CN112710486A - Equipment fault detection method, equipment fault detection device and computer storage medium - Google Patents
Equipment fault detection method, equipment fault detection device and computer storage medium Download PDFInfo
- Publication number
- CN112710486A CN112710486A CN201911019720.1A CN201911019720A CN112710486A CN 112710486 A CN112710486 A CN 112710486A CN 201911019720 A CN201911019720 A CN 201911019720A CN 112710486 A CN112710486 A CN 112710486A
- Authority
- CN
- China
- Prior art keywords
- fault
- spectrogram
- learning model
- fault type
- sound signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M99/00—Subject matter not provided for in other groups of this subclass
- G01M99/005—Testing of complete machines, e.g. washing-machines or mobile phones
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H9/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves by using radiation-sensitive means, e.g. optical means
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M99/00—Subject matter not provided for in other groups of this subclass
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The application discloses a method for detecting equipment faults, which comprises the following steps: emitting a measuring beam to the target device and collecting a detection beam of the measuring beam reflected by the target device; extracting a vibration signal from the detected light beam; the vibration signal is input to a trained fault learning model to identify a fault type associated with the vibration signal to determine a fault type of the target device. The application also discloses a corresponding device fault detection device and a computer storage medium. In this way, adopt non-contact vibration signal measurement scheme on the one hand, need not external syllable-dividing room, high-efficient check out test set trouble problem, on the other hand when big batch equipment quality control, save time and human cost, improve quality control efficiency.
Description
Technical Field
The present disclosure relates to the field of fault detection, and in particular, to a method and an apparatus for detecting a device fault and a computer storage medium.
Background
The equipment can generate various sounds in the operation process, wherein one part of the sounds are the sounds generated by the equipment in a normal operation state, and the other part of the sounds are abnormal sounds generated by the equipment in a fault condition and the like. When abnormal sound occurs, the abnormal sound generally indicates that the equipment has a fault and needs to be maintained.
However, the current method for detecting the equipment failure is quite backward, for example, a method of artificial listening discrimination is adopted to judge whether the equipment is failed at that time. For example, before the equipment is shipped, quality sampling inspection is performed, an independent sound insulation room needs to be established, a large amount of equipment needs to be detected, the detection efficiency is low, and the implementation cost is high. In some cases, before leaving the factory, the device needs to be subjected to destructive quality sampling inspection, for example, detection of a welding spot, and the sampled sample needs to be subjected to knocking destruction to check whether the quality of the welding spot is qualified, and this method is also low in detection efficiency and high in implementation cost.
Disclosure of Invention
In order to solve the above problems, the present application provides an equipment failure detection method, an equipment failure detection apparatus, and a computer storage medium, which adopt a non-contact vibration signal measurement scheme, do not need an external sound insulation room, and efficiently detect the problem of equipment failure, and on the other hand, save time and labor cost and improve quality inspection efficiency during quality inspection of large-scale equipment.
The technical scheme adopted by the application is to provide an equipment fault detection method, which comprises the following steps: emitting a measuring beam to the target device and collecting a detection beam of the measuring beam reflected by the target device; extracting a vibration signal from the detected light beam; the vibration signal is input to a trained fault learning model to identify a fault type associated with the vibration signal to determine a fault type of the target device.
The trained fault learning model is obtained by training based on vibration signal sample data and a predetermined fault type label.
Wherein inputting the vibration signal to a trained fault learning model to identify a fault type associated with the vibration signal comprises: converting the vibration signal into a sound signal; inputting the voice signal to a trained fault learning model to identify a fault type associated with the voice signal to determine a fault type of the target device; the trained fault learning model is obtained by training based on the sound signal sample data and a predetermined fault type label.
Wherein inputting the voice signal to a trained fault learning model to identify a fault type associated with the voice signal to determine a fault type of the target device comprises: preprocessing the sound signal; and inputting the sound signal after preprocessing into the trained fault learning model to identify the fault type associated with the sound signal so as to determine the fault type of the target equipment.
Wherein, carry out the preliminary treatment to the sound signal, include: pre-emphasis the sound signal to compensate for high frequency components in the sound signal; and performing frame division and windowing processing on the sound signal by adopting a preset window function to obtain the sound signal after preprocessing.
Wherein inputting the vibration signal to a trained fault learning model to identify a fault type associated with the vibration signal comprises: converting the vibration signal into a sound signal; preprocessing the sound signal; converting the preprocessed sound signals into spectrogram; inputting the spectrogram into a trained fault learning model to identify a fault type associated with the spectrogram, thereby determining a fault type of the target device; the trained fault learning model is obtained by training based on spectrogram sample data and a predetermined fault type label.
Wherein inputting the spectrogram into the trained fault learning model to identify a fault type associated with the spectrogram, comprises: extracting time information, frequency information and energy information of the spectrogram to obtain to-be-processed characteristic information; inputting the characteristic information to be processed into the trained fault learning model so as to identify the fault type associated with the spectrogram.
Wherein inputting the spectrogram into the trained fault learning model to identify a fault type associated with the spectrogram, comprises: inputting the spectrogram into a trained fault learning model; carrying out region blocking on the spectrogram by using a fault learning model, and distributing corresponding weights to different regions to obtain a blocked spectrogram to be processed; carrying out weighting processing on the to-be-processed block spectrogram by using a fault learning model so as to obtain a weighted similarity comparison result; and determining the fault type associated with the sound signal as the fault type corresponding to the spectrogram in response to the weighted similarity comparison result being larger than a set threshold.
Wherein, to target device emission measuring beam, and gather measuring beam and before the detection light beam that target device reflects, still include: splitting a laser beam into a measurement beam and a reference beam; extracting a vibration signal from a detected light beam, comprising: and interfering the detection beam and the reference beam to detect and obtain a vibration signal.
Wherein the target device includes a motor, and the target device is operated by the motor to generate vibration.
Wherein, to target device emission measuring beam, and gather measuring beam and before the detection light beam that target device reflects, still include: and connecting the target device through the external driving device to drive the target device to generate vibration.
Another technical solution adopted by the present application is to provide an apparatus fault detection device, including: the laser transmitter is used for transmitting a measuring beam to the target equipment; the laser receiver is used for collecting a detection light beam reflected by the target device through the measuring light beam; a processor connected with the laser receiver and used for extracting vibration signals from the detection light beams and inputting the vibration signals to the trained fault learning model so as to identify fault types associated with the vibration signals and determine the fault types of the target equipment; the trained fault learning model is obtained by training based on vibration signal sample data and a predetermined fault type label.
The processor is further used for converting the vibration signal into a sound signal and inputting the sound signal into the trained fault learning model so as to identify a fault type associated with the sound signal, and therefore the fault type of the target equipment is determined; the trained fault learning model is obtained by training based on the sound signal sample data and a predetermined fault type label.
The processor is further used for preprocessing the sound signal and inputting the preprocessed sound signal into the trained fault learning model to identify the fault type associated with the sound signal so as to determine the fault type of the target device.
The processor is further used for converting the vibration signal into a sound signal, preprocessing the sound signal, converting the preprocessed sound signal into a spectrogram, and inputting the spectrogram into the trained fault learning model to identify a fault type associated with the spectrogram; the trained fault learning model is obtained by training based on spectrogram sample data and a predetermined fault type label.
The processor is further used for extracting time information, frequency information and energy information of the spectrogram to obtain to-be-processed characteristic information; inputting the characteristic information to be processed into the trained fault learning model so as to identify the fault type associated with the spectrogram.
Wherein the processor is further configured to input the spectrogram into a trained fault learning model; carrying out region blocking on the spectrogram by using a fault learning model, and distributing corresponding weights to different regions to obtain a blocked spectrogram to be processed; carrying out weighting processing on the to-be-processed block spectrogram by using a fault learning model so as to obtain a weighted similarity comparison result; and determining the fault type associated with the sound signal as the fault type corresponding to the spectrogram in response to the weighted similarity comparison result being larger than a set threshold.
The processor is connected with the laser transmitter and is used for dividing the laser beam emitted by the laser transmitter into a measuring beam and a reference beam and interfering the detecting beam and the reference beam to detect and obtain a vibration signal.
Another technical solution adopted by the present application is to provide a computer storage medium for storing program data, which when executed by a processor, is used to implement any one of the methods provided in the above-mentioned solution.
The beneficial effect of this application is: different from the prior art, the device fault detection method of the present application includes: emitting a measuring beam to the target device and collecting a detection beam of the measuring beam reflected by the target device; extracting a vibration signal from the detected light beam; the vibration signal is input to a trained fault learning model to identify a fault type associated with the vibration signal to determine a fault type of the target device. In this way, adopt non-contact vibration signal measurement scheme on the one hand, need not external syllable-dividing room, high-efficient check out test set trouble problem, on the other hand when big batch equipment quality control, save time and human cost, improve quality control efficiency.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts. Wherein:
fig. 1 is a schematic flowchart of a first embodiment of an apparatus fault detection method provided in the present application;
FIG. 2 is a schematic structural diagram of an embodiment of an apparatus for detecting device failure provided in the present application;
FIG. 3 is a schematic diagram of an application of the device failure detection method provided in the present application;
FIG. 4 is a schematic flow chart diagram illustrating a second embodiment of a method for detecting device failure provided by the present application;
FIG. 5 is a schematic flow chart diagram illustrating a third embodiment of a device failure detection method provided by the present application;
FIG. 6 is a schematic flow chart diagram illustrating a fourth embodiment of the device failure detection method provided in the present application;
FIG. 7 is a schematic flow chart diagram illustrating a fifth embodiment of a method for detecting device failure according to the present application;
fig. 8 is a schematic diagram of weighting processing in the device fault detection method provided in the present application;
fig. 9 is a schematic structural diagram of an embodiment of the device fault detection apparatus provided in the present application;
fig. 10 is a schematic structural diagram of another embodiment of the device failure detection apparatus provided in the present application.
FIG. 11 is a schematic structural diagram of an embodiment of a computer storage medium provided in the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. It is to be understood that the specific embodiments described herein are merely illustrative of the application and are not limiting of the application. It should be further noted that, for the convenience of description, only some of the structures related to the present application are shown in the drawings, not all of the structures. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.
Referring to fig. 1, fig. 1 is a schematic flow chart of a first embodiment of an apparatus fault detection method provided in the present application, where the method includes:
step 11: the measuring beam is emitted toward the target device, and a detection beam reflected by the measuring beam via the target device is collected.
In some embodiments, the measuring beam is emitted by a laser emitter in the device failure detection apparatus to the target device, a laser reflector is mounted on the target device for reflecting the detecting beam after receiving the measuring beam, and the detecting beam is received by a laser receiver.
Taking the target device as an air conditioner as an example, when an outdoor unit of the air conditioner is assembled, quality detection is performed, the outdoor unit of the air conditioner is electrified to enable the outdoor unit to work, the laser reflector is installed at a preset position of the outdoor unit, and the position can be changed differently according to user requirements. The equipment fault detection device transmits a measuring beam to the laser reflector, the measuring beam is reflected by the laser reflector to generate a detection beam, and the detection beam is received by a laser receiver of the equipment fault detection device. The outdoor unit of the air conditioner is in a working state at this time.
In some embodiments, a laser transmitter in the device failure detection apparatus transmits a measurement beam to the target device, the measurement beam is reflected by a surface of the target device to generate a detection beam, and the detection beam is received by a laser receiver.
In some embodiments, the equipment failure detection device of fig. 2 is used for failure detection, and the equipment failure detection device 20 includes a laser transmitter 21, a laser receiver 22, a beam splitter 23, a reference beam reflector 24, and a measurement beam reflector 25.
The laser emitter 21 emits he-ne laser, and the he-ne laser is divided into a reference beam and a measuring beam by the beam splitter 23, the reference beam is reflected by the reference beam reflector 24 and then enters the laser receiver 22 through the beam splitter 23, the measuring beam is reflected by the measuring beam reflector 25 to obtain a detection beam, and the detection beam enters the laser receiver 22 through the beam splitter 23. Wherein the measuring beam reflector 25 is arranged on the object to be measured. The laser receiver receives the detected beam and proceeds to step 12.
In some embodiments, the beam splitter of the device failure detection apparatus may perform laser beam splitting using a partial wavefront method, a partial amplitude method, or a partial polarization method.
The wavefront dividing method divides the wavefront of a point light source into two parts, and the two parts pass through two optical systems respectively, are overlapped after reflection, refraction or diffraction, and form interference in a certain area. Since any part of the wavefront can be regarded as a new light source and each part of the same wavefront has the same phase, these separated parts of the wavefront can be regarded as light sources with the same initial phase, and the initial phase difference of these light sources is constant no matter how fast the phase of the point light source changes.
The partial amplitude method is a method in which when a beam of light is projected onto the interface between two transparent media, a part of the light energy is reflected and the other part is refracted.
The polarization splitting method adopts a polarization beam splitter which is formed by gluing a pair of glass prisms, and magnesium fluoride and zinc sulfide film layers are alternately evaporated on the gluing surface of one prism. Incident light enters the dielectric layer at the Brewster angle, and is transmitted and reflected for multiple times to obtain S-component reflected light and P-component transmitted light with high polarization degree. The polarizing beam splitter may be composed of polarizing prisms with orthogonal crystal axes, such as wollaston prisms.
In some embodiments, the reference beam reflector or the measurement beam reflector in the equipment failure detection device may be a plane reflector, a corner cube reflector, a right angle cube reflector, a cat-eye reflector.
Step 12: a vibration signal is extracted from the detected light beam.
In some embodiments, the device failure detection apparatus includes an optical path unit, a demodulation unit, and a signal generation unit.
The signal generating unit comprises a signal generator, an output signal of the signal generator is divided into two paths, the first path outputs two paths of orthogonal signals sin (Csin (ω t)) and cos (Csin (ω t)) after passing through a signal processor, and the second path outputs a phase-shift driving signal sin (ω t) after passing through a DA converter; c is a sine coefficient, and omega is a phase shift frequency caused by a carrier wave modulated by the electro-optical modulator in the optical path unit.
The light path unit comprises a laser and an electro-optic modulator, light output by the laser is divided into a measuring beam and a reference beam through a beam splitter, the reference beam is input to a signal input end of the electro-optic modulator, the measuring beam is input to a first beam splitter prism, a phase shift driving signal sin (ω t) output by the signal generating unit is input to an electric signal driving end of the electro-optic modulator, and the phase shift reference beam output by the electro-optic modulator under the action of the phase shift driving signal sin (ω t) is input to a second beam splitter prism; returning measuring beams formed after the measuring beams passing through the first beam splitter prism are reflected on the surface of the target to be measured enter the first beam splitter prism again, the returning measuring beams are reflected to the second beam splitter prism again, the returning measuring beams and the phase-shifted reference beams are mixed and interfered in the second beam splitter prism to form interference light, and the interference light is input into the photoelectric detector and then is output as an interference signal through the processor; omega is the phase shift frequency caused by the carrier wave modulated by the electro-optical modulator.
The demodulation unit is a hardware module realized in FPGA through digital logic, and comprises four multipliers, two low-pass filters, two differentiators, a subtracter and an integrator, wherein the input end of the demodulation unit is connected with two paths of orthogonal signals sin (Csin (ω t)) and cos (Csin (ω t)) from the signal generation unit and an interference signal from the optical path unit, the interference signal is divided into two paths after passing through an AD converter, one path of the interference signal and the cos (Csin (ω t)) are sent into the first multiplier I to be multiplied, and the other path of the interference signal and the sin (Csin (ω t)) are sent into the second multiplier to be multiplied; the product output by the first multiplier is divided into two paths after passing through a low-pass filter, wherein one path is sent into a third multiplier after passing through a differentiator d/dt, the other path is directly sent into a fourth multiplier, the product output by the second multiplier is divided into two paths after passing through another low-pass filter, one path is sent into the fourth multiplier after passing through another differentiator d/dt, and the other path is directly sent into the third multiplier; the product output by the third multiplier and the product output by the fourth multiplier are sent to the subtracter, and the difference output by the subtracter is sent to the integrator.
The quadrature phase locking + DCM demodulation method comprises the following steps:
(1) two paths of orthogonal signals sin (Csin (ω t)) and cos (Csin (ω t)) and one path of analog interference signal are selected, and the analog interference signal is divided into two paths of digital interference signals serving as demodulation input parameters after passing through an AD converter.
(2) One path of digital interference signal and cos (Csin (ω t)) are sent to a first multiplier to be multiplied, and the other path of digital interference signal and sin (Csin (ω t)) are sent to a second multiplier to be multiplied;
(3) dividing the product output by the first multiplier into two paths after passing through a low-pass filter, wherein one path is differentiated and then sent to a third multiplier, and the other path is directly sent to a fourth multiplier; dividing the product output by the second multiplier into two paths after passing through a low-pass filter, wherein one path is differentiated and then sent to a fourth multiplier, and the other path is directly sent to a third multiplier;
(4) subtracting the product output by the third multiplier and the product output by the fourth multiplier to obtain a difference value;
(5) and integrating the difference value to obtain an integral value, and demodulating a vibration signal of the target to be detected.
By demodulating the detection beam, the vibration signal, such as amplitude, frequency, phase of the vibration signal, is extracted.
In some embodiments, the vibration speed of the measured device is obtained by calculating the reference beam and the detection beam, and then the vibration frequency of the measured device is calculated.
Step 13: the vibration signal is input to a trained fault learning model to identify a fault type associated with the vibration signal to determine a fault type of the target device.
The trained fault learning model is obtained by training based on vibration signal sample data and a predetermined fault type label.
In some embodiments, the vibration signal is input to the trained fault learning model, and the feature extraction is performed on the vibration signal, which may be performed by time-frequency analysis processing to extract feature parameters of signal processing. The idea of the time-frequency analysis method is to design a joint function of time and frequency, and the joint function is used for simultaneously describing the energy density or the intensity of a signal at different times and frequencies. The signal is analyzed by utilizing time-frequency distribution, the instantaneous frequency and the amplitude thereof at each moment can be given, and time-frequency filtering and time-varying signal research can be carried out. The time-frequency analysis method comprises a time-domain waveform analysis method, a probability density analysis method, an autocorrelation analysis method, a trend analysis method, an amplitude spectrum analysis method, a power spectrum analysis method, a cepstrum analysis method, an envelope spectrum analysis method, a refined spectrum analysis method, a short-time Fourier analysis method and a wavelet analysis method. The above is only an example, and the obtaining of the vibration characteristic parameter is not limited to the above method, and may be implemented by other means.
The extracted characteristic information is identified to identify a fault type associated with the vibration signal, thereby determining the fault type of the target device.
In some embodiments, the fault learning model may be constructed by way of machine learning. By using a supervised learning method, different vibration signal samples are artificially input, the vibration signal samples are labeled according to fault types corresponding to the vibration signal samples, and the vibration signal samples are input to a learning model to be trained so as to train the learning model to be trained to form a fault learning model. The corresponding flag may be a fault type corresponding to the vibration signal. And performing feature extraction on the vibration signals, setting corresponding marks for the feature information, and constructing a fault learning model by using the feature information and the marks as training data. When an unknown vibration signal is input into the fault learning model, a corresponding mark is output, and the mark is a fault type corresponding to the vibration signal.
In some embodiments, a gaussian mixture model, a hidden markov model, K-nearest neighbor, a neural network, a support vector machine, etc. may be used for model training to train a fault learning model, and after training is completed, the model may be used to identify an unknown vibration signal.
In some embodiments, the vibration signal is input into a trained fault learning model, pre-processed in the fault learning model, the pre-processed vibration signal is identified, and a fault type associated with the pre-processed vibration signal is identified, thereby feeding back the user result.
In some embodiments, the target device includes a motor, and the motor operates to generate vibrations, and if the target device fails, the vibration signals generated by the vibrations of the motor are different. On the basis of big data, a plurality of vibration signals of faults are collected firstly, so that the vibration signals can be used for deep learning to train a deep learning model, and after the deep learning model is built, the deep learning model is used for identifying subsequent vibration signals to judge the fault type.
In some embodiments, the target device is not capable of generating vibration, and requires an external driving device to be connected, and the target device is driven by the external driving device to generate vibration, and the target device may be some welded component. Referring to fig. 3, taking an application in the welding technology as an example, a welded metal plate 31 is connected to an external driving device 32, so that the external driving device 32 vibrates to drive the metal plate 31 to vibrate, where P is a welding point, and a device failure detection device 33 collects vibration signals. The sheet metal 31 has the common defects of undercut, welding beading, dent, welding deformation and the like during welding, and sometimes has surface pores and surface cracks. The root of the single-side welding is not welded completely, and the like, and also has air holes, slag inclusion, cracks, non-fusion and the like. In addition, after welding is finished, faults such as welding spot loosening exist. It will be appreciated that each defect will cause an adverse effect in subsequent applications of the sheet metal 31 and is therefore an unacceptable weld. And vibration signals generated by each defect are different, and the collected vibration signals are subjected to feature extraction and input into a trained fault learning model so as to identify the associated fault type.
Different from the situation of the prior art, the equipment fault detection method comprises the following steps: emitting a measuring beam to the target device and collecting a detection beam of the measuring beam reflected by the target device; extracting a vibration signal from the detected light beam; the vibration signal is input to a trained fault learning model to identify a fault type associated with the vibration signal to determine a fault type of the target device. In this way, adopt non-contact vibration signal measurement scheme on the one hand, need not external syllable-dividing room, high-efficient check out test set trouble problem, on the other hand when big batch equipment quality control, save time and human cost, improve quality control efficiency.
Referring to fig. 4, fig. 4 is a schematic flowchart of a second embodiment of the device fault detection method provided in the present application, where the method includes:
step 41: the measuring beam is emitted toward the target device, and a detection beam reflected by the measuring beam via the target device is collected.
Step 42: a vibration signal is extracted from the detected light beam.
Step 43: the vibration signal is converted into a sound signal.
Step 44: the sound signal is pre-processed.
In some embodiments, the sound signal is pre-emphasized in order to emphasize high frequency portions of the speech, increasing the high frequency resolution of the speech to compensate for the high frequency components in the sound signal. Digital filters are typically employed to implement pre-emphasis.
After the pre-emphasis digital filtering processing is performed, windowing and frame division processing is performed. A speech signal is a signal that varies with time and is mainly classified into voiced speech and unvoiced speech. The pitch period of voiced sounds, the voiced and unvoiced signal amplitude, and the vocal tract parameters, etc. all vary slowly with time. Due to the inertial motion of the sounding organ, the voice signal is considered to be approximately constant in a short time (generally 10-30 ms), that is, the voice signal has short-time stationarity. In this way, the speech signal can be divided into short segments (called analysis frames) for processing. Framing of speech signals is achieved by weighting with movable finite-length windows. The number of frames per second is generally 33-100 frames, depending on the actual situation. The frame division can adopt a continuous segmentation method and also can adopt an overlapped segmentation method, which is used for ensuring the smooth transition between frames and keeping the continuity of the frames. The overlapping part of the previous frame and the next frame is called frame shift, and the ratio of the frame shift to the frame length is generally 0-1/2. Here, a rectangular window or a hamming window may be used for windowing.
Step 45: and inputting the sound signal after preprocessing into the trained fault learning model to identify the fault type associated with the sound signal so as to determine the fault type of the target equipment.
The trained fault learning model is obtained by training based on the sound signal sample data and a predetermined fault type label.
In some embodiments, the sound signal after preprocessing is input into a trained fault learning model, and feature extraction is performed on the sound signal. The extracted feature information may include: time domain features, frequency domain features, cepstral features, time-frequency features, and the like.
The time domain features mainly include short-time energy, short-time average amplitude, short-time zero crossing rate and the like. The short-time average amplitude reflects the amount of energy of a frame of speech signal. The short-time zero-crossing rate represents the number of times the waveform crosses the horizontal axis in one frame of the audio signal.
The frequency domain features are features obtained by performing fourier transform on the signal to convert the signal to the frequency domain and then calculating the frequency domain. Mainly comprising a frequency spectrum centroid, a bandwidth, a frequency spectrum roll-off coefficient and the like. The spectral centroid reflects the mean of the energy per frame. The bandwidth reflects the degree of fluctuation of the energy of the sampling points around the mean value. The spectral roll-off describes the degree of spectral tilt.
Cepstrum is the inverse fourier transform of the log of the power or energy spectrum.
In some embodiments, the sound signal after preprocessing is input into a deep learning network, preprocessing is carried out in the deep learning network, the preprocessed sound is identified, and a fault type associated with the preprocessed sound signal is identified, so that a user result is fed back.
Referring to fig. 5, fig. 5 is a schematic flow chart of a third embodiment of the device fault detection method provided in the present application, where the method includes:
step 51: the measuring beam is emitted toward the target device, and a detection beam reflected by the measuring beam via the target device is collected.
Step 52: a vibration signal is extracted from the detected light beam.
Step 53: the vibration signal is converted into a sound signal.
Step 54: the sound signal is pre-processed.
Step 55: and converting the sound signal after the pretreatment into a spectrogram.
A spectrogram is a distribution of time and frequency. The spectrogram not only reflects the frequency domain and time domain characteristics of the acoustic signal, but also shows the correlation between the time domain and the frequency domain, and the situation that some characteristics of the frequency domain change along with the occurrence of the acoustic signal can be observed from the spectrogram; energy variations with acoustic processes can also be observed. Therefore, the information of the sound signal carried by the spectrogram is far larger than the information carried by a pure time domain signal and a pure frequency domain signal. The spectrogram integrates the characteristics of a spectrogram and a time domain waveform, and obviously shows the change condition of the spectrogram of the sound along with time, or the spectrogram is a dynamic spectrogram.
In some embodiments, the framing window length is 512 points, the window function is hamming window, the frame stack is 0.75 times the window length, and the specgram function is called by a Matlab third party tool, VoiceBox, to draw the spectrogram.
Step 56: the spectrogram is input to a trained fault learning model to identify a fault type associated with the spectrogram.
The trained fault learning model is obtained by training based on spectrogram sample data and a predetermined fault type label.
In some embodiments, the converted sound signal is pre-emphasized, and a specific high-pass filter is used to compensate the high-frequency part of the collected sound signal; then, performing end point detection on the pre-emphasized sound signal, and determining a starting point of an effective signal in the sound signal, generally speaking, because there may be silence or blank in the collected sound signal for a period of time, in order to improve the detection efficiency of abnormal sound, this embodiment may determine the starting point of effective sound information (i.e., effective signal) in the sound signal, and then perform abnormal sound matching or detection on the effective signal; then, the characteristic parameters of the effective signals with the determined starting points in a certain range are subjected to frame windowing, so that the statistical properties are stable.
And converting the preprocessed sound signals into a spectrogram, wherein the spectrogram is a two-dimensional graph, the horizontal axis of the spectrogram is time, and the vertical axis of the spectrogram is frequency. The point corresponding to the coordinate (x, y) represents the sound intensity at time x, frequency y, which is represented by different colors. From the spectrogram of the sound signal, the distribution and the change condition of the sound intensity in the whole time-frequency range can be obtained; which cannot be represented in the waveform diagram. To obtain a spectrogram, the sound signal is divided into very short frames, with some overlap between adjacent frames. And then, short-time Fourier transform is carried out on each frame to obtain corresponding frequency spectrum information, and the spectrogram consists of three dimensional information of frequency, time and sound intensity, so that the value of the sound intensity needs to be calculated. And finally, connecting the frequency spectrum information into a complete spectrogram.
For example, the embodiment of the present application divides the effective signal segment into several frames by windowing; carrying out short-time Fourier transform on each frame to obtain frequency spectrum information of the frame, wherein the frequency spectrum information is used for expressing the relation between the frequency and the sound intensity of the frame; and connecting the spectrum information of all the frames to obtain a spectrogram of the effective signal segment, wherein the spectrogram consists of a plurality of points, and the coordinates (x, y) of any point are used for representing the corresponding sound intensity of the point at the x moment and the y frequency. In the embodiment of the present application, a time series signal of a sound is subjected to short-time fourier transform, the length of the fourier transform is 2N points, so that a frequency spectrum with a length of N can be obtained from a signal of each frame, and a sound pressure value of each point is represented as: p20 × log10| x (1/N) |; wherein, P is the sound pressure value of the point, and x is the spectrum value of the frame signal.
And distinguishing abnormal sounds and background noises according to the spectrogram. And generating a spectrogram matrix according to the frame, the frequency and the sound intensity of each point in the spectrogram. And extracting a to-be-tested identification feature matrix for representing the sound intensity distribution condition of the spectrogram from the spectrogram matrix. And inputting the identification feature matrix to be tested into the trained fault learning model, and outputting the fault type of the identification feature matrix to be tested through internal identification of the model.
Referring to fig. 6, fig. 6 is a schematic flow chart of a fourth embodiment of the device fault detection method provided in the present application, where the method includes:
step 61: the measuring beam is emitted toward the target device, and a detection beam reflected by the measuring beam via the target device is collected.
Step 62: a vibration signal is extracted from the detected light beam.
And step 63: the vibration signal is converted into a sound signal.
Step 64: the sound signal is pre-processed.
Step 65: and converting the sound signal after the pretreatment into a spectrogram.
Steps 61-65 have the same or similar technical solutions as the above embodiments, and are not described herein.
And step 66: and extracting time information, frequency information and energy information of the spectrogram to obtain to-be-processed characteristic information.
In the present embodiment, the spectrogram has an abscissa representing time, an ordinate representing frequency, and a value of a point on the graph of time and frequency representing sound signal energy. Typically the magnitude of the energy value is represented by a color, the darker the color indicates the stronger the energy at that point.
In some embodiments, features of the spectrogram can be extracted from multiple dimensions of frequencies using a Gabor transform for feature extraction.
In some embodiments, feature extraction is performed using a projection method. The projection method is that a spectrogram is projected in four directions (0 degree, 45 degrees, 90 degrees and 135 degrees) respectively, and projected in the 0 degree direction (abscissa), namely pixel brightness values of components at the same time and different frequency components are accumulated; projecting towards the 90-degree direction (vertical coordinate), namely accumulating the pixel brightness values of the same frequency component and different time components; and projecting in the directions of 45 degrees and 135 degrees, accumulating the pixel brightness values vertical to the directions, and dividing the pixel brightness values by the number of pixels after each accumulation in order to ensure the consistency of data quantity.
Step 67: inputting the characteristic information to be processed into the trained fault learning model so as to identify the fault type associated with the spectrogram.
The trained fault learning model is obtained by training based on spectrogram sample data and a predetermined fault type label.
In the present embodiment, the failure learning model is modeled by supervised learning in machine learning. The fault type is corresponding to the characteristic information in advance, the characteristic information is labeled, the labeled content can be the fault type, the characteristic information to be processed is input after the fault learning model is established, and the fault learning model outputs the corresponding labeled information, namely the fault type.
In some embodiments, the fault types are classified when training the fault learning model. Taking a washing machine as an example, unfixed screws are classified into one type, and motor faults are classified into one type. And when the characteristic information of the current spectrogram is determined to be the unfixed type of the screw, continuously comparing the current spectrogram in the unfixed type of the screw.
In some embodiments, when the similarity between the feature information to be processed of the sound information of the target device and the standard feature information is greater than a set threshold, it is determined that the fault type associated with the sound signal is the fault type corresponding to the standard feature information. Taking a target device as a refrigerator as an example, converting a vibration signal acquired through a fault detection device into a sound signal, converting the sound signal into a spectrogram, performing feature extraction on the spectrogram in a deep learning network, performing similarity comparison on feature information and pre-stored standard feature information, wherein the compared similarity is eighty-five percent, the set threshold value is eighty percent, and the similarity is greater than the set threshold value, and determining that the fault type associated with the sound signal is the fault type corresponding to the standard feature information.
Referring to fig. 7, fig. 7 is a schematic flowchart of a fifth embodiment of a voice interaction method of an electronic device provided in the present application, where the method includes:
step 71: the measuring beam is emitted toward the target device, and a detection beam reflected by the measuring beam via the target device is collected.
Step 72: a vibration signal is extracted from the detected light beam.
Step 73: the vibration signal is converted into a sound signal.
Step 74: the sound signal is pre-processed.
Step 75: and converting the sound signal after the pretreatment into a spectrogram.
Step 76: the spectrogram is input to a trained fault learning model.
Steps 71-76 have the same or similar technical solutions as the above embodiments, and are not described herein.
Step 77: and performing region blocking on the spectrogram by using a fault learning model, and distributing corresponding weights to different regions to obtain a blocked spectrogram to be processed.
Step 78: and performing weighting processing on the to-be-processed block spectrogram by using a fault learning model to obtain a weighted similarity comparison result.
Step 79: and determining the fault type associated with the sound signal as the fault type corresponding to the spectrogram in response to the weighted similarity comparison result being larger than a set threshold.
Referring to fig. 8, steps 77 to 79 will be described, where the left side of fig. 8 is the block spectrogram to be processed, and the right side is the standard block spectrogram. Dividing the block spectrogram to be processed on the left side into four blocks of ABCD, distributing weights to the four blocks of ABCD, and performing weighting processing on A and a, B and B, C and C, and D and D. For example, A, B, C, D are assigned weights of 30%, 20%, 40%, 10%, respectively; wherein the weight values of A, B, C, D add up to 1. The formula of similarity alignment is S ═ a × 30% + B × 20% + C × 40% + D × 10%.
Wherein, A is the comparison result of the similarity between A and a. B denotes the similarity alignment of B and B. C indicates the similarity alignment of C and C. D x D represents the similarity comparison result between D and D.
And when the value of S is larger than the set threshold, determining the fault type associated with the sound signal as the fault type corresponding to the set threshold.
For example, the threshold value is set to 70%, and
s > 70% because S is 90% + 30% + 80% + 20% + 70% + 40% + 60% + 10% + 77%; determining the fault type associated with the sound signal as the fault type corresponding to the spectrogram.
Referring to fig. 9, fig. 9 is a schematic structural diagram of an embodiment of the device fault detection apparatus provided in the present application, and the device fault detection apparatus 90 includes a laser transmitter 91, a laser receiver 92, and a processor 93.
And a laser transmitter 91 for transmitting a measuring beam to the target device.
A laser receiver 92 for collecting a detection beam of the measuring beam reflected via the target device.
A processor 93 connected to the laser receiver for extracting the vibration signal from the detected light beam and inputting the vibration signal to the trained fault learning model to identify a fault type associated with the vibration signal to determine a fault type of the target device; the trained fault learning model is obtained by training based on vibration signal sample data and a predetermined fault type label.
The processor 93 is further configured to convert the vibration signal into a sound signal, and input the sound signal to the trained fault learning model to identify a fault type associated with the sound signal, so as to determine a fault type of the target device; the trained fault learning model is obtained by training based on the sound signal sample data and a predetermined fault type label.
The processor 93 is further configured to pre-process the sound signal, and input the pre-processed sound signal to the trained fault learning model to identify a fault type associated with the sound signal, so as to determine the fault type of the target device.
The processor 93 is further configured to convert the vibration signal into a sound signal, pre-process the sound signal, convert the pre-processed sound signal into a spectrogram, and input the spectrogram into a trained fault learning model to identify a fault type associated with the spectrogram; the trained fault learning model is obtained by training based on spectrogram sample data and a predetermined fault type label.
The processor 93 is further configured to extract time information, frequency information, and energy information of the spectrogram, so as to obtain feature information to be processed; inputting the characteristic information to be processed into the trained fault learning model so as to identify the fault type associated with the spectrogram.
The processor 93 is also used to input the spectrogram into the trained fault learning model; carrying out region blocking on the spectrogram by using a fault learning model, and distributing corresponding weights to different regions to obtain a blocked spectrogram to be processed; carrying out weighting processing on the to-be-processed block spectrogram by using a fault learning model so as to obtain a weighted similarity comparison result; and determining the fault type associated with the sound signal as the fault type corresponding to the spectrogram in response to the weighted similarity comparison result being larger than a set threshold.
The processor 93 is connected to the laser emitter 91, and is configured to divide the laser beam emitted by the laser emitter 91 into a measurement beam and a reference beam, and interfere the detection beam with the reference beam to detect a vibration signal.
It can be understood that the technical solution of any of the above embodiments can be implemented by using the device failure detection apparatus 90.
In some embodiments, referring to fig. 10, the device failure detection apparatus 100 includes a laser emitter 101, a beam splitter 102, a beam splitter 103, a beam splitter 104, a bragg cell 105, a fixed reflector 106, a photodetector 107. It is to be understood that fig. 10 only illustrates the transmitting and receiving processes of the laser beam, and a partial structural diagram of the subsequent process is not shown. The target device is shown at 200.
In the following, a specific working flow is described, wherein the laser transmitter 101 transmits a laser beam to the beam splitter 102, and the beam splitter 102 divides the laser beam into a measuring beam and a reference beam; as shown in the direction of the measuring beam toward the beam splitter 103, the reference beam is directed toward the fixed reflector 106, the reference beam is reflected by the fixed reflector 106, is directed into the bragg cell 105, passes through the bragg cell 105 and is directed toward the beam splitter 104, and the reference beam is directed toward the photodetector 107 through the beam splitter 104; the measuring beam is emitted to the target device 200 after passing through the beam splitter 103, and the reflected detecting beam is collected and deflected downwards by the beam splitter 103 to be emitted to the beam splitter 104; the detection beam is directed to a photodetector 107 via a beam splitter 104; at this time, both the reference beam and the detection beam are incident on the photodetector 107, and the vibration signal of the target device is calculated by the photodetector 107 using the relationship that the doppler shift is proportional to the vibration velocity of the target device 200.
Specifically, since the optical path of the reference beam is constant, the vibration of the target apparatus 200 generates bright/dark fringes on the photodetector 107, which is an interference method by use. A full bright/dark periodic fringe on the photodetector 107 corresponds exactly to the half wavelength shift of the laser used. This corresponds to a displacement of 316nm in the case of helium neon lasers which are often used for laser emitters 101.
The change in optical path length per unit time appears as a doppler shift of the measuring beam. In metrology, it is meant that the doppler shift is directly proportional to the sample vibration velocity. This arrangement does not allow the direction of object movement to be specified, since the bright and dark fringes (and modulation frequency) produced by the movement of an object away from the interferometer are the same as those produced by the movement of an object towards the interferometer. For this reason, an acousto-optic modulator with an optical frequency shift of typically 40MHz is placed in the reference beam (for comparison purposes, the laser frequency is 4.74-1014 Hz). When the target device 200 is in a quiescent state, a typical interferometric modulation frequency of 40MHz will be produced. Therefore, when the target device 200 moves toward the device failure detection apparatus, the modulation frequency increases; when the target device 200 is moved away from the interferometer, then the frequency received by the detector is less than 40 MHz. Not only can the optical path length be accurately detected, but also the direction of motion can be detected.
Specifically, the device failure detection apparatus 100 can directly measure the displacement amount in addition to directly measuring the vibration speed. The maximum amplitude of the harmonic vibration can be expressed as follows:
v 2 pi f s; wherein v represents the velocity of the target device; f represents the vibration frequency of the target device; s represents the vibrational displacement of the target device. As the frequency increases, the vibration speed increases and the vibration displacement decreases.
Referring to fig. 11, fig. 11 is a schematic structural diagram of an embodiment of a computer storage medium provided in the present application, where the computer storage medium 110 is used for storing program data 111, and the program data 111, when being executed by a processor, is used for implementing the following method steps:
emitting a measuring beam to the target device and collecting a detection beam of the measuring beam reflected by the target device; extracting a vibration signal from the detected light beam; the vibration signal is input to a trained fault learning model to identify a fault type associated with the vibration signal to determine a fault type of the target device.
It will be appreciated that the program data 111, when executed by a processor, is also for implementing any of the embodiment methods described above.
In the several embodiments provided in the present application, it should be understood that the disclosed method and apparatus may be implemented in other manners. For example, the above-described device embodiments are merely illustrative, and for example, the division of the modules or units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated units in the other embodiments described above may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, a network device, or the like) or a processor (processor) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above description is only for the purpose of illustrating embodiments of the present application and is not intended to limit the scope of the present application, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the specification and the drawings of the present application or are directly or indirectly applied to other related technical fields, are also included in the scope of the present application.
Claims (19)
1. A method of device fault detection, the method comprising:
emitting a measuring beam to a target device and collecting a detection beam reflected by the measuring beam via the target device;
extracting a vibration signal from the detected light beam;
inputting the vibration signal to a trained fault learning model to identify a fault type associated with the vibration signal to determine a fault type of the target device.
2. The method of claim 1,
the trained fault learning model is obtained after training based on vibration signal sample data and a predetermined fault type label.
3. The method of claim 1,
the inputting the vibration signal to a trained fault learning model to identify a fault type associated with the vibration signal includes:
converting the vibration signal into a sound signal;
inputting the sound signal to a trained fault learning model to identify a fault type associated with the sound signal to determine a fault type of the target device;
the trained fault learning model is obtained by training based on sound signal sample data and a predetermined fault type label.
4. The method of claim 3,
the inputting the sound signal into a trained fault learning model to identify a fault type associated with the sound signal to determine the fault type of the target device includes:
preprocessing the sound signal;
inputting the sound signal after preprocessing into a trained fault learning model to identify a fault type associated with the sound signal, thereby determining the fault type of the target device.
5. The method of claim 3,
the preprocessing the sound signal comprises:
pre-emphasis the sound signal to compensate for high frequency components in the sound signal;
and performing frame division and windowing processing on the sound signal by adopting a preset window function to obtain the sound signal after preprocessing.
6. The method of claim 1,
the inputting the vibration signal to a trained fault learning model to identify a fault type associated with the vibration signal includes:
converting the vibration signal into a sound signal;
preprocessing the sound signal;
converting the preprocessed sound signal into a spectrogram;
inputting the spectrogram into a trained fault learning model to identify a fault type associated with the spectrogram, thereby determining a fault type of the target device;
the trained fault learning model is obtained by training based on spectrogram sample data and a predetermined fault type label.
7. The method of claim 6,
the inputting the spectrogram into a trained fault learning model to identify a fault type associated with the spectrogram comprises:
extracting time information, frequency information and energy information of the spectrogram to obtain to-be-processed characteristic information;
inputting the feature information to be processed into a trained fault learning model to identify a fault type associated with the spectrogram.
8. The method of claim 6,
the inputting the spectrogram into a trained fault learning model to identify a fault type associated with the spectrogram comprises:
inputting the spectrogram into a trained fault learning model;
carrying out region blocking on the spectrogram by using the fault learning model, and distributing corresponding weights to different regions to obtain a blocked spectrogram to be processed;
weighting the to-be-processed block spectrogram by using the fault learning model to obtain a weighted similarity comparison result;
and determining the fault type associated with the sound signal as the fault type corresponding to the spectrogram in response to the weighted similarity comparison result being larger than a set threshold value.
9. The method of claim 1,
before the emitting the measuring beam to the target device and collecting the detecting beam reflected by the target device, the method further includes:
splitting a laser beam into a measurement beam and a reference beam;
the extracting of the vibration signal from the detected light beam includes:
and interfering the detection beam and the reference beam to detect and obtain a vibration signal.
10. The method of claim 1,
the target device includes a motor, and the target device is operated by the motor to generate vibration.
11. The method of claim 1,
before the emitting the measuring beam to the target device and collecting the detecting beam reflected by the target device, the method further includes:
and connecting the target device through an external driving device to drive the target device to generate vibration.
12. An equipment failure detection apparatus, characterized in that the equipment failure detection apparatus comprises:
the laser transmitter is used for transmitting a measuring beam to the target equipment;
a laser receiver for collecting a detection beam of the measurement beam reflected by the target device;
a processor connected to the laser receiver for extracting a vibration signal from the detected light beam and inputting the vibration signal to a trained fault learning model to identify a fault type associated with the vibration signal to determine a fault type of the target device;
the trained fault learning model is obtained by training based on vibration signal sample data and a predetermined fault type label.
13. The apparatus of claim 12,
the processor is further used for converting the vibration signal into a sound signal and inputting the sound signal into a trained fault learning model so as to identify a fault type associated with the sound signal, and therefore the fault type of the target equipment is determined;
the trained fault learning model is obtained by training based on sound signal sample data and a predetermined fault type label.
14. The apparatus of claim 13,
the processor is further used for preprocessing the sound signal and inputting the sound signal after preprocessing into a trained fault learning model to identify a fault type associated with the sound signal so as to determine the fault type of the target device.
15. The apparatus of claim 12,
the processor is further used for converting the vibration signal into a sound signal, preprocessing the sound signal, converting the preprocessed sound signal into a spectrogram, and inputting the spectrogram into a trained fault learning model to identify a fault type associated with the spectrogram;
the trained fault learning model is obtained by training based on spectrogram sample data and a predetermined fault type label.
16. The apparatus of claim 15,
the processor is further used for extracting time information, frequency information and energy information of the spectrogram to obtain to-be-processed characteristic information; inputting the feature information to be processed into a trained fault learning model to identify a fault type associated with the spectrogram.
17. The apparatus of claim 15,
the processor is further configured to input the spectrogram into a trained fault learning model; carrying out region blocking on the spectrogram by using the fault learning model, and distributing corresponding weights to different regions to obtain a blocked spectrogram to be processed; weighting the to-be-processed block spectrogram by using the fault learning model to obtain a weighted similarity comparison result; and determining the fault type associated with the sound signal as the fault type corresponding to the spectrogram in response to the weighted similarity comparison result being larger than a set threshold value.
18. The apparatus of claim 12,
the processor is connected with the laser transmitter and is used for dividing the laser beam emitted by the laser transmitter into a measuring beam and a reference beam and interfering the detecting beam and the reference beam to obtain a vibration signal through detection.
19. A computer storage medium for storing program data for implementing the method according to any one of claims 1-10 when executed by a processor.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911019720.1A CN112710486B (en) | 2019-10-24 | 2019-10-24 | Equipment fault detection method, equipment fault detection device and computer storage medium |
| PCT/CN2019/124118 WO2021077567A1 (en) | 2019-10-24 | 2019-12-09 | Device failure detection method, device failure detection apparatus and computer storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911019720.1A CN112710486B (en) | 2019-10-24 | 2019-10-24 | Equipment fault detection method, equipment fault detection device and computer storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112710486A true CN112710486A (en) | 2021-04-27 |
| CN112710486B CN112710486B (en) | 2022-01-25 |
Family
ID=75540471
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911019720.1A Active CN112710486B (en) | 2019-10-24 | 2019-10-24 | Equipment fault detection method, equipment fault detection device and computer storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112710486B (en) |
| WO (1) | WO2021077567A1 (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114689161A (en) * | 2022-03-30 | 2022-07-01 | 珠海格力电器股份有限公司 | Detection method, nonvolatile storage medium, and detection system |
| CN114754413A (en) * | 2022-04-11 | 2022-07-15 | 青岛海信日立空调系统有限公司 | Multi-split air conditioning system and fault positioning method |
| CN115420977A (en) * | 2022-08-26 | 2022-12-02 | 正泰集团研发中心(上海)有限公司 | Electric appliance fault detection method, training method, computer equipment and storage medium |
| CN115691537A (en) * | 2022-12-28 | 2023-02-03 | 江苏米笛声学科技有限公司 | Analysis and processing system of earphone audio signal |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113268924B (en) * | 2021-05-18 | 2023-01-31 | 国网福建省电力有限公司电力科学研究院 | Time-frequency characteristic-based fault identification method for on-load tap-changer of transformer |
| CN113820109A (en) * | 2021-08-23 | 2021-12-21 | 广东电力发展股份有限公司 | Power plant auxiliary engine rotating equipment inspection device, method, equipment and medium |
| CN113820624B (en) * | 2021-09-30 | 2024-04-16 | 南方电网科学研究院有限责任公司 | High-resistance ground fault identification device for power distribution network |
| CN115629930B (en) * | 2022-12-23 | 2023-08-08 | 北京东远润兴科技有限公司 | Fault detection method, device, equipment and storage medium based on DSP system |
| CN115951002B (en) * | 2023-03-10 | 2023-05-16 | 山东省计量科学研究院 | Gas chromatograph-mass spectrometer fault detection device |
| CN116933170B (en) * | 2023-09-18 | 2024-01-02 | 福建福清核电有限公司 | Mechanical seal fault classification method |
| CN118312796B (en) * | 2024-06-05 | 2024-08-06 | 深圳市华测半导体设备有限公司 | High-precision FFT test optimization method, system and storage medium of tester |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2090853C1 (en) * | 1993-08-06 | 1997-09-20 | Павел Анатольевич Давыдов | Method of vibroacoustic diagnosis of machinery |
| US5679899A (en) * | 1995-03-06 | 1997-10-21 | Holographics Inc. | Method and apparatus for non-destructive testing of structures |
| CN203241143U (en) * | 2013-05-14 | 2013-10-16 | 西安邮电大学 | Online motor vibration monitoring device based on laser Doppler effect |
| CN105737965A (en) * | 2016-02-29 | 2016-07-06 | 莆田学院 | Aerogenerator vibration detection device and analysis method |
| CN105810213A (en) * | 2014-12-30 | 2016-07-27 | 浙江大华技术股份有限公司 | Typical abnormal sound detection method and device |
| CN105890740A (en) * | 2016-06-17 | 2016-08-24 | 佛山科学技术学院 | Air conditioner six-dimensional vibration testing system and method thereof |
| CN106404388A (en) * | 2016-09-13 | 2017-02-15 | 西安科技大学 | Scraper conveyor chain fluttering fault diagnosis method |
| CN106546892A (en) * | 2016-11-10 | 2017-03-29 | 华乘电气科技（上海）股份有限公司 | The recognition methodss of shelf depreciation ultrasonic audio and system based on deep learning |
| CN106599185A (en) * | 2016-12-14 | 2017-04-26 | 北京微智信业科技有限公司 | HSV-based image similarity identification method |
| CN107024331A (en) * | 2017-03-31 | 2017-08-08 | 中车工业研究院有限公司 | A kind of neutral net is to train motor oscillating online test method |
| CN107103270A (en) * | 2016-02-23 | 2017-08-29 | 云智视像科技（上海）有限公司 | A kind of face identification system of the dynamic calculation divided group coefficient based on IDF |
| CN108710756A (en) * | 2018-05-18 | 2018-10-26 | 上海电力学院 | The method for diagnosing faults of lower multicharacteristic information Weighted Fusion is analyzed based on spectral clustering |
| CN109697501A (en) * | 2017-10-20 | 2019-04-30 | 丹东东方测控技术股份有限公司 | Grinding machine sound voice print analysis equipment based on deep learning |
| CN110322896A (en) * | 2019-06-26 | 2019-10-11 | 上海交通大学 | A kind of transformer fault sound identification method based on convolutional neural networks |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7667827B2 (en) * | 2006-02-01 | 2010-02-23 | General Electric Company | Systems and methods for remote monitoring of vibrations in machines |
| CN104215320B (en) * | 2014-09-22 | 2017-01-18 | 珠海格力电器股份有限公司 | Vibration measuring device and system for electric machine |
-
2019
- 2019-10-24 CN CN201911019720.1A patent/CN112710486B/en active Active
- 2019-12-09 WO PCT/CN2019/124118 patent/WO2021077567A1/en active Application Filing
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2090853C1 (en) * | 1993-08-06 | 1997-09-20 | Павел Анатольевич Давыдов | Method of vibroacoustic diagnosis of machinery |
| US5679899A (en) * | 1995-03-06 | 1997-10-21 | Holographics Inc. | Method and apparatus for non-destructive testing of structures |
| CN203241143U (en) * | 2013-05-14 | 2013-10-16 | 西安邮电大学 | Online motor vibration monitoring device based on laser Doppler effect |
| CN105810213A (en) * | 2014-12-30 | 2016-07-27 | 浙江大华技术股份有限公司 | Typical abnormal sound detection method and device |
| CN107103270A (en) * | 2016-02-23 | 2017-08-29 | 云智视像科技（上海）有限公司 | A kind of face identification system of the dynamic calculation divided group coefficient based on IDF |
| CN105737965A (en) * | 2016-02-29 | 2016-07-06 | 莆田学院 | Aerogenerator vibration detection device and analysis method |
| CN105890740A (en) * | 2016-06-17 | 2016-08-24 | 佛山科学技术学院 | Air conditioner six-dimensional vibration testing system and method thereof |
| CN106404388A (en) * | 2016-09-13 | 2017-02-15 | 西安科技大学 | Scraper conveyor chain fluttering fault diagnosis method |
| CN106546892A (en) * | 2016-11-10 | 2017-03-29 | 华乘电气科技（上海）股份有限公司 | The recognition methodss of shelf depreciation ultrasonic audio and system based on deep learning |
| CN106599185A (en) * | 2016-12-14 | 2017-04-26 | 北京微智信业科技有限公司 | HSV-based image similarity identification method |
| CN107024331A (en) * | 2017-03-31 | 2017-08-08 | 中车工业研究院有限公司 | A kind of neutral net is to train motor oscillating online test method |
| CN109697501A (en) * | 2017-10-20 | 2019-04-30 | 丹东东方测控技术股份有限公司 | Grinding machine sound voice print analysis equipment based on deep learning |
| CN108710756A (en) * | 2018-05-18 | 2018-10-26 | 上海电力学院 | The method for diagnosing faults of lower multicharacteristic information Weighted Fusion is analyzed based on spectral clustering |
| CN110322896A (en) * | 2019-06-26 | 2019-10-11 | 上海交通大学 | A kind of transformer fault sound identification method based on convolutional neural networks |
Non-Patent Citations (2)
| Title |
|---|
| 宋琳琳 等: "小波域颜色和纹理特征提取及图像检索方法研究", 《小型微型计算机系统》 * |
| 李嘉頔 等: "分块CS-LBP和加权PCA的低分辨率人脸识别", 《光电子·激光》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114689161A (en) * | 2022-03-30 | 2022-07-01 | 珠海格力电器股份有限公司 | Detection method, nonvolatile storage medium, and detection system |
| CN114754413A (en) * | 2022-04-11 | 2022-07-15 | 青岛海信日立空调系统有限公司 | Multi-split air conditioning system and fault positioning method |
| CN114754413B (en) * | 2022-04-11 | 2023-10-27 | 青岛海信日立空调系统有限公司 | Multi-split air conditioning system and fault positioning method |
| CN115420977A (en) * | 2022-08-26 | 2022-12-02 | 正泰集团研发中心(上海)有限公司 | Electric appliance fault detection method, training method, computer equipment and storage medium |
| CN115691537A (en) * | 2022-12-28 | 2023-02-03 | 江苏米笛声学科技有限公司 | Analysis and processing system of earphone audio signal |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021077567A1 (en) | 2021-04-29 |
| CN112710486B (en) | 2022-01-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112710486B (en) | Equipment fault detection method, equipment fault detection device and computer storage medium | |
| Li et al. | A novel ensemble deep learning model for cutting tool wear monitoring using audio sensors | |
| Castellini et al. | Laser Doppler Vibrometry: Development of advanced solutions answering to technology's needs | |
| CA2390244C (en) | Methods and apparatuses for signal analysis | |
| Staszewski et al. | Wavelet signal processing for enhanced Lamb-wave defect detection in composite plates using optical fiber detection | |
| Patel et al. | Cochlear filter and instantaneous frequency based features for spoofed speech detection | |
| Lv et al. | The effect of speckles noise on the Laser Doppler Vibrometry for remote speech detection | |
| CN110333285A (en) | Ultrasonic Lamb waves Defect signal recognition method based on variation mode decomposition | |
| Xu et al. | Remote eavesdropping at 200 meters distance based on laser feedback interferometry with single-photon sensitivity | |
| Scislo | Quality assurance and control of steel blade production using full non-contact frequency response analysis and 3d laser doppler scanning vibrometry system | |
| Ohodnicki et al. | Fusion of distributed fiber optic sensing, acoustic NDE, and artificial intelligence for infrastructure monitoring | |
| CN114414659A (en) | Non-linear ultrasonic guided wave non-parametric damage identification method and system based on frequency fusion | |
| CN116243111A (en) | Cable bundle fault detection and positioning method and system | |
| Saratxaga et al. | Use of harmonic phase information for polarity detection in speech signals. | |
| CN114994185A (en) | Fatigue damage detection method for energy distribution variation under drive of excitation data | |
| KR20230152936A (en) | Method for contactless diagnosing power facility using artificial intelligence and signal processing technology and device using the same | |
| Kereliuk et al. | Improved hidden Markov model partial tracking through time-frequency analysis | |
| Govind et al. | Automatic speech polarity detection using phase information from complex analytic signal representations | |
| Xiao et al. | Full-field laser heterodyne imaging vibrometry using a CMOS–DVR system | |
| Lv et al. | Nonlinear dispersive component decomposition: Algorithm and applications | |
| CN110686710B (en) | All-fiber quantum chromatographic scanning device and scanning method | |
| CN112304415B (en) | Diffuse reflection type double-light-path air vibration detection system and method | |
| Tian et al. | Ultrasensitive detection of remote acoustic vibrations at 300 m distance by optical feedback enhancement | |
| Mitchell | Optical modal analysis using white-light projected fringes | |
| CN114705654B (en) | Gas concentration measuring method and device compatible with optical interference effect |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |