect two types of machine sound and audio sample data in the normal operating state and in the abnormal operating state.

[0046] S2. Divide the collected audio sample data to generate two types of audio data sets, a training set and a test set. Wherein, the training set only includes the machine sound audio samples in the normal operation state, and the test set includes both the machine sound audio samples in the normal operation state and the abnormal operation state.

[0047] S3. Perform pre-processing operations such as pre-emphasis, framing, and windowing on the sample data in the training set to obtain several audio frames

[0...

example 1

[0053] In the actual test, the present invention has selected a plurality of fans (fan) in the fan factory to carry out the experiment, specifically selected 20 fans in operation, utilized the sound collection device to collect the sound of mechanical operation when they were in operation, and analyzed each The specific operating conditions of each machine are recorded, so as to judge whether the method of the present invention is good or bad for fault sound detection.

[0054] First, use matlab software to pre-emphasize the collected fan data set, select the pre-emphasis coefficient as 0.95, and divide the audio signal into frames and add windows in the time domain. According to the length of the audio, the frame length is set to 20ms, and the Hamming window is selected. , set the length of the window to 1024 points, and the single-hop size to 512 points. Then extract the corresponding characteristic parameters, and select the mean index, variance index, peak index, pulse ind...

example 2

[0056] Belt conveyors are widely used in coal, slag, powder and other production lines due to their advantages such as simple structure, large transportation capacity, strong adaptability, low energy consumption, and long service life. Due to the characteristics of large internal friction and wear of the transported material, some failures often occur in the actual work of the belt conveyor, such as belt deviation, belt tearing, wear, belt softening, scorching and even fire, causing casualties . Through long-term observation and research, it is found that the noise of the belt conveyor is generally very small when it is working normally, and a large abnormal noise will suddenly appear when a fault occurs, which provides inherently favorable conditions for abnormal sound detection. In the present invention, a certain number of auscultation sonars are deployed along the blower to collect the working noise emitted by the equipment, and after a period of sample collection, a train...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an unsupervised machine abnormal sound detection method and device based on a single classification algorithm, and the method comprises the steps: obtaining an audio sample of a to-be-detected device in a running state, a training set only containing an audio sample in a normal running state, and a test set containing audio samples in the normal running state and an abnormal running state at the same time; performing traditional feature extraction on the audio frames; solving by adopting a K-SVD algorithm and an OMP algorithm to obtain an updated dictionary and a training set sparse coefficient, and importing the updated dictionary and the training set sparse coefficient into a single-classification support vector machine classification model; and importing the test set sparse coefficient into the classification model for detection, and outputting a label category corresponding to the test set sparse coefficient. The problem that abnormal sound samples of machine equipment are few or even unavailable is solved, sparse representation and dictionary learning are carried out after feature extraction, compared with direct input of original features, the distinction degree between the features is increased, the accuracy of abnormal data detection is improved, and meanwhile the detection time is greatly shortened.

Description

technical field [0001] The invention relates to the technical field of auditory abnormality detection, in particular to an unsupervised machine abnormal sound detection method and device based on a single classification algorithm. Background technique [0002] Automatic detection of mechanical faults is an important technology in the fourth industrial revolution, and abnormal sound detection (ASD) is the task of identifying whether the sound emitted by the target machine is normal or abnormal. Sounds and vibrations emitted by equipment are important information, and quick detection of equipment anomalies by observing the sound may be useful for equipment status monitoring. [0003] With the rapid development of industrial technology, mechanized production has become the mainstream. The stable operation of mechanical equipment in the factory plays a very important role in the cost, efficiency, quality, and even safety of mechanical production. Early warning is of great signi...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/2411G06F18/214
Inventor 王晨苏新萍邵曦姚瑶
Owner NANJING UNIV OF POSTS & TELECOMM
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap