CN112947368A - Fault diagnosis device and method for three-phase asynchronous motor - Google Patents

Fault diagnosis device and method for three-phase asynchronous motor Download PDF

Info

Publication number
CN112947368A
CN112947368A CN202110145260.8A CN202110145260A CN112947368A CN 112947368 A CN112947368 A CN 112947368A CN 202110145260 A CN202110145260 A CN 202110145260A CN 112947368 A CN112947368 A CN 112947368A
Authority
CN
China
Prior art keywords
asynchronous motor
phase asynchronous
fault diagnosis
data set
training
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
Application number
CN202110145260.8A
Other languages
Chinese (zh)
Other versions
CN112947368B (en
Inventor
黄友锐
杭俊
徐善永
丁石川
韩涛
周小杰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Anhui University of Science and Technology
Anhui University
Original Assignee
Anhui University of Science and Technology
Anhui University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Anhui University of Science and Technology, Anhui University filed Critical Anhui University of Science and Technology
Priority to CN202110145260.8A priority Critical patent/CN112947368B/en
Priority to LU500017A priority patent/LU500017B1/en
Publication of CN112947368A publication Critical patent/CN112947368A/en
Application granted granted Critical
Publication of CN112947368B publication Critical patent/CN112947368B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00—Testing or monitoring of control systems or parts thereof
    • G05B23/02—Electric testing or monitoring
    • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
    • G05B23/0262—Confirmation of fault detection, e.g. extra checks to confirm that a failure has indeed occurred
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00—Testing or monitoring of control systems or parts thereof
    • G05B23/02—Electric testing or monitoring
    • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
    • G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
    • G05B23/024—Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
    • 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
    • G01R31/343—Testing dynamo-electric machines in operation
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00—Testing or monitoring of control systems or parts thereof
    • G05B23/02—Electric testing or monitoring
    • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
    • G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
    • G05B23/0254—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model based on a quantitative model, e.g. mathematical relationships between inputs and outputs; functions: observer, Kalman filter, residual calculation, Neural Networks
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/20—Pc systems
    • G05B2219/24—Pc safety
    • G05B2219/24065—Real time diagnostics

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • Control Of Electric Motors In General (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

The invention discloses a fault diagnosis device and a method for a three-phase asynchronous motor, which comprises a terminal sensing system for sensing various electrical parameters of the three-phase asynchronous motor; the remote data center is used for constructing a digital twin body and a fault diagnosis model of the three-phase asynchronous motor; a three-phase asynchronous motor fault diagnosis model is constructed through deep learning, and is trained through a large amount of abundant data sets generated by a digital twin body, so that the accuracy of fault diagnosis model detection is greatly improved; through transfer learning, the fault diagnosis model can be updated only by a small amount of new samples, and the robustness and the stability of the fault diagnosis model detection of the three-phase asynchronous motor are improved.

Description

Fault diagnosis device and method for three-phase asynchronous motor
Technical Field
The invention relates to the field of a fault diagnosis device and a fault diagnosis method for a three-phase asynchronous motor, in particular to a fault diagnosis device and a fault diagnosis method for a three-phase asynchronous motor.
Background
Once the three-phase asynchronous motor fails, the whole industrial operation system may face the risk of comprehensive breakdown, causing unpredictable economic loss and even endangering the life safety of field workers. Therefore, the fault detection and fault diagnosis of the three-phase asynchronous motor have important research significance.
At present, the fault diagnosis of the three-phase asynchronous motor is mostly in a field on-site detection diagnosis stage, and a small part of the fault diagnosis starts to use remote virtual simulation diagnosis. The field on-site diagnosis requires expert personnel to go to a working field for measurement and diagnosis, and is limited by space and low in working efficiency. The remote virtual simulation is limited by a sensor and a communication technology, and cannot be mapped in real time, so that the accuracy of fault diagnosis of the three-phase asynchronous motor is low, and the real-time performance is poor. With the continuous development of big data and artificial intelligence technology, fault diagnosis methods of three-phase asynchronous motors are also continuously advanced, the existing fault diagnosis method of three-phase asynchronous motors based on data driving needs to collect a large amount of actual data before training, but the three-phase asynchronous motors are inevitably influenced by self abrasion, working environment change and the like in the use process, the previously trained fault diagnosis model is not suitable any more, the training is repeated, a large amount of running data of the three-phase asynchronous motors in various states needs to be collected again, the time consumption is long, and the efficiency is low.
Disclosure of Invention
In order to solve the above-mentioned drawbacks in the background art, the present invention provides a fault diagnosis apparatus and method for a three-phase asynchronous motor.
The purpose of the invention can be realized by the following technical scheme:
a three-phase asynchronous motor fault diagnosis device and method, it is made up of terminal sensing system and remote data center, the said terminal sensing system is made up of ARM processor, raspberry group module, vibration sensor X, vibration sensor Y, vibration sensor Z, current transformer A, current transformer B, current transformer C, voltage converter A, voltage converter B, voltage converter C, sound sensor, rotational speed sensor, infrared thermal camera, 3D stereoscopic camera and 5G communication module 1;
the remote data center consists of a server and a 5G communication module 2; the vibration sensor X, the vibration sensor Y, the vibration sensor Z, the current transformer A, the current transformer B, the current transformer C, the voltage converter A, the voltage converter B, the voltage converter C and the sound sensor are respectively connected with different AD interfaces of the ARM processor; the rotating speed sensor is connected with a GPIO interface of the ARM processor; the ARM processor, the infrared thermal camera, the 3D stereo camera and the 5G communication module 1 are respectively connected with different USB interfaces of the raspberry pi module; the 5G communication module 2 is connected with the server through a USB interface; the terminal sensing system is connected with the remote data center through a mobile internet.
A three-phase asynchronous motor fault diagnosis device and method, it is made up of virtual and real fusion stage, fault diagnosis stage and model updating stage, the said virtual and real fusion stage is formed by data perception and digital twinning; the fault diagnosis stage consists of a fault diagnosis model, offline training and online detection; the model updating phase consists of updating detection and transfer learning.
Further, the data sensing module is characterized in that the ARM processor respectively collects three-phase current values IA, IB and IC of the three-phase asynchronous motor through three current transformers, respectively collects three-phase voltage values UA, UB and UC of the three-phase asynchronous motor through three voltage transformers, respectively collects vibration values HX, HY and HZ in three directions of the three-phase asynchronous motor X, Y, Z through three vibration sensors, respectively collects sound information Z of the three-phase asynchronous motor through a sound sensor, measures the rotating speed V of the three-phase asynchronous motor through a rotating speed sensor, and sends the rotating speed V to the raspberry sending processing module through a USB interface; the raspberry processing module acquires temperature information T of the three-phase asynchronous motor through an infrared thermal camera, acquires the three-dimensional size, shape and surface texture of the three-phase asynchronous motor through a 3D (three-dimensional) camera, and sends the three-dimensional size, shape and surface texture together with IA (internal information), IB (internal information), IC (integrated circuit), UA (user agent), UB (user agent), UC (user agent), HX (human X), HY (human H), HZ (human Z), Z and V (human V) sent by an ARM (advanced RISC machines) processor to; the server of the remote data center can acquire all information such as voltage, current, vibration, sound, rotating speed, temperature, three-dimensional size, shape, surface texture and the like of the three-phase asynchronous motor in real time through the 5G communication module 2;
the digital twin is characterized in that three-dimensional virtual reconstruction is carried out on the three-dimensional size, shape and surface texture information of the real three-phase asynchronous motor acquired in real time on the server through three-dimensional simulation software, the voltage, current, vibration, sound, rotating speed and temperature data of the real three-phase asynchronous motor are acquired in real time, the running condition of a physical space is displayed in real time in a virtual space, and the virtual three-phase asynchronous motor and the real three-phase asynchronous motor are mapped in real time, so that the digital twin of the three-phase asynchronous motor is formed.
Further, the fault diagnosis model is a deep feedforward neural network realized in a server by software, and the structure of the fault diagnosis model is composed of an input layer 0, a hidden layer 1, a hidden layer 2, a hidden layer 3, a hidden layer 4, a hidden layer 5 and a Softmax layer; input layer 0 contains 12 neurons
Figure BDA0002929795950000031
Input X ═ IA, IB, IC, UA, UB, UC, HX, HY, HZ, Z, V, T for 12 operating parameters of the three-phase asynchronous motor]T, the hidden layers 1 to 5 respectively comprise 15 neurons, 12 neurons, 10 neurons, 8 neurons and 5 neurons, and are mainly used for feature extraction and compression of input parameters; input layer a [0 ]]＝X，a[0]A is obtained after passing through a hidden layer 1 to a hidden layer 5[ly]＝Sigmoid(w[ly]a[ly-1]+b[ly]) Ly e (1,2,3,4,5), the output through the Softmax layer is
Figure BDA0002929795950000032
The method is used for predicting the probabilities of the five three-phase asynchronous motor states, selecting the state with the highest probability as the final three-phase asynchronous motor fault diagnosis result,
Figure BDA0002929795950000033
wherein, w [ ly]And b [ ly]Ly belongs to (1,2,3,4,5) as unknown quantity, and needs to be determined after training;
Figure BDA0002929795950000034
function representation parameter taking
Figure BDA0002929795950000041
The maximum sigmoid function of (A) is calculated as
Figure BDA0002929795950000042
The off-line training comprises establishing a training data set, a testing data set and a training fault diagnosis model determining parameter;
firstly, using digital twin body simulation to obtain N running parameters X of the three-phase asynchronous motor corresponding to five conditions of 'normal', 'stator fault', 'rotor fault', 'bearing fault' and 'air gap eccentricity', wherein N is not less than 800, and forming a state data set of the three-phase asynchronous motor; secondly, a Label (Label) value table of five three-phase asynchronous motor states is established: the method comprises the following steps of 1,2 for "normal", 3 for "stator fault", 4 for "bearing fault", and 5 for air gap eccentricity; thirdly, setting a corresponding state label value for each three-phase asynchronous motor operation parameter X; finally, 400 running parameters of the three-phase asynchronous motor in five states of 'normal', 'stator fault', 'rotor fault', 'bearing fault' and 'air gap eccentricity' are selected from the state data set of the three-phase asynchronous motor, wherein the running parameters are 2000 in total and are placed according to a random sequence, and the 2000 running parameters X and corresponding label values thereof form a training data set; according to the same method, randomly selecting 200 operation parameters X from the remaining operation parameters X of the state data set of the three-phase asynchronous motor, placing the operation parameters according to a random sequence, and taking the 200 operation parameters and corresponding label values thereof as a test data set;
the process of training the fault diagnosis model to determine parameters is as follows:
(1-1) initializing parameters, setting w [ ly ], b [ ly ], ly epsilon (1,2,3,4,5) as random values, setting the iteration number as S, the learning rate as Lr, a diagnosis accuracy threshold value Tac, setting the operation parameter of the jth three-phase asynchronous motor in the training data set as X (j), and setting j as 1;
(1-2) inputting the three-phase asynchronous motor operation parameters X (j) into a fault diagnosis model, and calculating the estimated value of the detection
Figure BDA0002929795950000044
(1-3) according to the label value y corresponding to X (j) and the calculated estimation value
Figure BDA0002929795950000045
Computing cross entropy loss function
Figure BDA0002929795950000043
(1-4) calculating each parameter w [ ly ] in each layer of the fault diagnosis model]And b [ ly]Change value of delta w [ ly]And Δ b [ ly]，
Figure BDA0002929795950000051
(1-5) according to the formula w[ly]＝w[ly]-Lr*Δw[ly]，b[ly]＝b[ly]-Lr*Δb[ly]Update w [ ly]And b [ ly]；
(1-6) judging whether the data is the last data in the training data set, if not, inputting the next data, and returning to (4-2); otherwise, turning to (4-7) to calculate the detection accuracy;
(1-7) inputting the operating parameters of the test data set, and calculating a detection estimation value of each operating parameter
Figure BDA0002929795950000054
Comparing with the corresponding label value y to calculate the detection accuracy
Figure BDA0002929795950000052
Wherein
Figure BDA0002929795950000053
Representing the number of the operation parameters with the same detection estimated value and label value, and sigma num (y) representing the total number of the operation parameters in the test data set;
(1-8) judging whether the detection accuracy rate meets the requirement, and if Ac is more than or equal to Tac, turning to (4-9) and finishing training; if Ac is less than Tac, judging whether the iteration times are finished, if S is not equal to 0, turning to (4-2), and reusing the training data set to perform a new round of training until the iteration is finished; if S is equal to 0, finishing the training;
(1-9) storing all parameters w [ ly ] and b [ ly ], and finishing the training of the fault diagnosis model;
the online detection is characterized in that the operation parameter X of the three-phase asynchronous motor is obtained in real time, the operation state of the three-phase asynchronous motor is calculated through a trained fault diagnosis model, and the real-time fault diagnosis of the three-phase asynchronous motor is realized.
Further, the model updating phase consists of updating detection and transfer learning;
the updating detection is characterized in that a fault diagnosis error threshold Te of the three-phase asynchronous motor is set, when the fault diagnosis model detects the three-phase asynchronous motor on line, the condition that the detection result is wrong is recorded to form a migration data set Xe, when the quantity Ce of the migration data set is more than or equal to Te, migration learning is carried out to update the fault diagnosis model, otherwise, model updating is not carried out;
the process of the transfer learning is as follows:
(2-1) initializing parameters, setting fault diagnosis model parameters w [ ly ], b [ ly ], ly belonging to (1,2,3,4,5), setting the iteration times to be ST, the learning rate to be Ltr, the global loss function threshold to be TG, the operation parameter of the jth three-phase asynchronous motor in the training data set to be X (j), setting the operation parameter of the jth three-phase asynchronous motor in the migration data set to be Xe (j), and setting j to be 1;
(2-2) inputting X (j) into a fault diagnosis model, and calculating an estimated value of detection thereof
Figure BDA0002929795950000068
And obtaining the characteristic value output by the hidden layer 4Is recorded as XS;
(2-3) inputting Xe (j) into a fault diagnosis model, and acquiring a characteristic value output by the hidden layer 4 and recording the characteristic value as XT; (2-4) according to the label value y and the estimation value corresponding to X (j)
Figure BDA0002929795950000061
Computing cross entropy loss function
Figure BDA0002929795950000062
Figure BDA0002929795950000063
(2-5) calculating a domain loss function LD (XS, XT) according to XS and XT,
Figure BDA0002929795950000064
wherein φ (·) > X → H represents a feature mapping from a feature space to a regenerative kernel Hilbert space;
(2-6) according to
Figure BDA0002929795950000065
And LD (XS, XT), computing a global loss function LG,
Figure BDA0002929795950000066
wherein λ represents a weight coefficient occupied by migration to the target domain;
(2-7) calculating each parameter w [ ly ] in each layer of the failure diagnosis model]And b [ ly]Change value of delta w [ ly]And Δ b [ ly]，
Figure BDA0002929795950000067
(2-8) according to the formula w[ly]＝w[ly]-Ltr*Δw[ly]，b[ly]＝b[ly]-Ltr*Δb[ly]Update w [ ly]And b [ ly]；
(2-9) judging whether the data is the last data in the data set, if not, inputting the next data, and returning to (5-2); otherwise, go to (2-10);
(2-10) judging whether the global loss function value meets the requirement, and if LG is less than or equal to TG, turning to (5-11) and finishing transfer learning; if LG is larger than TG, judging whether iteration times are finished, if S is not equal to 0, turning to (5-2), and reusing the training data set and the migration data set to perform a new round of training until iteration is finished; if S is equal to 0, the training is finished, and the process goes to (5-11);
and (2-11) saving all parameters w [ ly ] and b [ ly ] and finishing updating the fault diagnosis model.
The invention has the beneficial effects that:
1. information such as three-phase current, three-phase voltage, X, Y, Z three-direction vibration, abnormal sound, rotating speed, temperature distribution, three-dimensional geometric shape, surface theory and the like of the three-phase asynchronous motor can be sent to a remote server in real time through a current transformer, a voltage converter, a vibration sensor, a sound sensor, a rotating speed sensor, an infrared thermal camera, a 3D stereo camera and a 5G communication module, the digital twin body of the three-phase asynchronous motor is constructed in a virtual space, real-time mapping of the virtual space and a physical space is realized, a large amount of running state data of the three-phase asynchronous motor can be obtained in a short time through virtual-real fusion of the digital twin body, and a rich training data set is provided for training a fault diagnosis model.
2. A three-phase asynchronous motor fault diagnosis model is constructed through deep learning, and is trained through a large amount of abundant data sets generated by a digital twin body, so that the accuracy of fault diagnosis model detection is greatly improved.
3. With the long-term operation of the three-phase asynchronous motor, the fault diagnosis model can be updated only by a small amount of new samples through transfer learning, and the detection robustness and stability of the fault diagnosis model of the three-phase asynchronous motor are improved.
Drawings
The invention will be further described with reference to the accompanying drawings.
FIG. 1 is a schematic diagram of the process of the present invention;
FIG. 2 is a schematic diagram of the structure of the apparatus of the present invention;
FIG. 3 is a schematic structural diagram of a fault diagnosis model according to the method of the present invention;
FIG. 4 is a schematic diagram of the construction of a training data set and a test data set according to the method of the present invention;
FIG. 5 is a flow chart of a fault diagnosis model training process of the method of the present invention;
FIG. 6 is a schematic diagram of a migration model according to the present invention;
FIG. 7 is a flow chart of transfer learning according to the method of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. 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 invention.
In the description of the present invention, it is to be understood that the terms "opening," "upper," "lower," "thickness," "top," "middle," "length," "inner," "peripheral," and the like are used in an orientation or positional relationship that is merely for convenience in describing and simplifying the description, and do not indicate or imply that the referenced component or element must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be considered as limiting the present invention.
As shown in fig. 1, the method for diagnosing a fault of a three-phase asynchronous motor is schematically illustrated, and is characterized by comprising a virtual-real fusion stage, a fault diagnosis stage and a model update stage; the virtual-real fusion stage consists of a data sensing part and a digital twin part; the fault diagnosis stage consists of a fault diagnosis model, an off-line training part and an on-line detection part; the model updating phase is composed of an updating detection part and a transfer learning part. During operation, firstly, in a virtual-real fusion stage, a data sensing part acquires abundant running data of the three-phase asynchronous motor in real time, so that a digital twin body is constructed in a virtual space, and a large number of running state parameters of the three-phase asynchronous motor are generated through real-time mapping and simulation of a physical space three-phase asynchronous motor entity and the virtual space digital twin body to form a three-phase asynchronous motor state data set; secondly, in a fault diagnosis stage, a fault diagnosis model is built based on a deep feedforward neural network, the fault diagnosis model is trained and tested by using a three-phase asynchronous motor state data set generated in a virtual-real fusion stage, and after training is finished, the fault diagnosis model can be used for carrying out real-time fault detection on the three-phase asynchronous motor; and finally, in the model updating stage, the updating detection part judges whether the fault diagnosis model needs to be updated according to the number of the detection errors of the fault diagnosis model, and when the number of the detection errors of the fault reaches a certain threshold value, migration learning is carried out by using the data of the detection errors, the fault diagnosis model is updated, so that the detection accuracy is improved, and the detection robustness is enhanced.
As shown in fig. 2, the device and the method for diagnosing the fault of the three-phase asynchronous motor are schematically shown in the composition structure, and are characterized by comprising a terminal sensing system and a remote data center; the terminal sensing system consists of an ARM processor, a raspberry group module, a vibration sensor X, a vibration sensor Y, a vibration sensor Z, a current transformer A, a current transformer B, a current transformer C, a voltage converter A, a voltage converter B, a voltage converter C, a sound sensor, a rotation speed sensor, an infrared thermal camera, a 3D stereo camera and a 5G communication module 1; the remote data center consists of a server and a 5G communication module 2; the vibration sensor X, the vibration sensor Y, the vibration sensor Z, the current transformer A, the current transformer B, the current transformer C, the voltage converter A, the voltage converter B, the voltage converter C and the sound sensor are respectively connected with different AD interfaces of the ARM processor; the rotating speed sensor is connected with a GPIO interface of the ARM processor; the ARM processor, the infrared thermal camera, the 3D stereo camera and the 5G communication module 1 are respectively connected with different USB interfaces of the raspberry pi module; the 5G communication module 2 is connected with the server through a USB interface; the terminal sensing system is connected with the remote data center through a mobile internet.
The ARM processor respectively collects three-phase current values IA, IB and IC of the three-phase asynchronous motor through three current transformers, respectively collects three-phase voltage values UA, UB and UC of the three-phase asynchronous motor through three voltage transformers, respectively collects vibration values HX, HY and HZ in three directions of X, Y, Z of the three-phase asynchronous motor through three vibration sensors, respectively collects sound information Z of the three-phase asynchronous motor through a sound sensor, measures the rotating speed V of the three-phase asynchronous motor through a rotating speed sensor, and sends the rotating speed V to the raspberry group processing module through a USB interface; the raspberry processing module acquires temperature information T of the three-phase asynchronous motor through the infrared thermal camera, acquires the three-dimensional size, shape and surface texture of the three-phase asynchronous motor through the 3D stereo camera, and sends the three-dimensional size, shape and surface texture together with IA, IB, IC, UA, UB, UC, HX, HY, HZ, Z and V sent by the ARM processor to the mobile internet in real time through the 5G communication module 1; finally, a server of the remote data center can acquire all information of the three-phase asynchronous motor such as voltage, current, vibration, sound, rotating speed, temperature, three-dimensional size, shape, surface texture and the like in real time through the 5G communication module 2;
the digital twin body part in the virtual-real fusion stage is characterized in that a server carries out three-dimensional virtual reconstruction on the three-dimensional size, shape and surface texture information of a real three-phase asynchronous motor acquired in real time through three-dimensional simulation software; the method aims at acquiring voltage, current, vibration, sound, rotating speed and temperature data of a real three-phase asynchronous motor in real time, and displays the running condition of a physical space in real time in a virtual space, so that the virtual three-phase asynchronous motor and the real three-phase asynchronous motor are mapped in real time, and a digital twin body of the three-phase asynchronous motor is formed.
As shown in fig. 3, which is a schematic structural diagram of a fault diagnosis model of the fault diagnosis method for the three-phase asynchronous motor, the deep feedforward neural network is implemented in a server by software, and the deep feedforward neural network is composed of an input layer 0, a hidden layer 1, a hidden layer 2, a hidden layer 3, a hidden layer 4, a hidden layer 5 and a Softmax layer; input layer 0 contains 12 neurons
Figure BDA0002929795950000106
Input X ═ IA, IB, IC, UA, UB, UC, HX, HY, HZ, Z, V, T for 12 operating parameters of the three-phase asynchronous motor]T, the hidden layers 1 to 5 are respectively 15, 12, 10 and 8And 5 neurons, mainly used for the input parameter feature extraction and compression; input layer a [0 ]]＝X，a[0]A is obtained after passing through a hidden layer 1 to a hidden layer 5[ly]＝Sigmoid(w[ly]a[ly-1]+b[ly]) Ly e (1,2,3,4,5), the output through the Softmax layer is
Figure BDA0002929795950000101
The method is used for predicting the probabilities of the five three-phase asynchronous motor states, selecting the state with the highest probability as the final three-phase asynchronous motor fault diagnosis result,
Figure BDA0002929795950000102
wherein, w [ ly]And b [ ly]Ly belongs to (1,2,3,4,5) as unknown quantity, and needs to be determined after training;
Figure BDA0002929795950000103
function representation parameter taking
Figure BDA0002929795950000104
The maximum sigmoid function of (A) is calculated as
Figure BDA0002929795950000105
Fig. 4 is a schematic diagram for constructing a training data set and a test data set of the method for diagnosing the fault of the three-phase asynchronous motor. Firstly, using digital twin body simulation to obtain three-phase asynchronous motor operation parameters X corresponding to five conditions of 'normal', 'stator fault', 'rotor fault', 'bearing fault' and 'air gap eccentricity' of a three-phase asynchronous motor, wherein N is more than or equal to 800, and forming a three-phase asynchronous motor state data set; secondly, a Label (Label) value table of five three-phase asynchronous motor states is established: the method comprises the following steps of 1,2 for "normal", 3 for "stator fault", 4 for "bearing fault", and 5 for air gap eccentricity; thirdly, setting a label value of a corresponding state for each three-phase asynchronous motor operation parameter X; finally, 400 running parameters of the three-phase asynchronous motor in five states of 'normal', 'stator fault', 'rotor fault', 'bearing fault' and 'air gap eccentricity' are selected from the state data set of the three-phase asynchronous motor, wherein the running parameters are 2000 in total and are placed according to a random sequence, and the 2000 running parameters X and corresponding label values thereof form a training data set; according to the same method, 200 operation parameters X in the state data set of the three-phase asynchronous motor are randomly selected and placed according to a random sequence, and the 200 operation parameters and corresponding label values are used as a test data set.
As shown in fig. 5, it is a flowchart of a fault diagnosis model training process of the method for diagnosing a fault of a three-phase asynchronous motor, and the process is as follows:
(3-1) initializing parameters, setting w [ ly ], b [ ly ], ly epsilon (1,2,3,4,5) as random values, setting the iteration times as S, the learning rate as Lr, a diagnosis accuracy threshold value Tac, setting the operation parameter of the jth three-phase asynchronous motor in the training data set as X (j), and setting j as 1;
(3-2) inputting the three-phase asynchronous motor operation parameters X (j) into a fault diagnosis model, and calculating the estimated value of the detection
Figure BDA0002929795950000114
(3-3) calculating an estimated value according to the label value y corresponding to X (j)
Figure BDA0002929795950000113
Computing cross entropy loss function
Figure BDA0002929795950000111
(3-4) calculating each parameter w [ ly ] in each layer of the fault diagnosis model]And b [ ly]Change value of delta w [ ly]And Δ b [ ly]，
Figure BDA0002929795950000112
(3-5) according to the formula w[ly]＝w[ly]-Lr*Δw[ly]，b[ly]＝b[ly]-Lr*Δb[ly]Update w [ ly]And b [ ly]；
(3-6) judging whether the data is the last data in the training data set, if not, inputting the next data, and returning to (3-2); otherwise, turning to (3-7) to calculate the detection accuracy;
(3-7) inputting the operational parameters of the test data set, and calculating a detection estimation value of each operational parameter
Figure BDA0002929795950000115
Comparing with the corresponding label value y to calculate the detection accuracy
Figure BDA0002929795950000121
Wherein
Figure BDA0002929795950000122
Representing the number of the operation parameters with the same detection estimated value and label value, and sigma num (y) representing the total number of the operation parameters in the test data set;
(3-8) judging whether the detection accuracy rate meets the requirement, and if Ac is more than or equal to Tac, turning to (3-9) and finishing training; if Ac is less than Tac, judging whether the iteration times are finished, if S is not equal to 0, turning to (3-2), and reusing the training data set to perform a new round of training until the iteration is finished; if S is equal to 0, finishing the training;
(3-9) storing all parameters w [ ly ] and b [ ly ], and finishing the training of the fault diagnosis model;
the online detection is characterized in that the operation parameter X of the three-phase asynchronous motor is obtained in real time, the state of the three-phase asynchronous motor is calculated through a trained fault diagnosis model, and the real-time fault diagnosis of the three-phase asynchronous motor is realized.
The updating detection is characterized in that a fault diagnosis error threshold Te of the three-phase asynchronous motor is set, when the fault diagnosis model detects the three-phase asynchronous motor on line, the condition that the detection result is wrong is recorded to form a migration data set Xe, when the quantity Ce of the migration data set is larger than or equal to Te, migration learning is carried out to update the fault diagnosis model, otherwise, model updating is not carried out.
As shown in fig. 6, which is a schematic diagram of a structure of a migration model of the method for diagnosing a fault of a three-phase asynchronous motor, because the operating conditions of the fault diagnosis model change, the fault diagnosis model is no longer adapted to new operating conditions, the fault diagnosis model is subjected to migration training by using migration learning so as to be adapted to the new operating conditions, the models used by the migration learning are two fault diagnosis models, and the same parameters and structures are provided on the same layer, and the models are updated synchronously; one of the two sets of input training data set data is used for extracting the output of the hidden layer 4 to be XS, the other set of input migration data set data is used for extracting the output of the hidden layer 4 to be XT, a loss function in the field is calculated through XS and XT and used for judging the distance between the original running condition and the new running condition of the fault diagnosis model, and the distance is shortened through training, so that the purpose of migration and updating of the fault diagnosis model is achieved.
As shown in fig. 7, it is a flow chart of the transfer learning of the fault diagnosis method for the three-phase asynchronous motor, and the process is as follows:
(4-1) initializing parameters, setting parameters w [ ly ], b [ ly ], ly belonging to (1,2,3,4,5), setting iteration times ST, a learning rate Ltr, a global loss function threshold value TG, a j-th three-phase asynchronous motor operation parameter of a training data set as X (j), setting a j-th three-phase asynchronous motor operation parameter of a migration data set as Xe (j), and setting j to 1;
(4-2) inputting X (j) into a fault diagnosis model, and calculating an estimated value of detection thereof
Figure BDA0002929795950000131
And acquiring a characteristic value XS output by the hidden layer 4;
(4-3) inputting Xe (j) into the fault diagnosis model, and acquiring a characteristic value XT output by the hidden layer 4;
(4-4) according to the label value y and the estimation value corresponding to X (j)
Figure BDA0002929795950000132
Computing cross entropy loss function
Figure BDA0002929795950000133
Figure BDA0002929795950000134
(4-5) calculating a domain loss function LD (XS, XT) according to XS and XT,
Figure BDA0002929795950000135
wherein φ (·) > X → H represents a feature mapping from a feature space to a regenerative kernel Hilbert space;
(4-6) according to
Figure BDA0002929795950000136
And LD (XS, XT), computing a global loss function LG,
Figure BDA0002929795950000137
wherein λ represents a weight coefficient occupied by migration to the target domain;
(4-7) calculating each parameter w [ ly ] in each layer of the failure diagnosis model]And b [ ly]Change value of delta w [ ly]And Δ b [ ly]，
Figure BDA0002929795950000138
(4-8) according to the formula w[ly]＝w[ly]-Ltr*Δw[ly]，b[ly]＝b[ly]-Ltr*Δb[ly]Update w [ ly]And b [ ly]；
(4-9) judging whether the data is the last data in the data set, if not, inputting the next data, and returning to (4-2); otherwise, go to (4-10);
(4-10) judging whether the global loss function value meets the requirement, and if LG is less than or equal to TG, turning to (4-11) and finishing transfer learning; if LG is larger than TG, judging whether iteration times are finished, if S is not equal to 0, turning to (4-2), and reusing the training data set and the migration data set to perform a new round of training until iteration is finished; if S is equal to 0, the training is finished, and the process goes to (4-11);
and (4-11) saving all parameters w [ ly ] and b [ ly ] and finishing updating the fault diagnosis model.
In the description herein, references to the description of "one embodiment," "an example," "a specific example" or the like are intended to mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
The foregoing shows and describes the general principles, essential features, and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed.

Claims (5)

1. A three-phase asynchronous motor fault diagnosis device comprises a terminal sensing system and a remote data center, and is characterized in that the terminal sensing system comprises an ARM processor, a raspberry group module, a vibration sensor X, a vibration sensor Y, a vibration sensor Z, a current transformer A, a current transformer B, a current transformer C, a voltage converter A, a voltage converter B, a voltage converter C, a sound sensor, a rotation speed sensor, an infrared thermal camera, a 3D stereo camera and a 5G communication module 1;
the remote data center consists of a server and a 5G communication module 2; the vibration sensor X, the vibration sensor Y, the vibration sensor Z, the current transformer A, the current transformer B, the current transformer C, the voltage converter A, the voltage converter B, the voltage converter C and the sound sensor are respectively connected with different AD interfaces of the ARM processor; the rotating speed sensor is connected with a GPIO interface of the ARM processor; the ARM processor, the infrared thermal camera, the 3D stereo camera and the 5G communication module 1 are respectively connected with different USB interfaces of the raspberry pi module; the 5G communication module 2 is connected with the server through a USB interface; the terminal sensing system is connected with the remote data center through a mobile internet.
2. The method of the fault diagnosis device of the three-phase asynchronous motor according to claim 1, which consists of a virtual-real fusion stage, a fault diagnosis stage and a model updating stage, wherein the virtual-real fusion stage consists of data perception and digital twinning; the fault diagnosis stage consists of a fault diagnosis model, offline training and online detection; the model updating phase consists of updating detection and transfer learning.
3. The method of the fault diagnosis device for the three-phase asynchronous motor as claimed in claim 2, wherein the data sensing, the ARM processor collects three-phase current values IA, IB, IC of the three-phase asynchronous motor through three current transformers respectively, collects three-phase voltage values UA, UB, UC of the three-phase asynchronous motor through three voltage transformers respectively, collects three-directional vibration values HX, HY, HZ of the three-phase asynchronous motor X, Y, Z through three vibration sensors respectively, collects sound information Z of the three-phase asynchronous motor through a sound sensor, measures the rotating speed V of the three-phase asynchronous motor through a rotating speed sensor, and sends the measured rotating speed V to the raspberry pi processing module through the USB interface; the raspberry processing module acquires temperature information T of the three-phase asynchronous motor through an infrared thermal camera, acquires the three-dimensional size, shape and surface texture of the three-phase asynchronous motor through a 3D (three-dimensional) camera, and sends the three-dimensional size, shape and surface texture together with IA (internal information), IB (internal information), IC (integrated circuit), UA (user agent), UB (user agent), UC (user agent), HX (human X), HY (human H), HZ (human Z), Z and V (human V) sent by an ARM (advanced RISC machines) processor to; the server of the remote data center can acquire all information such as voltage, current, vibration, sound, rotating speed, temperature, three-dimensional size, shape, surface texture and the like of the three-phase asynchronous motor in real time through the 5G communication module 2;
the digital twin is characterized in that three-dimensional virtual reconstruction is carried out on the three-dimensional size, shape and surface texture information of the real three-phase asynchronous motor acquired in real time on the server through three-dimensional simulation software, the voltage, current, vibration, sound, rotating speed and temperature data of the real three-phase asynchronous motor are acquired in real time, the running condition of a physical space is displayed in real time in a virtual space, and the virtual three-phase asynchronous motor and the real three-phase asynchronous motor are mapped in real time, so that the digital twin of the three-phase asynchronous motor is formed.
4. The method for diagnosing the fault of the three-phase asynchronous motor according to claim 2, wherein the fault diagnosis model is a deep feedforward neural network implemented in a server by software, and the structure of the deep feedforward neural network is composed of an input layer 0, a hidden layer 1, a hidden layer 2, a hidden layer 3, a hidden layer 4, a hidden layer 5 and a Softmax layer; input layer 0 contains 12 neurons
Figure FDA0002929795940000021
Input X ═ IA, IB, IC, UA, UB, UC, HX, HY, HZ, Z, V, T for 12 operating parameters of the three-phase asynchronous motor]T, the hidden layers 1 to 5 respectively comprise 15 neurons, 12 neurons, 10 neurons, 8 neurons and 5 neurons, and are mainly used for feature extraction and compression of input parameters; input layer a [0 ]]＝X，a[0]A is obtained after passing through a hidden layer 1 to a hidden layer 5[ly]＝Sigmoid(w[ly]a[ly-1]+b[ly]) Ly e (1,2,3,4,5), the output through the Softmax layer is
Figure FDA0002929795940000022
The method is used for predicting the probabilities of the five three-phase asynchronous motor states, selecting the state with the highest probability as the final three-phase asynchronous motor fault diagnosis result,
Figure FDA0002929795940000023
wherein, w [ ly]And b [ ly]Ly belongs to (1,2,3,4,5) as unknown quantity, and needs to be determined after training;
Figure FDA0002929795940000024
function representation parameter taking
Figure FDA0002929795940000031
The maximum sigmoid function of (A) is calculated as
Figure FDA0002929795940000032
The off-line training comprises establishing a training data set, a testing data set and a training fault diagnosis model determining parameter;
firstly, using digital twin body simulation to obtain N running parameters X of the three-phase asynchronous motor corresponding to five conditions of 'normal', 'stator fault', 'rotor fault', 'bearing fault' and 'air gap eccentricity', wherein N is not less than 800, and forming a state data set of the three-phase asynchronous motor; secondly, a Label (Label) value table of five three-phase asynchronous motor states is established: the method comprises the following steps of 1,2 for "normal", 3 for "stator fault", 4 for "bearing fault", and 5 for air gap eccentricity; thirdly, setting a corresponding state label value for each three-phase asynchronous motor operation parameter X; finally, 400 running parameters of the three-phase asynchronous motor in five states of 'normal', 'stator fault', 'rotor fault', 'bearing fault' and 'air gap eccentricity' are selected from the state data set of the three-phase asynchronous motor, wherein the running parameters are 2000 in total and are placed according to a random sequence, and the 2000 running parameters X and corresponding label values thereof form a training data set; according to the same method, randomly selecting 200 operation parameters X from the remaining operation parameters X of the state data set of the three-phase asynchronous motor, placing the operation parameters according to a random sequence, and taking the 200 operation parameters and corresponding label values thereof as a test data set;
the process of training the fault diagnosis model to determine parameters is as follows:
(4-1) initializing parameters, setting w [ ly ], b [ ly ], ly epsilon (1,2,3,4,5) as random values, setting the iteration times as S, the learning rate as Lr, a diagnosis accuracy threshold value Tac, setting the operation parameter of the jth three-phase asynchronous motor in the training data set as X (j), and setting j as 1;
(4-2) inputting the three-phase asynchronous motor operation parameters X (j) into a fault diagnosis model, and calculating the estimated value of the detection
Figure FDA0002929795940000036
(4-3) calculating according to the label value y corresponding to X (j)The obtained estimated value
Figure FDA0002929795940000033
Computing cross entropy loss function
Figure FDA0002929795940000034
Figure FDA0002929795940000035
(4-4) calculating each parameter w [ ly ] in each layer of the failure diagnosis model]And b [ ly]Change value of delta w [ ly]And Δ b [ ly]，
Figure FDA0002929795940000041
(4-5) according to the formula w[ly]＝w[ly]-Lr*Δw[ly]，b[ly]＝b[ly]-Lr*Δb[ly]Update w [ ly]And b [ ly]；
(4-6) judging whether the data is the last data in the training data set, if not, inputting the next data, and returning to (4-2); otherwise, turning to (4-7) to calculate the detection accuracy;
(4-7) inputting the operational parameters of the test data set, and calculating a detection estimation value of each operational parameter
Figure FDA0002929795940000044
Comparing with the corresponding label value y to calculate the detection accuracy
Figure FDA0002929795940000042
Wherein
Figure FDA0002929795940000043
Representing the number of the operation parameters with the same detection estimated value and label value, and sigma num (y) representing the total number of the operation parameters in the test data set;
(4-8) judging whether the detection accuracy rate meets the requirement, and if Ac is more than or equal to Tac, turning to (4-9) and finishing training; if Ac is less than Tac, judging whether the iteration times are finished, if S is not equal to 0, turning to (4-2), and reusing the training data set to perform a new round of training until the iteration is finished; if S is equal to 0, finishing the training;
(4-9) storing all parameters w [ ly ] and b [ ly ], and finishing the training of the fault diagnosis model;
the online detection is characterized in that the operation parameter X of the three-phase asynchronous motor is obtained in real time, the operation state of the three-phase asynchronous motor is calculated through a trained fault diagnosis model, and the real-time fault diagnosis of the three-phase asynchronous motor is realized.
5. The method of a fault diagnosis device for a three-phase asynchronous motor according to claim 2, characterized in that said model updating phase consists of update detection and transfer learning;
the updating detection is characterized in that a fault diagnosis error threshold Te of the three-phase asynchronous motor is set, when the fault diagnosis model detects the three-phase asynchronous motor on line, the condition that the detection result is wrong is recorded to form a migration data set Xe, when the quantity Ce of the migration data set is more than or equal to Te, migration learning is carried out to update the fault diagnosis model, otherwise, model updating is not carried out;
the process of the transfer learning is as follows:
(5-1) initializing parameters, setting fault diagnosis model parameters w [ ly ], b [ ly ], ly belonging to (1,2,3,4,5), setting the iteration times to be ST, the learning rate to be Ltr, the global loss function threshold value to be TG, the operation parameter of the jth three-phase asynchronous motor in the training data set to be X (j), setting the operation parameter of the jth three-phase asynchronous motor in the migration data set to be Xe (j), and setting j to be 1;
(5-2) inputting X (j) into a fault diagnosis model, and calculating an estimated value of detection thereof
Figure FDA0002929795940000051
And acquiring a characteristic value output by the hidden layer 4 and recording the characteristic value as XS;
(5-3) inputting Xe (j) into the fault diagnosis model, and acquiring a characteristic value output by the hidden layer 4 and recording the characteristic value as XT;
(5-4) according to the label value y and the estimation value corresponding to X (j)
Figure FDA0002929795940000052
Computing cross entropy loss function
Figure FDA0002929795940000053
Figure FDA0002929795940000054
(5-5) calculating a domain loss function LD (XS, XT) according to XS and XT,
Figure FDA0002929795940000055
wherein φ (·) > X → H represents a feature mapping from a feature space to a regenerative kernel Hilbert space;
(5-6) according to
Figure FDA0002929795940000056
And LD (XS, XT), computing a global loss function LG,
Figure FDA0002929795940000057
wherein λ represents a weight coefficient occupied by migration to the target domain;
(5-7) calculating each parameter w [ ly ] in each layer of the failure diagnosis model]And b [ ly]Change value of delta w [ ly]And Δ b [ ly]，
Figure FDA0002929795940000058
(5-8) according to the formula w[ly]＝w[ly]-Ltr*Δw[ly]，b[ly]＝b[ly]-Ltr*Δb[ly]Update w [ ly]And b [ ly]；
(5-9) judging whether the data is the last data in the data set, if not, inputting the next data, and returning to (5-2); otherwise, go to (5-10);
(5-10) judging whether the global loss function value meets the requirement, and if LG is less than or equal to TG, turning to (5-11) and finishing transfer learning; if LG is larger than TG, judging whether iteration times are finished, if S is not equal to 0, turning to (5-2), and reusing the training data set and the migration data set to perform a new round of training until iteration is finished; if S is equal to 0, the training is finished, and the process goes to (5-11);
and (5-11) saving all parameters w [ ly ] and b [ ly ] and finishing updating the fault diagnosis model.
CN202110145260.8A 2021-02-02 2021-02-02 Three-phase asynchronous motor fault diagnosis device based on digital twinning and transfer learning Active CN112947368B (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202110145260.8A CN112947368B (en) 2021-02-02 2021-02-02 Three-phase asynchronous motor fault diagnosis device based on digital twinning and transfer learning
LU500017A LU500017B1 (en) 2021-02-02 2021-04-09 Fault diagnosis device for three-phase asynchronous motor based on digital twinning and transfer learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110145260.8A CN112947368B (en) 2021-02-02 2021-02-02 Three-phase asynchronous motor fault diagnosis device based on digital twinning and transfer learning

Publications (2)

Publication Number Publication Date
CN112947368A true CN112947368A (en) 2021-06-11
CN112947368B CN112947368B (en) 2022-06-07

Family

ID=76241722

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110145260.8A Active CN112947368B (en) 2021-02-02 2021-02-02 Three-phase asynchronous motor fault diagnosis device based on digital twinning and transfer learning

Country Status (2)

Country Link
CN (1) CN112947368B (en)
LU (1) LU500017B1 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113569475A (en) * 2021-07-21 2021-10-29 上海工程技术大学 Subway axle box bearing fault diagnosis system based on digital twinning technology
CN113835060A (en) * 2021-11-23 2021-12-24 武汉格蓝若智能技术有限公司 Power transformer online state monitoring method and system based on digital twinning
CN115184563A (en) * 2022-09-08 2022-10-14 北京中环高科环境治理有限公司 Chemical workshop field data acquisition method based on digital twinning
CN115238925A (en) * 2022-07-25 2022-10-25 北京卓尔忠诚科技有限公司 Motor equipment supervision method and system

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109933881A (en) * 2019-03-06 2019-06-25 武汉大学 A kind of Fault Diagnosis of Power Electronic Circuits method based on optimization deepness belief network
CN110243595A (en) * 2019-07-09 2019-09-17 福州大学 A kind of long-range gearbox fault monitoring system based on LabVIEW
JP2019180112A (en) * 2018-03-30 2019-10-17 三菱重工業株式会社 Diagnosis device, drive system, diagnosis method and program
CN111257751A (en) * 2020-02-25 2020-06-09 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） Motor fault diagnosis device, method, apparatus and storage medium
CN111400930A (en) * 2020-04-09 2020-07-10 武汉大学 Power equipment small sample fault diagnosis method and system based on virtual and real twin space
CN111830408A (en) * 2020-06-23 2020-10-27 朗斯顿科技(北京)有限公司 Motor fault diagnosis system and method based on edge calculation and deep learning
CN112162519A (en) * 2020-10-21 2021-01-01 桂林电子科技大学 Compound machine tool digital twin monitoring system

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2019180112A (en) * 2018-03-30 2019-10-17 三菱重工業株式会社 Diagnosis device, drive system, diagnosis method and program
CN109933881A (en) * 2019-03-06 2019-06-25 武汉大学 A kind of Fault Diagnosis of Power Electronic Circuits method based on optimization deepness belief network
CN110243595A (en) * 2019-07-09 2019-09-17 福州大学 A kind of long-range gearbox fault monitoring system based on LabVIEW
CN111257751A (en) * 2020-02-25 2020-06-09 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） Motor fault diagnosis device, method, apparatus and storage medium
CN111400930A (en) * 2020-04-09 2020-07-10 武汉大学 Power equipment small sample fault diagnosis method and system based on virtual and real twin space
CN111830408A (en) * 2020-06-23 2020-10-27 朗斯顿科技(北京)有限公司 Motor fault diagnosis system and method based on edge calculation and deep learning
CN112162519A (en) * 2020-10-21 2021-01-01 桂林电子科技大学 Compound machine tool digital twin monitoring system

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113569475A (en) * 2021-07-21 2021-10-29 上海工程技术大学 Subway axle box bearing fault diagnosis system based on digital twinning technology
CN113569475B (en) * 2021-07-21 2023-09-01 上海工程技术大学 Subway axle box bearing fault diagnosis system based on digital twin technology
CN113835060A (en) * 2021-11-23 2021-12-24 武汉格蓝若智能技术有限公司 Power transformer online state monitoring method and system based on digital twinning
CN113835060B (en) * 2021-11-23 2022-02-18 武汉格蓝若智能技术有限公司 Power transformer online state monitoring method and system based on digital twinning
CN115238925A (en) * 2022-07-25 2022-10-25 北京卓尔忠诚科技有限公司 Motor equipment supervision method and system
CN115238925B (en) * 2022-07-25 2023-12-29 北京卓尔忠诚科技有限公司 Motor equipment supervision method and system
CN115184563A (en) * 2022-09-08 2022-10-14 北京中环高科环境治理有限公司 Chemical workshop field data acquisition method based on digital twinning
CN115184563B (en) * 2022-09-08 2022-12-02 北京中环高科环境治理有限公司 Chemical workshop field data acquisition method based on digital twinning

Also Published As

Publication number Publication date
LU500017B1 (en) 2022-08-09
CN112947368B (en) 2022-06-07

Similar Documents

Publication Publication Date Title
CN112947368B (en) Three-phase asynchronous motor fault diagnosis device based on digital twinning and transfer learning
CN108681747B (en) Rotary machine fault diagnosis and state monitoring system and method based on deep learning
Zhang et al. A small sample focused intelligent fault diagnosis scheme of machines via multimodules learning with gradient penalized generative adversarial networks
CN113092115A (en) Digital twin model construction method of digital-analog combined drive full-life rolling bearing
CN109856515A (en) A kind of direct current cables state of insulation judgment method and system
CN106210269A (en) A kind of human action identification system and method based on smart mobile phone
CN106707221A (en) Fault diagnosis method and system for sensor in electric energy metering device
Chen et al. Novel data-driven approach based on capsule network for intelligent multi-fault detection in electric motors
CN110216680A (en) A kind of service robot cloud ground collaborative fault diagnosis system and method
CN109948194B (en) High-voltage circuit breaker mechanical defect integrated learning diagnosis method
CN111273125A (en) RST-CNN-based power cable channel fault diagnosis method
CN111654392A (en) Low-voltage distribution network topology identification method and system based on mutual information
CN110504878B (en) Soft measurement method for rotor speed and displacement of bearingless permanent magnet synchronous motor
CN113391216A (en) Method and device for estimating available capacity of lead-acid battery based on response surface design
CN108280424A (en) A kind of rolling bearing method for predicting residual useful life based on sparse coding
CN108054768B (en) Power system transient stability evaluation method based on principal component analysis
CN114154779A (en) Health assessment and fault diagnosis method and system based on EMD-SOM
CN105260521B (en) A kind of load Fusion Modeling Method based on analytic hierarchy process (AHP)
CN111814408A (en) Motor data generation model modeling method based on conditional variation self-encoder
CN109510245A (en) A kind of electric system Coherent Generator Group discrimination method based on figure segmentation
CN113064075B (en) Motor service life prediction method based on edge calculation and deep learning
Wen et al. Fault diagnosis of rolling bearings of different working conditions based on multi-feature spatial domain adaptation
CN113269400A (en) Low-voltage distribution network equipment state evaluation method based on historical operation and maintenance information
CN113806974A (en) Stability evaluation method, device and system for transformer bushing electric connection
CN109842113B (en) Power system simplified equivalence method based on generator group dynamic feature analysis

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