A mine ventilator motor fault warning device based on audio recognition

Through audio identification technology and advanced signal processing algorithms, the real-time and cost problems of motor fault detection of mine ventilation fans are solved, unattended remote monitoring and intelligent fault diagnosis are realized, and the safe and stable operation of the mine ventilation system is ensured.

CN114035047BInactive Publication Date: 2024-11-26HUAIHU COAL & ELECTRICITY CO LTD DINGJI COAL MINE +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111387365.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-22
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing technology has poor real-time performance in the detection of motor faults of mine ventilation fans, and relies on experienced detectors. The traditional vibration signal detection contact method is high and the cost is high, so it is impossible to achieve unattended and real-time monitoring.

Method used

Using a fault warning device based on audio recognition, the noise sensor and empirical modal decomposition denoising algorithm, improved Mel frequency cepspectral coefficients and hidden Markov model fault diagnosis algorithm are used to realize real-time audio signals through Wi-Fi module. Monitor and remote troubleshooting.

Benefits of technology

It realizes fast, accurate, unattended remote monitoring and intelligent fault diagnosis of mine ventilation motor faults, reduces hardware equipment requirements and costs, and improves real-time and reliability of detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114035047B_ABST
    Figure CN114035047B_ABST
Patent Text Reader

Abstract

The invention discloses a mine ventilator motor fault early warning device based on audio recognition. The device consists of an acquisition node, an analysis node and a host computer monitoring center. The first main control chip in the acquisition node controls the acquisition module to collect the audio signal of the mine ventilator motor in real time, and sends the collected data to the analysis node via a Wi-Fi module a after preprocessing; the second main control chip in the analysis node controls the analysis module to extract and analyze the data sent by the acquisition node, and sends the data after feature extraction and analysis to the host computer monitoring center through a serial communication module, so as to complete the remote monitoring, operation status display and intelligent fault early warning of the mine ventilator motor; the fault diagnosis algorithm based on the combination of empirical mode denoising, improved Mel frequency cepstrum coefficient and hidden Markov model can improve the accuracy of mine ventilator motor fault identification, and ensure the safe and stable operation of the underground ventilation system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field, and in particular relates to a mine ventilator motor fault early warning device based on audio recognition. Background Art

[0002] Mine ventilators are important equipment in mine production. They are responsible for replacing fresh air in the mine, exhausting harmful gases, diluting and exhausting toxic, asphyxiating, explosive gases and dust in the mine, and ensuring safe production in the mine. They are known as the "lungs of the mine". The mine ventilator motor is the "heart" of the mine ventilation system. Whether it can operate safely will directly affect the safety and stability of the coal mine system. Therefore, as equipment directly related to safe production, mine ventilators require a device to provide real-time early warning of ventilator motor failures to maintain the safety and stability of the coal mine system.

[0003] At present, most of the faults of mine ventilator motors are detected rather than monitored. The real-time detection is poor and requires experienced inspectors to conduct regular inspections and "listen" to the abnormal noise of the ventilator motor to judge the operating status of the ventilator motor. However, this method is greatly affected by the experience of the inspectors themselves, and can only be performed by inspectors on site, and cannot be unmanned and monitored in real time; the few monitoring methods are mainly to complete the real-time monitoring of vibration signals by embedding sensors inside, but the detection of vibration signals is contact-type, which has certain limitations and is costly and technically difficult. Therefore, regular inspections by "listening" and only using vibration signals as the basis for fault judgment of ventilator motors cannot accurately achieve real-time fault monitoring and fault warning.

[0004] The audio signal is a vibration signal inherent in the motor. In actual operation, the audio signal is stable when the fan motor is operating normally. When the fan motor fails, its audio signal will change significantly. Unlike the traditional "listening" type of regular inspection, the use of noise sensors can achieve uninterrupted and real-time monitoring of the motor audio signal; compared with the detection of vibration signals, the detection of audio signals is non-contact, easy to collect, small in amount of detection data, and simple in signal processing circuit. According to the different audio signals generated by the fan motor during normal operation and when a failure occurs, the motor audio signal is monitored in real time and the audio characteristics when a failure occurs are analyzed, so as to achieve the function of early warning of fan motor failure.

[0005] To this end, the present invention proposes a fault diagnosis algorithm based on an empirical mode decomposition denoising algorithm, an improved Mel frequency cepstral coefficient and a hidden Markov model, and determines the fault type of the mine ventilator motor through corresponding calculations. Summary of the invention

[0006] In view of the above-mentioned problems, the purpose of the present invention is to provide a mine ventilator motor fault warning device based on audio recognition, which can monitor and identify the noise emitted by the mine ventilator motor in real time, and realize remote monitoring, operation status display and intelligent fault diagnosis of the mine ventilator motor through preprocessing of audio signals, extraction of characteristic parameters and fault identification. It can quickly, accurately and effectively warn of the occurrence of mine ventilator motor failures, thereby ensuring the safe and stable operation of the underground ventilation system.

[0007] The technical solution of the present invention is: a mine ventilator motor fault early warning device based on audio recognition, characterized in that it includes a collection node, an analysis node and a host computer monitoring center;

[0008] The acquisition node includes a power module a, a first main control chip, a Wi-Fi module a and an acquisition module. The corresponding pins of the acquisition module and the Wi-Fi module a are respectively connected to the corresponding I / O ports of the first main control chip through wires. The power module a is respectively connected to the acquisition module, the first main control chip and the Wi-Fi module a through wires, and is used to supply power to the acquisition module, the first main control chip and the Wi-Fi module a;

[0009] The analysis node includes a power module b, a second main control chip, a Wi-Fi module b, an analysis module and a serial communication module, wherein the corresponding pins of the analysis module, the serial communication module and the Wi-Fi module b are respectively connected to the corresponding I / O ports of the second main control chip through wires, and the power module b is respectively connected to the analysis module, the serial communication module, the second main control chip and the Wi-Fi module b through wires, and is used to supply power to the analysis module, the serial communication module, the second main control chip and the Wi-Fi module b;

[0010] The collection node and the analysis node send and receive data via the interconnected Wi-Fi module a and Wi-Fi module b;

[0011] The host computer monitoring center is connected to the serial communication module via a serial port line, and is used to receive data sent by the analysis node to obtain relevant data and fault identification results for real-time monitoring of the audio signal of the mine ventilator motor, thereby achieving functions such as remote monitoring of the mine ventilator motor, operating status display and intelligent fault diagnosis.

[0012] Furthermore, in the acquisition node, the first main control chip controls the acquisition module to monitor the audio signal of the mine ventilator motor in real time. All collected data are pre-processed by the empirical mode decomposition denoising algorithm through the first main control chip and then transmitted to the analysis node via Wi-Fi module a.

[0013] Furthermore, it is characterized in that the acquisition module uses a noise sensor, which has the characteristics of small size, easy placement, high sensitivity, etc.

[0014] Furthermore, it is characterized in that, in the analysis node, the second main control chip controls the Wi-Fi module b to receive the audio signal sent by the collection node, and the second main control chip of the analysis node controls the analysis module to diagnose the fault type of the received audio signal, and sends the diagnosis result to the host computer monitoring center through the serial communication module.

[0015] Further, it is characterized in that the algorithm used in the fault type diagnosis is a fault diagnosis algorithm based on empirical mode decomposition denoising algorithm, improved Mel frequency cepstrum coefficient and hidden Markov model. The fault type diagnosis consists of audio signal preprocessing and fault identification. First, the audio signal of the ventilator motor collected by the acquisition node is pre-emphasized, framed and windowed, endpoint detected and empirical mode denoising preprocessed, and then the pre-processed audio signal is feature extracted by improved Mel frequency cepstrum coefficient, and then the extracted characteristic parameters are put into the built fault model library as the input fault sound signal to be identified for analysis and calculation to obtain the output probability of the fault model, thereby diagnosing and determining the fault type of the mine ventilator motor. The empirical mode decomposition denoising algorithm is essentially the adaptive decomposition of non-stationary signals using basis functions. Different from denoising methods such as Fourier transform, wavelet denoising, and wavelet packet denoising, empirical mode decomposition has the advantages of flexibility of basis functions, variable multi-resolution and adaptive filtering characteristics; its basis function originates from the inside of the signal, not artificially specified by the outside world, so different basis functions are used for different signals. The improved Mel-frequency cepstral coefficient method can achieve better recognition effect by extracting Mel-frequency cepstral coefficients twice to obtain characteristic parameters reflecting fault signals, thereby achieving the purpose of improving system recognition performance. The hidden Markov model has dual randomness and can accurately describe audio signals. At the same time, as a probability statistical model, it can directly use probability statistical density functions for probability calculation, thereby calculating the output probability of the fault model, and the largest output is the recognition result. By utilizing the advantages of the above three, the effects of audio signal preprocessing, feature extraction and fault identification are greatly improved, which plays an important role in ensuring the normal operation of the fan motor and finding the cause of the fault as soon as possible when a fault occurs.

[0016] Furthermore, it is characterized in that the first main control chip in the acquisition node and the second main control chip in the analysis node are both STM32F103ZET6 chips, which have the characteristics of high integration, low energy consumption and good real-time performance.

[0017] The working method of the present invention is as follows: the acquisition node is placed around the mine ventilator motor, and the power module a in the acquisition node provides power to the first main control chip, the Wi-Fi module a and the acquisition module; after the first main control chip is powered on and initialized, when the ventilator motor is working, the first main control chip controls the acquisition module to collect the audio signal of the ventilator motor in real time, and sends the audio signal data to the analysis node through the Wi-Fi module a after preprocessing;

[0018] The analysis node is placed in the motor monitoring room. The power module b in the analysis node provides power to the second main control chip, Wi-Fi module b, analysis module and serial communication module. After the second main control chip is powered on and initialized, it controls the Wi-Fi module b to receive the data sent by the acquisition node, and controls the analysis module to extract and analyze the received audio signals. The second main control chip sends the analysis data results to the host computer monitoring center through the serial communication module.

[0019] After the host computer monitoring center receives the analysis data results sent from the analysis node, it can display them in real time on the computer side, realizing functions such as remote monitoring of mine ventilator motors, real-time display of operating status, and intelligent fault warning.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] 1. The present invention adopts sensor technology, modern digital signal processing technology and Internet of Things technology to achieve a more accurate and reliable monitoring method than the traditional "ear listening method", and can actually realize remote monitoring and real-time monitoring that cannot be achieved in traditional detection;

[0022] 2. The present invention adopts audio signal monitoring. Compared with the traditional monitoring of vibration signals, the monitoring of audio signals is non-contact and convenient to collect. The amount of audio detection data is small, the signal processing circuit is simple, the hardware equipment requirements are low, and the cost is low;

[0023] 3. The present invention adopts the empirical mode decomposition denoising algorithm for preprocessing audio signals, which does not require external personnel interference, is more flexible and reliable, has strong adaptability, is simple to operate, and has stable and reliable denoising effect;

[0024] 4. The present invention adopts an improved Mel-frequency cepstral coefficient to extract the features of the audio signal, and obtains the parameters reflecting the dynamic characteristics of the audio by extracting the Mel-frequency cepstral coefficient twice;

[0025] 5. The present invention uses a hidden Markov model to diagnose the fault type of the mine ventilator motor, matches and analyzes the input fault audio signal to be identified with the fault model library, and calculates the output probability of each model as the identification result. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a schematic diagram of the overall structure of the present invention;

[0027] Figure 2 It is a structural diagram of the acquisition node of the present invention;

[0028] Figure 3 It is the analysis node structure diagram of the present invention;

[0029] Figure 4 It is a schematic diagram of the principle block diagram of the fault diagnosis algorithm of the present invention.

[0030] Among them, 101-power module a, 102-power module b, 201-first main control chip, 202-second main control chip, 301-Wi-Fi module a, 302-Wi-Fi module b, 4-acquisition module, 5-analysis module, 6-serial communication module. DETAILED DESCRIPTION

[0031] The specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific implementation modes.

[0032] Example: Figure 1 As shown, a mine ventilator motor fault warning device based on audio recognition includes a collection node, an analysis node and a host computer monitoring center.

[0033] The acquisition node includes a power module a101, a first main control chip 201, a Wi-Fi module a301 and an acquisition module 4. The acquisition module 4 uses a noise sensor. The corresponding pins of the acquisition module 4 and the Wi-Fi module a301 are respectively connected to the corresponding I / O ports of the first main control chip 201 through wires. The power module a101 is respectively connected to the acquisition module 4, the first main control chip 201 and the Wi-Fi module a301 through wires, and is used to supply power to the acquisition module 4, the first main control chip 201 and the Wi-Fi module a301. The first main control chip 201 controls the acquisition module 4 to monitor the audio signal of the mine ventilator motor in real time. All the collected data are pre-processed by the empirical mode decomposition denoising algorithm through the first main control chip 201 and then transmitted to the analysis node through the Wi-Fi module a301. The first main control chip 201 uses the STM32F103ZET6 chip;

[0034] The analysis node includes a power module b102, a second main control chip 202, a Wi-Fi module b302, an analysis module 5 and a serial communication module 6. The corresponding pins of the analysis module 5, the serial communication module 6 and the Wi-Fi module b302 are respectively connected to the corresponding I / O ports of the second main control chip 202 through wires. The power module b102 is respectively connected to the analysis module 5, the serial communication module 6, the second main control chip 202 and the Wi-Fi module b302 through wires, and is used to supply power to the analysis module 5, the serial communication module 6, the second main control chip 202 and the Wi-Fi module b302. The second main control chip 202 controls the Wi-Fi module b302 to receive the audio signal sent by the acquisition node. The second main control chip 202 of the analysis node controls the analysis module 5 to diagnose the fault type of the received audio signal, and sends the diagnosis result to the upper computer monitoring center through the serial communication module 6. The second main control chip 202 uses the STM32F103ZET6 chip;

[0035] The algorithm used for fault type diagnosis is a fault diagnosis algorithm based on empirical mode decomposition denoising algorithm, improved Mel frequency cepstral coefficients and hidden Markov model. Fault type diagnosis consists of audio signal preprocessing and fault identification. First, the audio signal of the ventilator motor collected by the acquisition node is pre-emphasized, framed and windowed, endpoint detected and empirical mode denoising preprocessed, and then the preprocessed audio signal is feature extracted by improved Mel frequency cepstral coefficients. Then, the extracted feature parameters are used as the input fault sound signal to be identified and put into the established fault model library for analysis and calculation to obtain the output probability of the fault model, thereby diagnosing and determining the fault type of the mine ventilator motor;

[0036] The data collection node and the analysis node are sent and received via the interconnected Wi-Fi module a301 and Wi-Fi module b302;

[0037] The host computer monitoring center is connected to the serial communication module 6 via a serial port line to receive the data sent by the analysis node.

[0038] The working method of the above embodiment is: the collection node is placed around the mine ventilator motor, and the power module a101 in the collection node provides power to the first main control chip 201, the Wi-Fi module a301 and the collection module 4; after the first main control chip 201 is powered on and initialized, when the ventilator motor is working, the first main control chip 201 controls the collection module 4 to collect the audio signal of the ventilator motor in real time, and sends the audio signal data to the analysis node through the Wi-Fi module a301 after preprocessing;

[0039] The analysis node is placed in the motor monitoring room, and the power module b102 in the analysis node provides power to the second main control chip 202, the Wi-Fi module b302, the analysis module 5 and the serial communication module 6; after the second main control chip 202 is powered on and initialized, the Wi-Fi module b302 is controlled to receive the data sent by the acquisition node, and the analysis module 5 is controlled to extract and analyze the received audio signal. The second main control chip 202 sends the analysis data results to the host computer monitoring center through the serial communication module 6;

[0040] After the host computer monitoring center receives the analysis data results sent from the analysis node, it can display them in real time on the computer side, realizing functions such as remote monitoring of mine ventilator motors, real-time display of operating status, and intelligent fault warning.

[0041] The specific models of the above electronic components are not specially specified, and ordinary products available on the market can be selected as long as they can meet the use requirements of the present invention.

[0042] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and does not limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included in the protection scope of the present invention.

Claims

1. A mine ventilator motor fault warning device based on audio recognition, It is characterized in that Includes collection nodes, analysis nodes and host computer monitoring center; The acquisition node comprises a power module a (101), a first main control chip (201), a Wi-Fi module a (301) and an acquisition module (4); the corresponding pins of the acquisition module (4) and the Wi-Fi module a (301) are respectively connected to the corresponding I / O ports of the first main control chip (201) through wires; the power module a (101) is respectively connected to the acquisition module (4), the first main control chip (201) and the Wi-Fi module a (301) through wires, and is used to supply power to the acquisition module (4), the first main control chip (201) and the Wi-Fi module a (301); The analysis node comprises a power module b (102), a second main control chip (202), a Wi-Fi module b (302), an analysis module (5) and a serial communication module (6); the corresponding pins of the analysis module (5), the serial communication module (6) and the Wi-Fi module b (302) are respectively connected to the corresponding I / O ports of the second main control chip (202) through wires; the power module b (102) is respectively connected to the analysis module (5), the serial communication module (6), the second main control chip (202) and the Wi-Fi module b (302) through wires, and is used to supply power to the analysis module (5), the serial communication module (6), the second main control chip (202) and the Wi-Fi module b (302); The collection node and the analysis node send and receive data via the interconnected Wi-Fi module a (301) and Wi-Fi module b (302); The host computer monitoring center is connected to the serial communication module (6) via a serial port line, and is used to receive data sent by the analysis node; In the acquisition node, the first main control chip (201) controls the acquisition module (4) to monitor the audio signal of the mine ventilator motor in real time, and all the collected data are pre-processed by the first main control chip (201) using an empirical mode decomposition denoising algorithm and then transmitted to the analysis node via the Wi-Fi module a (301); The acquisition module (4) uses a noise sensor; In the analysis node, the second main control chip (202) controls the Wi-Fi module b (302) to receive the audio signal sent by the collection node, and the second main control chip (202) of the analysis node controls the analysis module (5) to perform fault type diagnosis on the received audio signal, and sends the diagnosis result to the upper computer monitoring center through the serial communication module (6); The algorithm used in the fault type diagnosis is a fault diagnosis algorithm based on empirical mode decomposition denoising algorithm, improved Mel frequency cepstral coefficients and hidden Markov model. The fault type diagnosis consists of audio signal preprocessing and fault identification. First, the audio signal of the ventilator motor collected by the acquisition node is pre-emphasized, framed and windowed, endpoint detected and empirical mode denoising preprocessed, and then the preprocessed audio signal is feature extracted by improved Mel frequency cepstral coefficients. Then, the extracted feature parameters are used as input to the fault sound signal to be identified and put into the established fault model library for analysis and calculation to obtain the output probability of the fault model, thereby diagnosing and determining the fault type of the mine ventilator motor; The first main control chip (201) in the collection node and the second main control chip (202) in the analysis node are both STM32F103ZET6 chips.