CN-122016315-A - Early degradation detection method and system for coal cutter bearing by adopting pulse neural network

CN122016315ACN 122016315 ACN122016315 ACN 122016315ACN-122016315-A

Abstract

The invention relates to the technical field of fault diagnosis of coal mining machines, in particular to a method and a system for detecting early degradation of a coal mining machine bearing by adopting a pulse neural network. The method comprises the steps of collecting acoustic signals of the bearing in real time, carrying out blind source separation and band-pass filtering pretreatment, converting the pretreated signals into logarithmic Mel time-frequency spectrograms, encoding the logarithmic Mel time-frequency spectrograms into pulse event sequences through peak detection, inputting the pulse sequences into a pulse neural network model comprising a pulse convolution layer and a pulse recurrent neural network layer, extracting state characteristics, and judging whether the bearing is degraded early or not based on classification or deviation calculation. The invention realizes the sensitive and accurate identification of the early degradation weak characteristics of the bearing under the strong noise environment by simulating the biological hearing and nerve processing mechanism and utilizing the characteristics of high sensitivity and low power consumption of the pulse neural network to the time sequence signals and combining the targeted noise reduction and coding, and is particularly suitable for the intelligent edge monitoring of underground equipment.

Inventors

ZHAO QINGLIANG
DING QIANG
ZHANG YUN
LI JINLONG
ZHANG ZHONGCUN
CHEN WAN
LIU FENG
YANG ZHIQIANG
BAI WENYING
LU YING
WANG FEN

Assignees

山东能源装备集团天地采掘设备再制造有限公司

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. The early degradation detection method for the coal cutter bearing by adopting the pulse neural network is characterized by comprising the following steps of: s1, acquiring an original acoustic signal of a coal cutter shaft operation part in real time, and preprocessing the original acoustic signal to obtain a preprocessed audio signal; s2, encoding the preprocessed audio signal into a pulse event sequence; s3, inputting the pulse event sequence into a pre-trained pulse neural network model for processing to obtain state characteristic data of the bearing; And S4, judging whether the bearing of the coal mining machine is degraded early or not according to the state characteristic data.
2. The method for detecting early degradation of a bearing of a coal mining machine according to claim 1, wherein the preprocessing in step S1 specifically includes: And carrying out band-pass filtering on the separated source signals, and reserving signal components of a preset frequency band to obtain the preprocessed audio signals.
3. The method for early degradation detection of a shearer bearing according to claim 2, wherein the preset frequency band is a characteristic frequency range which is predetermined based on a natural frequency of the shearer bearing and harmonic components thereof and is sensitive to bearing degradation.
4. The method for detecting early degradation of a bearing of a coal mining machine according to claim 1, wherein the step S2 specifically includes: The method comprises the steps of preprocessing an audio signal, converting the preprocessed audio signal into a time-frequency spectrogram, detecting energy peaks in time and frequency dimensions in the time-frequency spectrogram, encoding each detected energy peak into a pulse event, and generating the pulse event sequence in time sequence.
5. The method of claim 4, wherein the time-frequency spectrum is a logarithmic Mel spectrum, and the encoding as a pulse event is to generate a pulse at a time point and a corresponding frequency channel when the energy value of a frequency unit in a continuous time frame exceeds the energy values of adjacent time frames and frequency units and reaches a preset threshold.
6. The method for detecting early degradation of a bearing of a coal mining machine according to claim 1, wherein the pulse neural network model comprises a pulse convolution layer and a pulse recurrent neural network layer which are sequentially connected, wherein the pulse convolution layer is used for extracting spatial features related to bearing degradation from an input pulse event sequence, and the pulse recurrent neural network layer is used for capturing time sequence dependency of the spatial features evolving along with time and outputting state feature data.
7. The method for detecting early degradation of a shearer bearing of claim 6, wherein the pulse recurrent neural network layer is a pulse long and short term memory network layer.
8. The method for detecting early degradation of a bearing of a coal mining machine according to claim 1, wherein in the step S4, it is determined whether early degradation of the bearing of the coal mining machine occurs, specifically: And (3) inputting the state characteristic data into a classifier to obtain a classification result of the bearing in a healthy state or an early degradation state, or calculating the deviation degree between the state characteristic data and the standard characteristic data of the healthy bearing, and judging that the early degradation occurs if the deviation degree continuously exceeds a preset threshold value.
9. The method for detecting early degradation of a coal cutter bearing according to claim 8, wherein the classifier is a pulse neural layer or a full-connection layer, and the deviation degree is obtained by calculating Euclidean distance, cosine similarity or Markov distance.
10. A coal mining machine bearing early degradation detection system employing a pulsed neural network for implementing the detection method of any one of claims 1-9, comprising: The signal acquisition and preprocessing module is used for acquiring an original acoustic signal of a coal mining machine shaft operation part in real time, and preprocessing the original acoustic signal to obtain a preprocessed audio signal; the pulse coding module is used for coding the preprocessed audio signal into a pulse event sequence; The pulse neural network processing module is integrated with a pre-trained pulse neural network model, and is used for processing the pulse event sequence and outputting state characteristic data of the bearing; and the degradation judging module is used for judging whether the coal cutter bearing is degraded early or not according to the state characteristic data.

Description

Early degradation detection method and system for coal cutter bearing by adopting pulse neural network Technical Field The invention relates to the technical field of fault diagnosis of coal mining machines, in particular to a method and a system for detecting early degradation of a coal mining machine bearing by adopting a pulse neural network. Background The reliability of key components (such as a cutting part and a traction part bearing) of the coal mining machine serving as core equipment of a fully-mechanized coal mining face is directly related to production safety and efficiency. The bearing can gradually generate degradation such as abrasion, fatigue, peeling and the like when running for a long time under severe working conditions, and if not found in time, the bearing can cause sudden failure, so that production stopping and even safety accidents are caused. Therefore, the early degradation detection of the bearing of the coal mining machine and the predictive maintenance are implemented, so that the method has great significance. Currently, a vibration signal analysis-based method is a mainstream technology of bearing state monitoring, and is characterized by installing an acceleration sensor to collect signals and analyzing the frequency spectrum, the envelope spectrum and the like of the signals so as to identify faults. However, the method has significant challenges in the application of the coal mining machine, namely firstly, the working environment of the coal mining machine is extremely complicated in noise, strong background noise such as impact generated by cutting coal and rock, vibration of other rotating parts and the like are extremely easy to submerge weak vibration characteristics generated by early degradation of the bearing, so that report missing is caused, secondly, the traditional vibration analysis usually depends on expert experience to set a threshold value or a fixed frequency band, is insensitive to early and atypical degradation modes and early warning is delayed, and furthermore, the intelligent diagnosis method based on deep learning (such as CNN and RNN) can automatically extract characteristics, but is a computationally intensive 'static' network, has low processing efficiency and high power consumption on continuous time sequence signals, and is difficult to meet the edge calculation requirements of underground equipment on real-time performance and low power consumption. In recent years, a pulse neural network is used as a novel calculation model imitating a biological neuron information processing mechanism, has the natural advantages of event driving, sparse calculation and extremely low power consumption due to high sensitivity to time sequence signals, and has potential in the fields of voice recognition, dynamic vision and the like, but has not yet been applied to early degradation early warning of mechanical parts in the acoustic monitoring of industrial equipment, particularly in the complex noise background. Therefore, there is a need to develop a new intelligent detection method that can acutely capture early signs of degradation from strong noise and is suitable for edge deployment. Thus, the prior art is still to be further developed. Disclosure of Invention The invention aims to overcome the technical defects and provide a method and a system for detecting early degradation of a bearing of a coal mining machine by adopting a pulse neural network so as to solve the problems in the prior art. To achieve the above technical object, according to a first aspect of the present invention, there is provided a method for detecting early degradation of a bearing of a coal mining machine using a pulse neural network, comprising: s1, acquiring an original acoustic signal of a coal cutter shaft operation part in real time, and preprocessing the original acoustic signal to obtain a preprocessed audio signal; s2, encoding the preprocessed audio signal into a pulse event sequence; s3, inputting the pulse event sequence into a pre-trained pulse neural network model for processing to obtain state characteristic data of the bearing; And S4, judging whether the bearing of the coal mining machine is degraded early or not according to the state characteristic data. Specifically, the preprocessing in step S1 specifically includes: And carrying out band-pass filtering on the separated source signals, and reserving signal components of a preset frequency band to obtain the preprocessed audio signals. Specifically, the preset frequency band is a characteristic frequency range which is predetermined based on the natural frequency of the bearing of the coal mining machine and harmonic components thereof and is sensitive to the degradation of the bearing. Specifically, the step S2 specifically includes: The method comprises the steps of preprocessing an audio signal, converting the preprocessed audio signal into a time-frequency spectrogram, detecting energy peaks in time and freq