CN115015756A - Fault diagnosis method for fine-crushing fused new material mixer - Google Patents
Fault diagnosis method for fine-crushing fused new material mixer Download PDFInfo
- Publication number
- CN115015756A CN115015756A CN202210827199.XA CN202210827199A CN115015756A CN 115015756 A CN115015756 A CN 115015756A CN 202210827199 A CN202210827199 A CN 202210827199A CN 115015756 A CN115015756 A CN 115015756A
- Authority
- CN
- China
- Prior art keywords
- fault
- signal
- module
- mixer
- representing
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/34—Testing dynamo-electric machines
-
- 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
- G01M13/00—Testing of machine parts
- G01M13/02—Gearings; Transmission mechanisms
- G01M13/028—Acoustic or vibration analysis
-
- 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
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R15/00—Details of measuring arrangements of the types provided for in groups G01R17/00 - G01R29/00, G01R33/00 - G01R33/26 or G01R35/00
- G01R15/14—Adaptations providing voltage or current isolation, e.g. for high-voltage or high-current networks
- G01R15/20—Adaptations providing voltage or current isolation, e.g. for high-voltage or high-current networks using galvano-magnetic devices, e.g. Hall-effect devices, i.e. measuring a magnetic field via the interaction between a current and a magnetic field, e.g. magneto resistive or Hall effect devices
- G01R15/202—Adaptations providing voltage or current isolation, e.g. for high-voltage or high-current networks using galvano-magnetic devices, e.g. Hall-effect devices, i.e. measuring a magnetic field via the interaction between a current and a magnetic field, e.g. magneto resistive or Hall effect devices using Hall-effect devices
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/34—Testing dynamo-electric machines
- G01R31/346—Testing of armature or field windings
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a fault diagnosis method for a finely-divided fused new material mixing machine, which relates to the technical field of fault diagnosis and solves the technical problem of fault diagnosis of the new material mixing machine; classifying the acquired data information, and classifying and screening the data information of the running state of the mixing machine; the method comprises the steps of extracting fault information, monitoring an electric signal of a generator of the material mixer bearing fault through a current sensor and a voltage sensor, denoising acquired data information through a signal based on the combination of a Hankel matrix and singular value decomposition, and detecting the mixer fault through a fault diagnosis module, wherein the fault diagnosis module is a VMD-LSTM algorithm model. The invention greatly improves the fault diagnosis capability of the new material mixer.
Description
Technical Field
The invention relates to the technical field of fault diagnosis, in particular to a fault diagnosis method for a finely-crushed fused new material mixer.
Background
The mixer is a mechanical device for uniformly mixing two or more materials by using mechanical force, gravity and the like. During the mixing process, the material contact surface area can be increased to promote chemical reaction; but also to accelerate physical changes. The novel material mixing machine that fuses in small, broken bits of axial motion is a novel machine, in the motion process, can realize axial motion, mixes the novel material that fuses in small, broken bits, and this mixes machine various faults very easily appear in the application, for example because most mix quick-witted product all adopts the vibration principle, generally adopts the motor direct connection reducing mechanism, this kind of connected mode is simple, easy maintenance. So in general, when the motor rotor becomes non-concentric with the mixer rotor, the concentricity of the two rotors will not be consistent.
The mixed quick-witted fault data information among the conventional art appears bearing overheated usually, electric motor rotor and mixed quick-witted rotor decentraction, the axle center is skew, motor fault, the rotor is unusual, lubricating oil lacks, the sealing strip is damaged, cylinder atmospheric pressure is not enough, multiple problems such as cylinder atmospheric pressure is not enough, how to realize that multiple new material mixes quick-witted fault diagnosis is the problem that awaits the solution urgently, conventional art passes through check out test set and detects mixed quick-witted trouble, this kind of detection mode is more single, detection efficiency is low.
Disclosure of Invention
Aiming at the defects of the technology, the invention discloses a fault diagnosis method for a finely-divided fused new material mixing machine, which improves the fault diagnosis capability of the new material mixing machine by an intelligent fault detection method.
In order to achieve the technical effects, the invention adopts the following technical scheme:
a fault diagnosis method for a finely-divided fused new material mixer comprises the following steps:
acquiring a mechanical rotor assembly and an axial motion bearing measurement electric signal and operation state data information of a new material mixing machine;
classifying the acquired data information, and classifying and screening the data information of the running state of the mixing machine;
in this step, the failure detection means generates two types of vibrations, i.e., torsional vibration and radial vibration, for the detection failure of the power train; the fault detection device is matched with the vibration sensor for detecting faults of the transmission system; the vibration sensor senses radial vibration and is used for monitoring the state of a mechanical part in the rotary machine, measuring the amplitude of a sideband harmonic frequency component of a vibration signal and comparing the amplitude with the amplitude of adjacent harmonic frequency;
step three: extracting fault information, namely monitoring an electric signal of a generator with a material mixer bearing fault through a current sensor and a voltage sensor;
step four: denoising the acquired data information through a signal based on the combination of a Hankel matrix and singular value decomposition;
step five: and realizing the fault detection of the mixer through a fault diagnosis module, wherein the fault diagnosis module is a VMD-LSTM algorithm model.
As a further technical scheme of the invention, the fault detection device comprises a main control module, a sensor group connected with the main control module, a power supply module, an external memory, a communication interface, a signal filtering module, a signal amplifying module, a clock module, a Hall effect current sensor and a configuration module.
As a further technical scheme, the Hall-effect current sensor comprises an input end, a Hall current driving module, an analog switch, a dynamic compensation module, a sensing fine adjustment module, a signal recovery module and a current offset adjustment module, wherein the input end is connected with the analog switch, the analog switch is connected with the dynamic compensation module, the dynamic compensation module is connected with the signal recovery module, and the signal recovery module is further connected with an operational amplifier circuit.
As a further technical scheme of the invention, the current sensor is an ACS712 type current sensor, the nominal current of 0A to 5A is measured, and the sensitivity is 185 mV/A.
As a further technical scheme of the invention, the signal denoising method of the improved singular value decomposition comprises the following steps:
step (1): inputting new materials in the time domain analysis processHankel matrix of material mixer fault discrete signal, new material mixer fault signalSliding length by corresponding vector isWindow (2)The metering formula is as follows:
in the formula (1), the first and second groups of the compound,andall are the fault signal ordinal numbers of the new material mixing machine,is the total number of samples and is,the Hankel matrix in the formula (1) is used for singular value decomposition, and the method for quantizing the signal pulse characteristics is kurtosis which is expressed as:
in the formula (2), the first and second groups,、、andrespectively the kurtosis value, the probability density function, the standard deviation and the average value of the time domain signal of the new material mixer fault signal; obtaining a matrix related to the first singular valueThen, by performing an arithmetic mean calculation along the anti-diagonals of the matrix, the calculation function is expressed as:
in the formula (3), the first and second groups,representA maximum value parameter of the interval;to representA minimum value parameter of the interval; vector quantityRepresenting a denoised new material mixer fault time domain signal;
step (2): in the frequency domain analysis process, when a bearing of the new material mixer has a local fault, generating pulses at a basic fault frequency by an axial movement rotor assembly through the local fault, wherein the frequency spectrum of a fault signal comprises the fault frequency and harmonic waves of the mixer, and when local fault information exists on the rotor assembly, the local fault influences inner and outer axial roller paths;
using discrete Fourier transform to obtainThe obtained frequency spectrum removes noise and disperses Fourier series components of fault signalsy i Expressed as:
hankel matrix of discrete fault signal input similarly to the formula (1) to obtain matrix [ 2 ]A]Expressed as:
similar to equation (5), the spectrum vector of the discrete fault signal reconstruction process is:
the reconstructed spectrum vector in equation (6) is a denoised spectrum vector.
As a further technical scheme of the invention, the fault diagnosis model comprises a variational modal decomposition VMD model and a long-short term memory neural network LSTM model.
As a further technical scheme of the invention, the working method of the fault diagnosis model comprises the following steps:
the variable mode decomposition VMD model decomposes fault characteristic signals of the mixing machine, wavelet half-soft threshold denoising obtains denoising signals through wavelet coefficients subjected to reconstruction processing, the long and short term memory neural network LSTM model is responsible for extracting long mode information hidden in a mixing machine fault characteristic sequence, high-order information in a sequence mixing machine sample is mined, and the collected mixing machine fault characteristic signals are represented as follows:
in the formula (7), the first and second groups,showing mixing machinesThe dominant component of the fault signature information is,representing the interference-dominant component in the signal,a sequence number of the fault signal is indicated,which is indicative of the time of the fault sequence,representing the residual component of the fault, equation (7) represents the component of the mixer fault signal, the variational modal decomposition VMD model decomposes the fault signal into sub-signals and shifts the spectrum of the solid state mode to baseband, decomposing into:
in the formula (8), the first and second groups,a signal representative of a change in resolution of the mixer fault,representing a fourier transform of the mixer data information,representing the number of solid state modes of the signal resolved by the mixer,a signal indicative of a fault mode in operation of the mixer,representing the second penalty term coefficient in the mixer run,representing Lagrange operators in the calculation process of the state of the mixer, i representing the serial number of the decomposed sub-signals of the mixer, decomposing the original signals in the running state of the mixer into a plurality of eigenmode signals through a formula (8), and representing the function as:
in the formula (9), the reaction mixture,a wavelet based signal representative of the fault characteristics of the mixer,representing the wavelet coefficients after the processing,which is indicative of an initial fault sub-signal,representing the second penalty term coefficient in the mixer run,and (3) expressing the translation amount of the wavelet transform, completing the wavelet transform of the fault characteristic signal by a formula (9), wherein a wavelet half-soft threshold function can be expressed as follows:
in the formula (10), the first and second groups,representing the wavelet coefficients after thresholding,is shown smallThe threshold value of the wave is set to be,representing the second penalty term coefficient in the mixer run,
inputting the fault characteristic signal into the LSTM for convolution, and carrying out batch standardization processing and activating functions after a convolution module, wherein the function expression is as follows:
in the formula (11), the reaction mixture,a convolution module representing a fault identification model,a signal representative of the characteristic of the fault input,which represents the offset of the convolution module(s),which represents a process of standardization of a batch,representing the activation function, introducing non-linear features through equation (11) can represent more complex cases, and the new state of the LSTM module is represented as:
in the formula (12), the first and second groups,the output of the forgetting gate is represented,the status information of the last moment in time is represented,the output of the memory gate is represented,a new candidate vector is represented that is,
integrating the state information and the output signal, and the fault identification result output by the output gate of the LSTM module is expressed as:
in the formula (13), the first and second groups,an output vector representing the identification of the fault,a matrix of weights is represented by a matrix of weights,the output vector of the previous layer is represented,indicating the current output fault signature.
The invention has the following positive beneficial effects:
the invention obtains the measured electrical signal and the running state data information of the mechanical rotor component and the axial moving bearing of the new material mixing machine; classifying the acquired data information, and classifying and screening the data information of the running state of the mixing machine; the method comprises the steps of extracting fault information, monitoring an electric signal of a generator of the material mixer bearing fault through a current sensor and a voltage sensor, denoising acquired data information through a signal based on the combination of a Hankel matrix and singular value decomposition, and detecting the mixer fault through a fault diagnosis module, wherein the fault diagnosis module is a VMD-LSTM algorithm model.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without inventive exercise, wherein:
FIG. 1 is a schematic view of the overall structure of the present invention;
FIG. 2 is a circuit diagram of a Hall effect based current sensor according to the present invention;
fig. 3 is a schematic diagram of the fault detection apparatus of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, and it should be understood that the embodiments described herein are merely for the purpose of illustrating and explaining the present invention and are not intended to limit the present invention.
A fault diagnosis method for a finely-divided fused new material mixer comprises the following steps:
acquiring a mechanical rotor assembly and an axial motion bearing measurement electric signal and operation state data information of a new material mixing machine;
and a fault detection device is applied to the axial motion bearing to acquire bearing overheating, the motor rotor is not concentric with the mixer rotor, the axis deviates, the motor fault, the rotor is abnormal, lubricating oil is lacked, a sealing strip is damaged, air pressure of an air cylinder is insufficient or insufficient, a gear fault, a high-speed shaft gear fault, a high-speed intermediate gear fault, a planetary gear fault, a gear ring fault or a sun gear fault;
in this step, the rotor assembly is coupled to the generator through the axial motion bearing of the drive train, and the material-based mixer main controller receives the electrical signals from the current sensor and the voltage sensor to generate an electrical signal signature; receiving a vibration signal measured by a vibration sensor to generate a vibration signal signature; the vibration signal indicia and the electrical signal indicia are used to determine a material mixing machine fault detection signal, which may be representative of one or more diagnostic parameters;
classifying the obtained data information, and classifying and screening the mixer operation state data information;
in this step, the failure detection means generates two types of vibrations, i.e., torsional vibration and radial vibration, for the detection failure of the power train; the fault detection device is matched with the vibration sensor for detecting faults of the transmission system; the vibration sensor senses radial vibration and is used for monitoring the state of a mechanical part in the rotary machine, measuring the amplitude of a sideband harmonic frequency component of a vibration signal and comparing the amplitude with the amplitude of adjacent harmonic frequency;
step three: extracting fault information, namely monitoring an electric signal of a generator with a material mixer bearing fault through a current sensor and a voltage sensor; the electric signal is measured and sent to the controller by the electric sensor; the electrical signal is a voltage signal, the electrical signal is a current signal, the current signal is converted into the voltage signal through a Hall circuit to improve the signal analysis efficiency, the electrical characteristic code analysis of the electrical signal is formed by equipment based on a main controller of the material mixing machine, and one or more fault signals are generated through a computing unit;
step four: denoising the acquired data information through a signal based on the combination of a Hankel matrix and singular value decomposition; the algorithm is applied to the new material mixer fault signal and the frequency spectrum thereof, background noise is eliminated from two aspects of time domain and frequency domain, the reliability of the new material mixer fault diagnosis process is improved, and finally the fault diagnosis result display is realized through the server.
Step five: and realizing the fault detection of the mixer through a fault diagnosis module, wherein the fault diagnosis module is a VMD-LSTM algorithm model.
In the above embodiment, the fault detection apparatus includes a main control module, and a sensor group, a power module, an external memory, a communication interface, a signal filtering module, a signal amplifying module, a clock module, and a configuration module connected to the main control module.
In a specific embodiment, the sensor group includes a vibration sensor, a displacement sensor, a flow sensor, a speed sensor, a temperature sensor, a flame sensor, and the like, so as to obtain a plurality of fault data information. The main control module of the fault detection device uses an ARM Cortex-M332-bit STM32F103RBT6 chip, a built-in 20K RAM and an 8M crystal oscillator, a FLASH with 12-bit AD and 128KB, and resources including 4 16-bit timers, a 3-path serial communication interface and the like. The data acquisition module finishes acquisition of various electrical parameters such as voltage, electric power, frequency, power and the like by using a special metering chip ATT7022, has a specific data verification function to ensure acquisition accuracy, and is internally integrated with a high-precision ADC structure and data signal processing capacity. The ATT7022 special chip processor is used for measuring various information such as voltage effective value, current effective value, frequency measurement value, active power, reactive power, power factor and the like of the mixer motor. The signal filtering module can realize clutter data information interference in the circuit, the signal amplifying module can realize data information amplification, the clock module can output clock data information, and the data information is passed through
In the above embodiment, the current sensor with hall effect includes an input terminal, a hall current driving module, an analog switch, a dynamic compensation module, a sensing fine tuning module, a signal recovery module, and a current offset adjustment module, where the input terminal is connected to the analog switch, the analog switch is connected to the dynamic compensation module, the dynamic compensation module is connected to the signal recovery module, and the signal recovery module is further connected to an operational amplifier circuit.
In the above embodiment, the current sensor was an ACS712 type current sensor measuring a nominal current of 0A to 5A with a sensitivity of 185 mV/A.
In a specific embodiment, the Hall effect based current sensor circuit employs an ACS712 model current sensor, designed by the Worcester electronics plant, Mass., as a low cost alternative with off-the-shelf components that can monitor well the current encountered in most motor control circuits; the current sensor can measure nominal current from 0A to 5A, has the sensitivity of 185 mV/A, works under the power supply voltage of 5V, and has the bandwidth reaching 80 kHz; the generator is connected with the current sensor by adopting a copper conducting wire, and the path of the copper conducting wire is positioned near the surface of the current sensor; the current sensor does not support sleep mode and therefore includes an analog switch to control its turn off.
The current sensor consists of an accurate and low-offset linear Hall circuit, the current sensor is provided with four current signal input ports and is driven by Hall current after being input, and the Hall current drives an external open-circuit voltage VCC; the input current signal generates a magnetic field by dynamically compensating the current flowing through the Cooper conducting path, and the Hall circuit converts the magnetic field into proportional voltage; the current sensor measures the current flowing through the primary winding by using a Hall element, and the current in the secondary winding of the Hall circuit and the current in the primary winding are changed linearly; finally, the inverter amplifier with single-ended output converts the sensed current into a voltage signal; again, this output voltage is read by the internal ADC of the material mixer main controller; the offset of the current sensor is 2.5V, and the current for each phase is calculated as:I n =(analogread-compensation)/sensitivity, where analoglead is the input voltage of the material mixer master controller port;
the signal denoising method of the improved singular value decomposition comprises the following steps:
in a specific embodiment, the invention provides a signal denoising process based on a combination of a Hankel (Hankel) matrix and singular value decomposition, the algorithm is applied to a new material mixer fault signal and a frequency spectrum thereof, background noise is eliminated from two aspects of a time domain and a frequency domain, and the reliability of the new material mixer fault diagnosis process is improved, and the algorithm comprises the following steps:
step (1): in the time domain analysis process, a Hankel matrix of a fault discrete signal of the new material mixer is input, and a fault signal of the new material mixer is inputSliding length by corresponding vector isOf (2)Formula of measurement:
In the formula (1), the first and second groups,andall are the fault signal ordinal numbers of the new material mixing machine,is the total number of samples and is,the Hankel matrix in the formula (1) is used for singular value decomposition, and the method for quantizing the signal pulse characteristics is kurtosis, which is expressed as:
in the formula (2), the first and second groups,、、andrespectively the kurtosis value, the probability density function, the standard deviation and the average value of the time domain signal of the new material mixer fault signal; obtaining a matrix related to the first singular valueAfter that, by following the matrixThe inverse diagonals are calculated as the arithmetic mean, and the calculated function is expressed as:
in the formula (3), the first and second groups,to representA maximum value parameter of the interval;representA minimum value parameter of the interval; vector quantityRepresenting a denoised new material mixer fault time domain signal;
step (2): in the frequency domain analysis process, when a bearing of the new material mixer has a local fault, generating pulses at a basic fault frequency by an axial moving rotor assembly through the local fault, wherein the frequency spectrum of a fault signal comprises the fault frequency and harmonic waves of the mixer, and when local fault information exists on the rotor assembly, the local fault influences inner and outer axial roller paths;
denoising frequency spectrum obtained by using discrete Fourier transform, and Fourier series component of discrete fault signaly i Expressed as:
hankel matrix of discrete fault signal input similar to the formula (1) is obtained to obtain matrix 2A]Expressed as:
similar to equation (5), the spectrum vector of the discrete fault signal reconstruction process is:
the reconstructed spectrum vector in equation (6) is a denoised spectrum vector.
The fault diagnosis model comprises a variational modal decomposition VMD model and a long-short term memory neural network LSTM model.
The working method of the fault diagnosis model comprises the following steps:
the variable mode decomposition VMD model decomposes fault characteristic signals of the mixing machine, wavelet half-soft threshold denoising obtains denoising signals through wavelet coefficients subjected to reconstruction processing, the long and short term memory neural network LSTM model is responsible for extracting long mode information hidden in a mixing machine fault characteristic sequence, high-order information in a sequence mixing machine sample is mined, and the collected mixing machine fault characteristic signals are represented as follows:
in the formula (7), the first and second groups,a dominant component of fault characteristic information representing the mixing machine,representing the interference-dominant component in the signal,a sequence number of the fault signal is indicated,which is indicative of the time of the fault sequence,representing the residual component of the fault, equation (7) represents the component of the mixer fault signal, the variational modal decomposition VMD model decomposes the fault signal into sub-signals and shifts the spectrum of the solid state mode to baseband, decomposing into:
in the formula (8), the first and second groups,a signal representative of a change in resolution of the mixer fault,representing a fourier transform of the mixer data information,representing the number of solid state modes of the signal resolved by the mixer,a signal indicative of a fault mode in operation of the mixer,representing the second penalty term coefficient in the mixer run,representing Lagrange operators in the calculation process of the state of the mixer, i representing the serial number of the decomposed sub-signals of the mixer, decomposing the original signals in the running state of the mixer into a plurality of eigenmode signals through a formula (8), and representing the function as:
in the formula (9), the reaction mixture,a wavelet based signal representative of the fault characteristics of the mixer,representing the wavelet coefficients after the processing,which is indicative of an initial fault sub-signal,representing the second penalty term coefficient in the mixer run,the translation amount of wavelet transform is expressed, the wavelet transform of the fault characteristic signal is completed through a formula (9), and the wavelet half-soft threshold function can be expressed as:
in the formula (10), the first and second groups,representing the wavelet coefficients after thresholding,which represents a wavelet threshold value, is a function of,representing the second penalty term coefficient in the mixer run,
inputting the fault characteristic signal into the LSTM for convolution, and carrying out batch standardization processing and activating functions after a convolution module, wherein the function expression is as follows:
in the formula (11), the reaction mixture,a convolution module representing a fault identification model,a signal representative of the characteristic of the fault input,which represents the offset of the convolution module(s),which represents a batch normalization process, is shown,representing the activation function, introducing non-linear features through equation (11) can represent more complex cases, and the new state of the LSTM module is represented as:
in the formula (12), the first and second groups,the output of the forgetting gate is represented,the status information of the last moment in time is represented,the output of the memory gate is represented,a new candidate vector is represented that is,
integrating the state information and the output signal, and the fault identification result output by the output gate of the LSTM module is expressed as:
in the formula (13), the first and second groups,an output vector representing the identification of the fault,a matrix of weights is represented by a matrix of weights,the output vector of the previous layer is represented,indicating the current output fault signature.
And (4) finishing outputting the fault identification result of the LSTM neural network through a formula (13). In summary, the VMD decomposes the original mixer fault signature signal, transforms each modal signal to construct an analysis signal, removes noise components by wavelet transform, and inputs the analysis signal into the LSTM network for fault identification. The LSTM is provided with a plurality of equal-gate control units, so that the feature information and the state vector selectively pass through, the backward propagation of the fault feature can be carried out, and the long mode information hidden in the data can be better mined.
Although specific embodiments of the present invention have been described above, it will be understood by those skilled in the art that these specific embodiments are merely illustrative and that various omissions, substitutions and changes in the form of the detail of the methods and systems described above may be made by those skilled in the art without departing from the spirit and scope of the invention. For example, it is within the scope of the present invention to combine the steps of the methods described above to perform substantially the same function in substantially the same way to achieve substantially the same result. Accordingly, the scope of the invention is to be limited only by the following claims.
Claims (7)
1. A fault diagnosis method for a finely-divided fused new material mixer is characterized by comprising the following steps: the method comprises the following steps:
acquiring a mechanical rotor assembly and an axial motion bearing measurement electric signal and operation state data information of a new material mixing machine;
classifying the acquired data information, and classifying and screening the data information of the running state of the mixing machine;
in this step, the failure detection means generates two types of vibrations, i.e., torsional vibration and radial vibration, for the detection failure of the power train; the fault detection device is matched with the vibration sensor for detecting faults of the transmission system; the vibration sensor senses radial vibration and is used for monitoring the state of a mechanical part in the rotary machine, measuring the amplitude of a sideband harmonic frequency component of a vibration signal and comparing the amplitude with the amplitude of adjacent harmonic frequency;
step three: extracting fault information, namely monitoring an electric signal of a generator with a material mixer bearing fault through a current sensor and a voltage sensor;
step four: denoising the acquired data information through a signal based on the combination of a Hankel matrix and singular value decomposition;
step five: and realizing the fault detection of the mixer through a fault diagnosis module, wherein the fault diagnosis module is a VMD-LSTM algorithm model.
2. The method of claim 1, wherein the method comprises the steps of:
the fault detection device comprises a main control module, a sensor group, a power supply module, an external memory, a communication interface, a signal filtering module, a signal amplifying module, a clock module, a Hall-effect current sensor and a configuration module, wherein the sensor group, the power supply module, the external memory, the communication interface, the signal filtering module, the signal amplifying module, the clock module, the Hall-effect current sensor and the configuration module are connected with the main control module.
3. The method of claim 1, wherein the method comprises the steps of:
the current sensor with the Hall effect comprises an input end, a Hall current driving module, an analog switch, a dynamic compensation module, a sensing fine adjustment module, a signal recovery module and a current offset adjustment module, wherein the input end is connected with the analog switch, the analog switch is connected with the dynamic compensation module, the dynamic compensation module is connected with the signal recovery module, and the signal recovery module is further connected with an operational amplifier amplification circuit.
4. The method of claim 3, wherein the method comprises the steps of:
the current sensor was an ACS712 model current sensor measuring nominal currents of 0A to 5A with a sensitivity of 185 mV/A.
5. The method of claim 4, wherein the method comprises the steps of: the signal denoising method of the improved singular value decomposition comprises the following steps:
step (1): in the time domain analysis process, a Hankel matrix of a new material mixer fault discrete signal and a new material mixer fault signal are inputSliding length by corresponding vector isOf (2)The metering formula is as follows:
in the formula (1), the first and second groups,andall are the fault signal ordinal numbers of the new material mixing machine,is the total number of samples and is,the Hankel matrix in the formula (1) is used for singular value decomposition, and the method for quantizing the signal pulse characteristics is kurtosis, which is expressed as:
in the formula (2), the first and second groups,、、andrespectively the kurtosis value, the probability density function, the standard deviation and the average value of the time domain signal of the new material mixer fault signal; in obtaining a matrix associated with the first singular valueThen, by performing an arithmetic mean calculation along the anti-diagonals of the matrix, the calculation function is expressed as:
in the formula (3), the first and second groups,to representA maximum value parameter of the interval;to representA minimum value parameter of the interval; vector quantityRepresenting a denoised new material mixer fault time domain signal;
step (2): in the frequency domain analysis process, when a bearing of the new material mixer has a local fault, generating pulses at a basic fault frequency by an axial moving rotor assembly through the local fault, wherein the frequency spectrum of a fault signal comprises the fault frequency and harmonic waves of the mixer, and when local fault information exists on the rotor assembly, the local fault influences inner and outer axial roller paths;
denoising frequency spectrum obtained by using discrete Fourier transform, and Fourier series component of discrete fault signaly i Expressed as:
hankel matrix of discrete fault signal input similar to the formula (1) is obtained to obtain matrix 2A]Expressed as:
similar to equation (5), the spectrum vector of the discrete fault signal reconstruction process is:
the reconstructed spectrum vector in equation (6) is a denoised spectrum vector.
6. The method of claim 4, wherein the method comprises the steps of: the fault diagnosis model comprises a variational modal decomposition VMD model and a long-short term memory neural network LSTM model.
7. The method of claim 6, wherein the method comprises the steps of: the working method of the fault diagnosis model comprises the following steps:
the variable mode decomposition VMD model decomposes fault characteristic signals of the mixing machine, wavelet half-soft threshold denoising obtains denoising signals through wavelet coefficients subjected to reconstruction processing, the long and short term memory neural network LSTM model is responsible for extracting long mode information hidden in a mixing machine fault characteristic sequence, high-order information in a sequence mixing machine sample is mined, and the collected mixing machine fault characteristic signals are represented as follows:
in the formula (7), the first and second groups,representing the dominant component of the fault characteristic information of the mixing machine,representing the dominant component of interference in the signal,a sequence number of the fault signal is indicated,which is indicative of the time of the fault sequence,representing the residual component of the fault, equation (7) represents the component of the mixer fault signal, the variational modal decomposition VMD model decomposes the fault signal into sub-signals and shifts the spectrum of the solid state mode to baseband, decomposing into:
in the formula (8), the first and second groups,a translation resolution signal indicative of a mixer fault,representing a fourier transform of the mixer data information,representing the number of solid state modes of the signal resolved by the mixer,a signal indicative of a fault mode in operation of the mixer,representing the second penalty term coefficient in the mixer run,representing Lagrange operators in the calculation process of the state of the mixer, i representing the serial number of the decomposed sub-signals of the mixer, decomposing the original signals in the running state of the mixer into a plurality of eigenmode signals through a formula (8), and representing the function as:
in the formula (9), the reaction mixture,a wavelet based signal representative of the fault characteristics of the mixer,representing the wavelet coefficients after the processing,which is indicative of an initial fault sub-signal,representing the second order penalty term coefficient in the mixer operation,the translation amount of wavelet transform is expressed, the wavelet transform of the fault characteristic signal is completed through a formula (9), and the wavelet half-soft threshold function can be expressed as:
in the formula (10), the first and second groups,representing the wavelet coefficients after thresholding,which represents a wavelet threshold value, is a value,representing the second order penalty term coefficient in the mixer operation,
inputting the fault characteristic signal into the LSTM for convolution, and carrying out batch standardization processing and activating functions after a convolution module, wherein the function expression is as follows:
in the formula (11), the reaction mixture is,a convolution module representing a fault identification model,a signal representative of the characteristic of the fault input,which represents the offset of the convolution module(s),which represents a batch normalization process, is shown,representing the activation function, introducing non-linear features through equation (11) can represent more complex cases, and the new state of the LSTM module is represented as:
in the formula (12), the first and second groups,the output of the forgetting gate is represented,the status information of the last moment in time is represented,the output of the memory gate is represented,a new candidate vector is represented that is,
integrating the state information and the output signal, and the fault identification result output by the output gate of the LSTM module is expressed as:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210827199.XA CN115015756A (en) | 2022-07-14 | 2022-07-14 | Fault diagnosis method for fine-crushing fused new material mixer |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210827199.XA CN115015756A (en) | 2022-07-14 | 2022-07-14 | Fault diagnosis method for fine-crushing fused new material mixer |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115015756A true CN115015756A (en) | 2022-09-06 |
Family
ID=83081463
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210827199.XA Pending CN115015756A (en) | 2022-07-14 | 2022-07-14 | Fault diagnosis method for fine-crushing fused new material mixer |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115015756A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116736116A (en) * | 2023-08-15 | 2023-09-12 | 泰坦(天津)能源技术有限公司 | Fault sensing method and system for miniature motor |
CN117330816A (en) * | 2023-12-01 | 2024-01-02 | 南京中旭电子科技有限公司 | Monitoring data optimization method for Hall current sensor |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140230555A1 (en) * | 2012-12-20 | 2014-08-21 | Zapadoceska Univerzita V Plzni | Method of detecting and localizing partial rotor-stator rubbing during the operation of a turbine |
CN109855851A (en) * | 2019-01-25 | 2019-06-07 | 辽宁欣科电气股份有限公司 | A kind of A.C. contactor measuring mechanical characteristics method and device |
CN110082841A (en) * | 2019-04-18 | 2019-08-02 | 东华大学 | A kind of short-term wind speed forecasting method |
CN110848165A (en) * | 2019-11-04 | 2020-02-28 | 江苏科技大学 | Centrifugal pump mechanical seal fault diagnosis method and device based on sensorless monitoring technology |
CN113269169A (en) * | 2021-07-19 | 2021-08-17 | 武汉恩为科技有限公司 | Bearing fault detection method and device |
CN113495229A (en) * | 2020-04-01 | 2021-10-12 | 卡特彼勒公司 | System and method for detecting generator winding faults |
WO2022099855A1 (en) * | 2020-11-13 | 2022-05-19 | 烟台杰瑞石油装备技术有限公司 | Motor fault monitoring device, driving motor system, and motor fault monitoring method |
CN114638266A (en) * | 2022-03-21 | 2022-06-17 | 上海电力大学 | VMD-WT-CNN-based multi-fault coupling signal processing and diagnosis method for gas turbine rotor |
-
2022
- 2022-07-14 CN CN202210827199.XA patent/CN115015756A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140230555A1 (en) * | 2012-12-20 | 2014-08-21 | Zapadoceska Univerzita V Plzni | Method of detecting and localizing partial rotor-stator rubbing during the operation of a turbine |
CN109855851A (en) * | 2019-01-25 | 2019-06-07 | 辽宁欣科电气股份有限公司 | A kind of A.C. contactor measuring mechanical characteristics method and device |
CN110082841A (en) * | 2019-04-18 | 2019-08-02 | 东华大学 | A kind of short-term wind speed forecasting method |
CN110848165A (en) * | 2019-11-04 | 2020-02-28 | 江苏科技大学 | Centrifugal pump mechanical seal fault diagnosis method and device based on sensorless monitoring technology |
CN113495229A (en) * | 2020-04-01 | 2021-10-12 | 卡特彼勒公司 | System and method for detecting generator winding faults |
WO2022099855A1 (en) * | 2020-11-13 | 2022-05-19 | 烟台杰瑞石油装备技术有限公司 | Motor fault monitoring device, driving motor system, and motor fault monitoring method |
CN113269169A (en) * | 2021-07-19 | 2021-08-17 | 武汉恩为科技有限公司 | Bearing fault detection method and device |
CN114638266A (en) * | 2022-03-21 | 2022-06-17 | 上海电力大学 | VMD-WT-CNN-based multi-fault coupling signal processing and diagnosis method for gas turbine rotor |
Non-Patent Citations (2)
Title |
---|
GANGSHI ETAL.: "A VMD-EWT-LSTM-based multi-step prediction approach for shield tunneling machine cutterhead torque", 《KNOWLEDGE-BASED SYSTEMS》 * |
梁治华: "基于数据的旋转机械故障诊断和性能评估方法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116736116A (en) * | 2023-08-15 | 2023-09-12 | 泰坦(天津)能源技术有限公司 | Fault sensing method and system for miniature motor |
CN116736116B (en) * | 2023-08-15 | 2023-10-24 | 泰坦(天津)能源技术有限公司 | Fault sensing method and system for miniature motor |
CN117330816A (en) * | 2023-12-01 | 2024-01-02 | 南京中旭电子科技有限公司 | Monitoring data optimization method for Hall current sensor |
CN117330816B (en) * | 2023-12-01 | 2024-01-26 | 南京中旭电子科技有限公司 | Monitoring data optimization method for Hall current sensor |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115015756A (en) | Fault diagnosis method for fine-crushing fused new material mixer | |
Bessous et al. | Diagnosis of bearing defects in induction motors using discrete wavelet transform | |
Ordaz-Moreno et al. | Automatic online diagnosis algorithm for broken-bar detection on induction motors based on discrete wavelet transform for FPGA implementation | |
Wang et al. | A smart sensing unit for vibration measurement and monitoring | |
Contreras-Medina et al. | FPGA-based multiple-channel vibration analyzer for industrial applications in induction motor failure detection | |
CN104819841B (en) | Built-in-coding-information-based single sensing flexible angle-domain averaging method | |
JP2003528292A (en) | State-based monitoring of bearings by vibration analysis | |
Corne et al. | Comparing MCSA with vibration analysis in order to detect bearing faults—A case study | |
CN110553844A (en) | Method and system for detecting misalignment fault of rotary machine | |
Balakrishna et al. | An autonomous electrical signature analysis-based method for faults monitoring in industrial motors | |
CN105651412A (en) | Measurement method and measurement circuit for PT1000 temperature sensor | |
CN110161406A (en) | A kind of vibrating motor current mode failure diagnostic apparatus and diagnostic method | |
Sakhalkar et al. | Fault detection in induction motors based on motor current signature analysis and accelerometer | |
CN112115802A (en) | Crane slewing mechanism gear fault diagnosis method, system and storage medium | |
Ibragimov et al. | Analysis of the methods of diagnosing asynchronous motors according to vibration indicators | |
GB2122749A (en) | Electrical condition monitoring of electric motors | |
JP2011237459A (en) | Abnormality diagnosis device, rotary device and abnormality diagnosis method | |
CN105784364A (en) | Bearing fault diagnosis method based on total experience mode decomposition and fractal box dimensions | |
CN111855192A (en) | Singular value decomposition method for denoising encoder signal | |
CN114486252B (en) | Rolling bearing fault diagnosis method of vector mode maximum envelope | |
JP4848803B2 (en) | Abnormality diagnosis device, rotation device, and abnormality diagnosis method | |
Rodriguez-Donate et al. | FPGA based embedded system for induction motor failure monitoring at the start-up transient vibrations with wavelets | |
Sharma et al. | Evaluation of arduino based das for condition monitoring of induction motor | |
CN102713525B (en) | The methods, devices and systems that the rotor angle of rotary shaft is determined by monitoring by means of decomposer | |
Ágoston | Studying and measuring system for motor base unbalance |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220906 |
|
RJ01 | Rejection of invention patent application after publication |